Increased encapsulated cell biodelivery of nerve growth factor in the brain by transposon-mediated gene transfer


Nerve growth factor (NGF) is a potential therapeutic agent for Alzheimer's disease (AD) as it has positive effects on the basal forebrain cholinergic neurons whose degeneration correlates with the cognitive decline in AD. We have previously described an encapsulated cell biodelivery device, NsG0202, capable of local delivery of NGF by a genetically modified human cell line, NGC-0295. The NsG0202 devices have shown promising safety and therapeutic results in a small phase 1b clinical study. However, results also show that the NGF dose could advantageously be increased. We have used the sleeping beauty transposon expression technology to establish a new clinical grade cell line, NGC0211, with at least 10 times higher NGF production than that of NGC-0295. To test whether encapsulation of this cell line provides a relevant dose escalation step in delivering NGF for treatment of the cognitive decline in AD patients, we have validated the bioactivity of devices with NGC0211 and NGC-0295 cells in normal rat striatum as well as in the quinolinic acid striatal lesion model. These preclinical animal studies show that implantation of devices with NGC0211 cells lead to significantly higher NGF output, which in both cases correlate with highly improved potency.


Alzheimer's disease (AD) is a progressive, fatal neurodegenerative disorder characterised by deterioration in cognition and memory, progressive impairment in the ability to carry out activities of daily living and a number of neuropsychiatric symptoms. The degeneration of cholinergic neurons in the basal forebrain correlates with the cognitive decline in AD patients.1 While current drug treatments for AD offer transient benefits to some patients, there is an urgent need for improved treatments and, in particular, disease-modifying therapies. A promising therapeutic agent in this respect is Nerve Growth Factor (NGF), which has shown neuroprotective and regenerative effects on the cholinergic neurons in the basal forebrain in various animal models, such as axonal injury, spontaneous age-related atrophy and impaired learning and memory in lesioned and aged rats or primates.2, 3, 4, 5, 6, 7, 8 Besides bypassing the blood–brain barrier, a challenge for clinical application is the safe delivery of therapeutic doses of NGF to the target area, as impaired endogenous transport of NGF to the cholinergic neurons is part of the AD pathology.9 Encapsulated Cell (EC) Biodelivery combines the advantages of gene therapy with the safety of a retrievable and replaceable medical device.10 The implantable EC Biodelivery device containing a genetically engineered cell line secretes a low dose directly into the relevant anatomical region of the brain. The semipermeable membrane allows for influx of nutrients and the efflux of the therapeutic protein, and immunologically isolates the host brain from the ECs. As applied to AD, the device contains NGF-secreting cells and is placed in the medial and lateral aspects of the basal forebrain.

Several studies have shown beneficial effects of encapsulated NGF-secreting baby hamster kidney cells or rat fibroblasts in animal models with relevance for AD.11, 12, 13, 14 However, to avoid zoonosis and xenogenic reactions from the secretome, a human cell line is preferred for clinical applications. We have previously described clinically relevant, NGF-secreting EC Biodelivery devices, which can be safely implanted in the brain, retrieved and show long-term function.15 The clinical product, named NsG0202, contains NGF-secreting NGC-0295 cells, derived from a human retinal pigment epithelial (RPE) cell line, ARPE-19. This spontaneously immortalised cell line shows contact-inhibited growth and survives well in nutrient-poor culture environments,16 making it well suited for encapsulation. A small phase 1b clinical study to examine the safety of NsG0202 devices implanted in the basal forebrain of patients with mild to moderate AD has shown promising results (Eriksdotter-Jönhagen et al.; Wahlberg et al., manuscripts submitted). The successful safety profile in all patients and relevant improvement in cognition and biomarkers in some patients suggest that a dose escalation step is relevant.

To accomplish higher secretion, we have generated regulatory compliant clonal NGF-secreting ARPE-19 cell lines using the sleeping beauty (SB) transposon expression system.17 This technology uses an optimised SB transposase (SB100X), which mediates genomic integration of multiple copies of the transgene inserted between two transposon terminal inverted repeats. The preferred sites for genomic SB integration are palindromic AT repeats, and the insertion site distribution is nearly random. The vector system is capable of stable gene transfer with long-term gene expression at an efficiency comparable with that of viral systems.18 From the established clones, we have selected the cell line NGC0211, showing at least 10-fold higher NGF expression than cell line NGC-0295, generated using standard transfection techniques. Correct processing of NGF from NGC0211 was confirmed, and the bioactivities of devices with NGC0211 and NGC-0295 were validated in preclinical animal studies in normal rat brain and in the quinolinic acid (QA) striatal lesion model of neuronal degeneration.


NGC0211 cells secrete high amounts of correctly processed NGF

The transgene copy number in the NGC0211 cell line was determined to be four using a derivative of the splinkerette technique.19 NGF release from NGC0211 cells was more than 10-fold higher than NGC-0295, created using standard liposome transfection techniques (400 ng 10−6 cells per 24 h versus 30 ng 10−6 cells per 24 h).

Processing of NGF secreted from NGC0211 cells was investigated by western blotting analyses, which showed a monomer of processed mature NGF with the expected molecular weight of 13.2 kDa (Figure 1a). Faint bands representing the molecular weight of the NGF dimer was also seen. Importantly, no pro-NGF (migrates at 32–34 kDa) was detected. As previously observed for NGF produced from NGC-0295 cells,15 a small fraction (2–3%) of partially processed NGF, represented by a double band with molecular weight between 21.3 and 29 kDa, was present. As also described for NGC-0295 cells, we observed a higher molecular weight band (>36 kDa) in the NGC0211-produced NGF sample. This band was not recognised by a pro-NGF antibody (data not shown) and may represent an aggregated form of NGF, possibly formed during the processing of samples for SDS-PAGE. In the shown experiment, it was not detected in the NGC-0295-produced NGF sample. The exact processing of the NGF secreted from NGC0211 cells was checked by high-pressure liquid chromatography (HPLC) purification and MALDI-TOF mass spectrometry. NGF was purified by three HPLC steps and the last step resulted in several discrete ultraviolet peaks. One of these showed high NGF immunoreactivity and also the expected characteristics of mature NGF, as analysed by mass spectrometry and top–down sequencing (Figure 1b).

Figure 1

Characterisation of NGF produced from NGC0211 cells. (a) NGF western blotting of conditioned medium from NGC-0295 and NGC0211 cells. Purified recombinant human NGF (rh NGF) is included as reference. The mature monomer (NGF) is indicated by an arrow. A band, corresponding to the dimer, and the double band, representing intermediate partially processed forms (pp NGF), are also shown by arrows. (b) The immunoreactive peak from the last HPLC purification step analysed by MALDI-TOF mass spectrometry. The average molecular mass was 13 488.0 Da, compared with the theoretical 13.488,4 Da of intact NGF. Top–down sequencing confirmed the residues 10–15 of NGF. In addition, the remaining N-terminal fragment (Ser-Ser-Ser-His-Pro-Ile-Phe-Arg) had the molecular mass 1066.8 Da compared with the theoretical 1066.5 Da.

Increased delivery of NGF from devices with NGC0211 cells in normal rat brain

To compare the in vivo delivery of NGF, devices with NGC0211 cells, NGC-0295 cells or parental ARPE-19 cells were implanted in rat striata and explanted after 2 weeks. NGF release from devices was measured before implantation and after explantation (Figure 2a). Release from devices with NGC0211 cells (9.1±0.3 ng per 24 h) was significantly higher than from devices with NGC-0295 cells (1.0±0.1 ng per 24 h) before implantations (t-test, P<0001). After explantation, the measured NGF release was around seven times higher for both cell lines. No NGF release was detected from devices with parental ARPE-19 cells (Figure 2a). Quantifications of NGF immunostainings reflected weak staining around NGC-0295 devices and a prominent NGF immunoreactivity covering the main part of the striatum around NGC0211 devices (Figure 2b). The radius of visible immunoreactive NGF from the implant edge was up to 2 mm. Explanted devices contained healthy living cells (Figure 2c).

Figure 2

NGF release from devices with parental ARPE-19 cells, NGC-0295 or NGC0211 cells, implanted in normal rat brain for 2 weeks. (a) NGF release measured from devices before and after the 2 weeks in rat brain. Error bars denote s.e.m., n=8 (b) Quantification of the NGF immunostaining in tissue surrounding devices. Data are shown as the mean NGF-immunoreactive area per section (±s.e.m., n=8). Insert shows a brain having a device with NGC-0295 cells in the left side and one with NGC0211 cells in the right side. Scale bar=2 mm). (c) Hematoxylin- and eosin-stained sections of explanted devices with the three indicated cell types. Scale bar=300 μm.

In an additional experiment, NGF levels in the tissue surrounding devices with NGC0211 cells and NGC-0295 cells, respectively, were determined by NGF enzyme-linked immunosorbent assay (ELISA) analyses of homogenised punches, taken at the implant site. Results showed high NGF levels (128±26 pg mg−1 tissue) around devices with NGC0211 compared with devices with NGC-0295 cells (10.1±2.5 pg mg−1 tissue).

To further characterise the increased NGF release measured after explantation, the membranes from 50% of the explanted devices were separated from the cell-containing yarn matrix, and the NGF release was determined. Intact explanted devices and devices kept in vitro throughout the experimental period were used as references. In this experiment, NGF release from intact explanted devices was around five times higher than from the in vitro devices. However, from the explanted devices, a high NGF release was measured from the membranes with no cells (55.7±11.9% of release measured from intact explanted devices) compared with release from membranes taken from the devices kept in vitro (13.8±0.8% of release measured from intact devices).

Higher potency of NGC0211 devices in normal rat brain

NGF has previously been shown to increase the size of intact as well as degenerating cholinergic neurons.20, 21, 22 Accordingly, larger cholinergic neurons, expressing choline acetyltransferase (ChAT), were seen around devices with both of the NGF-producing cell lines, compared with devices containing the parental ARPE-19 cells (Figure 3a). The size of the ChAT-positive neurons in a distance of up to 2 mm from the devices was quantified by image analyses, which showed a significant effect of NGC-0295 devices (367±19 μm2 per neuron) compared with ARPE-19 devices (298±11 μm2 per neuron). Furthermore, the mean size of ChAT-positive neurons in the vicinity of devices with NGC0211 cells (386±12 μm2 per neuron) was significantly larger than around devices with NGC-0295 cells (P<0.05). There was no difference in the number of ChAT-immunoreactive neurons per section between the treatment groups (P>0.05). NGF also induces sprouting of cholinergic fibres.4, 6 As shown in Figure 3b, we observed a significant increase in sprouting of p75NGFR-immunoreactive fibres around devices with NGF-producing cells. Quantification by image analysis showed significantly more sprouting around devices with NGC0211 cells (2.281±0.142 mm2 p75NGFR-immunoreactive area per section) than around devices with ARPE-19 cells (0.349±0.027 mm2 per section) and NGC-0295 cells (1.186±0.063 mm2 per section), P<0.05.

Figure 3

Response of cholinergic cells around devices. (a) ChAT-immunoreactive cells (arrows) in the vicinity of implant site (i) of devices with the three cell types. NGF secreted from devices increases cell size. Scale bar=200 μm. (b) Sprouting of cholinergic (p75NGFR-immunoreactive) fibres (arrows) around the implant site (i) of devices with the three cell types. Scale bar=500 μm.

Improved protection by NGC0211 cells in QA-lesioned rats

Next, we wanted to validate the in vivo potency of devices with NGC-0295 and NGC0211 cells in a model of striatal neurodegeneration, the QA lesion model, in which NGF has previously shown protective effects.23, 24 Empty devices and devices with the parental ARPE-19 cell line were included as controls. Unilateral striatal implantations of devices were performed 2 weeks before inducing an excitotoxic lesion by intrastriatal QA injection. The animals were scored in two behavioural tests for motor deficits before QA injections and again 2 and 4 weeks after. After the last test, animals were killed, and histological analyses were performed to examine the cellular responses in the brain.

Figure 4a shows results from the forelimb placing test, where performance of the unimpaired ipsilateral forelimb was consistent with normal function with all treatment groups making >9.4 out of 10 correct responses. The QA lesion resulted in marked performance deficits in the contralateral side in the control groups with empty devices and devices with the parental cell line. Rats having devices with NGF-producing cells showed significantly improved performance. After 2 weeks, there was no difference between the devices with NGC0211 and NGC-0295 cells (5.0±0.3 vs 4.8±0.4), whereas there was a significantly better effect of devices with NGC0211 at 4 weeks after inducing the lesion (5.1±0.3 vs 4.4±0.3, P<0.05). The results of the spontaneous forelimb use in the cylinder test were in accordance with those from the placing test (Figure 4b). At both testing points, a prominent decrease in the forepaw use in the limb contralateral to the injected striatum was observed in the groups with empty devices (16.4±2.8 and 15.0±2.7% symmetry at 2 and 4 weeks, respectively) and devices with ARPE-19 cells (15.0±1.9 and 14.3±2.6% symmetry at the two time points). Animals implanted with devices containing NGF-producing cells performed significantly better, with at least 30% symmetry (P<0.05). In the cylinder test, there was no significant difference between the groups with NGC-0295 and NGC0211 cells.

Figure 4

Protective effect of devices with NGF-secreting cells in the rat QA lesion model. (a) Forelimb placing test. Data are presented as the mean±s.e.m. Number of correct places out of 10 consecutive trials with the forelimbs ipsilateral and contralateral to the striatum 2 and 4 weeks after the QA lesion, n=8. The * shows a significant difference (P<0.05) between the indicated treatment groups. (b) Spontaneous forelimb use in the cylinder test. Values displayed represent the percent symmetry as mean contralateral touches/total touches±s.e.m., 2 and 4 weeks after QA lesion (n=8). The * indicates a significant difference between empty devices and devices with ARPE-19 cells. (c) Weight curves for the four experimental groups. Data are shown as the mean increase±s.e.m. (n=8), and * indicates a significant difference between the NGC0211 group and the other three groups. (d) Devices with NGC0211 decreases lesion size at the implant site (Bregma 0.0). Selected sections at the indicated coordinates were analysed. Data are shown as the mean lesion area per section ±s.e.m. (n=8) and * indicates a significant difference from devices with empty cells, ** indicates a difference from devices with ARPE-19 cells and NGC-0295 cells. (e) Quantification of the number of ChAT-immunoreactive neurons up to 2 mm from the implant. Data are shown as the mean number per section ±s.e.m. (n=48 sections) and * indicates a significant difference between groups with NGF-producing cells and controls.

Weight loss is a common observation when injecting QA in rat striatum.25, 26 Though we did not record a QA-induced weight reduction, we observed a significant effect of the implants with NGC0211 cells, as the weight increase in the first week after QA lesion and the mean weight throughout the study was significantly larger for the NGC0211 group than the other three experimental groups (Figure 4c). No difference was found between the mean weight of the two control groups and the NGC-0295 group.

To visualise NGF-induced neuronal protection against QA toxicity, brains were immunostained using an antibody against NeuN, expressed in the nuclei of most neuronal cell types in the brain.27 The size of the striatal lesion, devoid of NeuN-positive neurons, was quantified on coronal sections at four AP coordinates: +1.5, +1.0 (QA injection site), 0.0 (device implant) and −1.0, relative to Bregma (Figure 4d). At the device implant level, the lesion area was significantly smaller in the NGC0211 group than in the other three experimental groups (P<0.05). The mean lesion area around devices with NGC-0295 cells was smaller than around empty devices but not different from the group with devices containing parental ARPE-19 cells. Posterior to the device, the groups with NGF-producing cells had a significantly smaller lesion than animals with empty devices. There was no difference in the lesion size between the groups at the QA injection site or 0.5 mm anterior. In accordance with the smaller lesion size, the number of ChAT-positive neurons in a distance of 2 mm from the devices was significantly higher around devices with NGC-0295 and NGC0211 cells than around empty devices and devices with ARPE-19 cells (Figure 4e). No cholinergic sprouting was observed in the QA-lesioned brain (data not shown), and the size of cholinergic neurons around control devices (242±6 μm2 per neuron) did not differ significantly from the neuronal size around NGF-secreting devices (241±8 for NGC0211 and 226±5 μm2 per neuron for NGC-0295, P>0.05).

In contrast to the data obtained from normal brain, the mean NGF release measured from explanted devices with NGC0211 cells decreased compared with the pre-implantation values (from 13.5±1.5 to 1.0±0.3 ng per 24 h) in the QA-lesioned brains. The NGF release from devices with NGC-0295 seemed more stable, but low (from 1.8±0.4 to 2.5±1.0 ng per 24 h). Accordingly, the NGF around the devices in the QA-lesioned brains was below the detection level for the immunohistochemistry method (data not shown). Devices with both the parental cells (data not shown) and the NGF-producing clones contained obviously healthy living cells (Figures 5a and b). However, in some devices a fraction of dead cells (without nuclei) was observed. Estimating the cell number of explanted NGC0211 and NGC-0295 cells by quantifying the area of nuclei per device section (Figure 5c) showed no significant difference (P=0.282).

Figure 5

Haematoxylin- and eosin-stained sections of explanted devices. Representative explanted devices with (a) NGC-0295 cells and (b) NGC0211 cells. Healthy cells with dark-stained nuclei are seen (arrows). Scale bar=300 μm. (c) Relative survival determined by quantification of the total nuclei area in device sections, shown as mean area per section ±s.e.m. (n=8 devices).


In the present study, we have obtained a 10-fold increase in NGF expression from the EC Biodelivery platform cell line, ARPE-19, by using the SB transposon technology. The resulting expression level is also at least 10 times higher than previously described for NGF-secreting rat fibroblasts used for grafting or encapsulation.6, 11, 23 The multiplicity of transgene copies may be the primary reason for the enhanced expression compared with standard transfection techniques. Prevention of transgene downregulation could be another factor. Transgenes delivered by non-viral approaches often lead to concatamer-induced gene silencing by heterochromatin formation. As the cut and paste mechanism of transposon-mediated gene transfer results in a single copy of the gene per insertion locus, this downregulation is avoided. Furthermore, results from SB transposon-mediated gene transfer in HeLa cells suggest that inactive heterochromatin regions are rarely targeted.28

NGF secretion as correctly processed bioactive molecules from NGC0211 cells was confirmed by mass spectrometry, and we found that the differences in the NGF release in vitro correlated well with NGF concentrations measured in tissue surrounding devices with the two cell lines. Quantification of NGF immunostainings gave an even higher difference (50 times). This result, however, is likely to reflect that the NGF concentrations around devices with NGC-0295 cells are at the borderline of the detection level for the NGF immunohistochemistry method, as previously described.15 Importantly, our study shows a marked improvement in bioactivity when using the NGC0211 cell line for delivery of NGF. The diffusion distance of visible immunoreactive NGF from the implants with NGC0211 cells was up to 2 mm. While outside the striatal target area, biologically relevant NGF doses may be present at even farther distances from the devices, an important notion for the clinical setting.

Though our primary indication for the NGF-releasing EC Biodelivery devices is AD, we have tested the bioactivity in the excitotoxic striatal QA model, which is modelling many aspects of Huntington's disease. However, QA is also likely to have a role in the pathogenesis of other neurodegenerative diseases, including AD.29, 30 Furthermore, the striatum is a sufficiently large structure to accommodate the device and it contains cholinergic interneurons, responsive to NGF. In accordance with earlier studies delivering NGF by virus or ECs in the QA lesion model,23, 24 we found a protective effect of the NGF-secreting devices. The protective effect was not as potent as for other neurotrophic factors, such as CNTF or meteorin.26 However, our main scope was to validate the NGF doses delivered by the two cell lines in a neurodegenerative model. The NGF dose delivered from NGC-0295 was sufficient to significantly improve performance in functional tests. However, the improved performance of NGC0211 cells was illustrated by the small, but significantly higher, improvement in the NGC0211 group in the placement test at the last testing point, the positive effect on animal growth rate and the observation that only the group with NGC0211 cells had a general lesion size significantly smaller than the group with parental cells.

In the normal rat brain, the NGF release from explanted devices with both cell lines increased, compared with pre-implantation values. The high release of NGF from the explanted membrane without cells indicates that this increase may to some extent be caused by release of NGF accumulated in the devices in vivo (in the unstirred fluid of the membrane and interior device components). In addition, the change from the brain environment into culture medium may possibly affect the NGF expression from cells in the devices. Though the post-explant release testing does not necessarily reflect the amount of NGF released in vivo, it yields important additional information if the cell line is viable and functioning. NGF secretion of devices in the naive rat was maintained throughout the implantation period. However, in the QA model, long-term NGF secretion from NGC0211 cells seemed compromised. The nuclei quantifications did not indicate less surviving NGC0211 cells, compared with NGC-0295 cells. However, the QA injections apparently affect the NGF release of the cells negatively, either by influencing cell health and/or metabolism. In accordance with this, a previous study with ECs in this model also showed decreased NGF release after device explantation.24 In cell culture, the survival and NGF release of both cell clones were not affected by QA concentrations up to 10 mM (data not shown), but the ECs may be distressed by the QA-induced secondary inflammatory responses in the brain. The exact timing of the decrease in NGF release is unknown, but evidently the higher levels of NGF present at the time of the QA injection were sufficient for a significant and lasting protection against the excitotoxic insult. In contrast to the findings in normal rat brain, cholinergic sprouting or hypertrophy around devices with NGF-secreting cells was not observed in the QA-lesioned brain. This is in agreement with the subsequently lower NGF levels, and that some NGF effects are not maintained after abolishing treatment.31, 32 The diseased AD brain environment also shows increased inflammatory responses and oxidative stress,33, 34, 35 which might affect cells negatively in implanted devices. However, the inflammatory environment in the AD brain is less severe than in the QA-lesioned rat and therefore cell viability is unlikely to be affected. Going forward, strategies to increase stress resistance of the ECs may still warrant investigation.

The EC biodelivery technology has potential challenges in the clinic, including safety of the surgery, the risk for infections at the implantation sites and the recurring need for device replacement. At this point, a safe and tolerable implantation of EC Biodelivery devices with NGC-0295 cells has been performed in AD patients. No infections at the implantation sites occurred and the subsequent explantations after 12 months were easily performed and revealed intact devices with no tissue adherence. Importantly, a relevant improvement in cognition and biomarkers in some of the AD patients was shown (Eriksdotter-Jönhagen et al., submitted manuscript). We have further boosted the NGF expression in our human ARPE-19 platform cell line using the SB transposon expression technology, and the current study validates that the resulting NGC0211 cell line shows improved performance and provides a promising advance for future clinical devices. Safety aspects and efficacy of clinically sized devices housing NGC0211 cells are therefore currently being tested in a large animal model with the future goal of performing a NGF dose escalation in EC Biodelivery treatment of the cognitive decline in AD patients.

Materials and methods

Establishment of the NGF-expressing NGC0211 cell line

The NGF expression vector pCIn.hNGF was cloned as previously described.15 From pCIn.hNGF, pCAn.hNGF was cloned by replacing the BglII/BamHI-CI promoter fragment with the corresponding cytomegalovirus early enhancer element/chicken beta-actin (CA) promoter sequence from pCAIB, a kind gift from Dr E Arenas, Karolinska Institute, Sweden. The contiguous hNGF and neomycin resistance expression cassettes were excised as one fragment from pCAn.hNGF with BglII and SspI. This fragment was inserted between the BglII and EcoRV sites in the SB substrate vector pT2BH to create the pT2.CAn.hNGF plasmid. The ARPE-19 cell line16 was cultured and transfected as previously described.15 Cells were co-transfected with plasmids pT2.CAn.hNGF and pCMV-SB-100X with the hyperactive version of the SB transposase (both the SB substrate vector pT2BH and the hyperactive SB transposase were kind gifts from Drs Zsuzsanna Izsvak and Zoltan Ivics, MDC, Berlin, Germany). The plasmid does not contain a eukaryotic selection marker cassette and is, thus, intentionally only transiently expressed. The transient expression window allows for the active, transposase-mediated integration of the SB transposon, that is, the inverted repeat SB substrate sequences and the sequences contained within these repeats. Clones were selected using G418 (Sigma-Aldrich, Copenhagen, Denmark) and single colonies were isolated and expanded. Clones producing high levels of NGF were further characterised, and NGC0211 was chosen from its ability to deliver long-term NGF expression in conventional cell culture and during encapsulation in vitro and in vivo. A stable long-term NGF production was determined in NGC0211 cells, cultured continuously under standard cell culture conditions (passaged twice a week) for 8.5 months (at least 85 passages) with regular NGF release measurements (1–2 times a month).

Characterisation of secreted NGF

Conditioned media samples from cells were processed for western blotting as previously described,36 and NGF was detected using a goat anti-human NGF antibody (R&D Systems, Arbingdon, UK). The western blots were run in triplicates and a representative blot is shown. The percentage of mature and partly processed NGF was quanitified using ImageJ 1.44 software (Bethesda, MD, USA).

For mass spectrometry, cell culture medium harvested from the NGC0211 cell line was acidified to pH 4 using 5 M HCl and then centrifuged for 30 min at 10 000 g. The supernatant was filtered through a 0.45 mixed cellulose filter (HAWP, Millipore, Copenhagen, Denmark), applied to a C4 HPLC column, 4.6 × 250 mm, and eluted with a gradient from solvent A (0.1% trifluoroacetic acid in H2O) to solvent B (0.1% trifluoroacetic acid in acetonitrile): 10–60% B in 50 min. Fractions (1 min) were collected and assayed by NGF ELISA as described below. The NGF-immunoreactive fractions were purified in two further steps using respectively a phenyl and a C8 column, both 2.1 × 150 mm, and both eluted with a 15–35% B gradient in 40 min; fractions of 0.5 min were collected. All HPLC columns were from Vydac (Hesperia, CA, USA) and were mounted in a 1090 HPLC (Hewlett-Packard, Waldbronn, Germany). Mass spectrometric analyses were performed using an Autoflex II Tof/Tof instrument (Bruker, Bremen, Germany). The molecular mass was obtained using α-cyano-4-hydroxy-cinnamic acid as matrix, while 1.5-diaminonaphtalene was used for top–down sequencing using in-source-decay.37

Encapsulation of cells in EC Biodelivery devices

The 7-mm-long polyether sulphone membranes (Akzo Membrana, Wuppertal, Germany) were fitted with polyvinylalcohol cylindrical foam (Clinicel, M-PACT, Eudora, KS, USA). A load tube was attached to the membrane in one end with ultraviolet-cured acrylic-based glue, and the other end was sealed with the same glue. Before filling, cells were cultured in growth medium. Cells were dissociated and suspended at a density of 10,000 cells μl−1 in human endothelial serum-free medium (HE-SFM) (Invitrogen, Copenhagen, Denmark). A volume of 5 μl of cell solution (5 × 104 cells in total) was injected into each device. Devices were kept in HE-SFM at 37 °C and 5% CO2 for 2–3 weeks before implantations. Devices without cells were treated in the same manner and included as negative controls in some experiments.

Device analyses

The amount of NGF released by each capsule over a 4-h incubation period in HE-SFM was measured using NGF sandwich ELISA (R&D Systems). Standards and samples were assayed in duplicate according to manufacturer's instructions, and results were expressed in ng NGF per 24 h. Devices were subsequently fixed in 4% formalin solution, dehydrated in graded ethanol series and embedded in historesin (Leica Microsystems, Ballerup, Denmark). Sections (5 μm) were mounted on poly-L-lysine-coated slides and stained with haematoxylin–eosin (Bie & Berntsen, Rødovre, Denmark). Images of sections were prepared by a Hamamatsu Nanozoomer 1.0-HT scanner (Ballerup, Denmark).


The experiments involving animals were conducted in accordance with the Danish Animal Protection Law and the American Association for Accreditation of Laboratory Animal Care and with experimental procedures approved by the Danish National Committee for Ethics in Animal Research or the Institutional Animal care and Use Committee (US), respectively. Female Sprague-Dawley (MolTac: s.d., Taconic, Lille Skensved, Denmark) weighing around 220 g at the start of the experiment were used for the study in normal rats. Male Sprague-Dawley (Harlan, Somerville, NJ, USA) weighing 250–300 g were used in the QA lesion study. Animals were housed 2–3 per cage with free access to food and water in a standardised 12-h light/dark cycle.

In vivo potency of devices in normal striatum

Before surgery, rats (n=12) were anaesthetised with isofluran (3–4%) and positioned in a stereotaxic frame (Stoelting, Dublin, Ireland). A midline incision was made in the scalp and two bilateral holes drilled through the scull. Devices filled with NGC0211 (n=8), NGC-0295 (n=8) or ARPE-19 cells (n=8) were bilaterally implanted in striatum by an implantation cannula mounted to the stereotaxic frame. The implantation coordinates with respect to Bregma were: AP:0.0, ML:±3.2, DV:−7.5 and TB:−3.3. After implantations, the skin was suture closed. After 2 weeks, rats were deeply anaesthetised, decapitated and the brains removed. The devices were retrieved and incubated at 37 °C in HE-SFM. Media samples for determination of NGF release were collected the next day. The brains were immersion fixed in formalin solution for 48 h before cryoprotection in 30% (W/V) sucrose solution in 0.1 M sodium phosphate-buffered saline. In an additional experiment for determination of tissue levels, rats (n=6) were bilaterally implanted as described above with devices filled with NGC0211 and NGC-0295 cells, respectively. After 3 weeks, rats were killed and the brains removed. Tissue punches were taken in a radius of 2 mm around the implant site and immediately frozen in dry ice. Reference punches were taken in cortex and cerebellum. Samples were homogenised as previously described,15 and NGF ELISA was performed on diluted samples. Explanted devices were placed directly in HE-SFM, either intact or with the membrane separated from the cell-containing yarn fraction. After 24 h, NGF release in the media samples was determined by NGF ELISA. Devices, manufactured at the same time, but kept in vitro during the experimental period were included as references. NGF release from the membranes is expressed as the mean±STDEV of that from intact devices. After taking the media samples for NGF measurements, intact devices as well as the membranes without cell-containing yarn were minced with a scalpel and sonicated. DNA contents were determined by addition of Hoechst H 33258 (Invitrogen, Tåstrup, Denmark) and measurement in a DQ300 Hoefer fluorometer. No DNA was detected in the membrane, confirming the absence of cells.

In vivo potency of devices in the rat QA lesion model

Devices (n=8 in each group) were unilaterally implanted in the right striatum as described above. Two weeks after device implantations, rats (n=32) received an intrastriatal QA injection of 225 nmol as previously described26 with the coordinates adjusted to: TB, −3.3, AP: 1.0; ML: 2.6 and DV: −5.0 with respect to Bregma. Animals were tested for striatal dysfunction before QA lesion and 2 and 4 weeks after using the cylinder and forelimb placing test, as also previously described.26 All behavioural analyses were performed in a blinded fashion. After the last behavioural test 4 weeks after the lesion, the animals were euthanised and immediately perfused with cold isotonic saline through the left ventricle. The brains and devices were processed as described above.


Frozen brains were cut in six series of 40 μm on a freezing microtome. Free-floating sections were quenched of endogenous peroxidase activity through incubation with 3% H2O2. Sections were then incubated overnight with goat anti human NGF (R&D Systems), mouse anti-ChAT (Chemicon, Copenhagen, Denmark), goat anti-mouse p75NGFR (R&D Systems) or mouse anti-NeuN (Chemicon) antibodies diluted in potassium phosphate-buffered saline containing 2% normal horse serum and 0.25% Triton X-100 followed by incubation with the appropriate (anti-mouse or anti-goat) biotinylated secondary antibody. This was followed by conjugation of horseradish peroxidase using a streptavidin–horseradish peroxidase complex (ABC elite kit, Vector Laboratories, Petersborough, UK), incubation with DAB and precipitation of the chromophore with 1% H2O2. Labelled sections were slide mounted and cover slipped. Whole brain images were prepared using a Hamamatsu Nanozoomer 1.0-HT scanner.

Image analyses

Visiomorph software (VisioPharm, Hørsholm, Denmark) was used to analyse images. The mean NGF-IR area was quantified in serial sections covering the implant. Data are expressed as mean area per section (mm2)±s.e.m., n=8 implants. The number and size of striatal ChAT-IR neurons was determined in a radius of 2 mm from the implant in serial sections covering the device implant sites. The number is expressed as neurons per section ±s.e.m. and the size as μm2 per neuron±s.e.m. (n=6–8 sections per implant and 8 implants per group)). The area of p75NGFR-IR fibres was determined in serial sections covering the device implant site and is expressed as the mean area per section (mm2)±s.e.m. (n=6–8 sections per implant and 8 implants per group). Quantification of lesion area was performed on NeuN-stained sections. This was done at sections +1.5, +0.5, −0.5 and −1.5 from Bregma, where the lesioned area was manually indicated. To estimate the relative number of living cells in device sections, the area of nuclei was quantified on 20 × objective images from scanned sections, covering the entire device (two sections per device). The researcher performing the analysis was blinded to the identity of the experimental groups.


SigmaPlot (Systat, Chicago, IL, USA) was used to analyse the data. For each data set, groups were compared using one-way analysis of variance followed by all pairwise multiple comparison procedures (Dunn's method or Holm–Sidaks method) with P-values <0.05 indicating statistical significance.


  1. 1

    Jonhagen ME . Nerve growth factor treatment in dementia. Alzheimer Dis Assoc Disord 2000; 14 (Suppl 1): S31–S38.

    CAS  Article  Google Scholar 

  2. 2

    Rosenberg MB, Friedmann T, Robertson RC, Tuszynski M, Wolff JA, Breakefield XO et al. Grafting genetically modified cells to the damaged brain: restorative effects of NGF expression. Science 1988; 242: 1575–1578.

    CAS  Article  Google Scholar 

  3. 3

    Koliatsos VE, Clatterbuck RE, Nauta HJ, Knusel B, Burton LE, Hefti FF et al. Human nerve growth factor prevents degeneration of basal forebrain cholinergic neurons in primates. Ann Neurol 1991; 30: 831–840.

    CAS  Article  Google Scholar 

  4. 4

    Kordower JH, Winn SR, Liu YT, Mufson EJ, Sladek Jr JR, Hammang JP et al. The aged monkey basal forebrain: rescue and sprouting of axotomized basal forebrain neurons after grafts of encapsulated cells secreting human nerve growth factor. Proc Natl Acad Sci USA 1994; 91: 10898–10902.

    CAS  Article  Google Scholar 

  5. 5

    Martinez-Serrano A, Fischer W, Soderstrom S, Ebendal T, Bjorklund A . Long-term functional recovery from age-induced spatial memory impairments by nerve growth factor gene transfer to the rat basal forebrain. Proc Natl Acad Sci USA 1996; 93: 6355–6360.

    CAS  Article  Google Scholar 

  6. 6

    Conner JM, Darracq MA, Roberts J, Tuszynski MH . Nontropic actions of neurotrophins: subcortical nerve growth factor gene delivery reverses age-related degeneration of primate cortical cholinergic innervation. Proc Natl Acad Sci USA 2001; 98: 1941–1946.

    CAS  Article  Google Scholar 

  7. 7

    Tuszynski MH, Grill R, Jones LL, McKay HM, Blesch A . Spontaneous and augmented growth of axons in the primate spinal cord: effects of local injury and nerve growth factor-secreting cell grafts. J Comp Neurol 2002; 449: 88–101.

    CAS  Article  Google Scholar 

  8. 8

    Li LY, Li JT, Wu QY, Li J, Feng ZT, Liu S et al. Transplantation of NGF-gene-modified bone marrow stromal cells into a rat model of Alzheimer’ disease. J Mol Neurosci 2008; 34: 157–163.

    Article  Google Scholar 

  9. 9

    Salehi A, Delcroix JD, Swaab DF . Alzheimer's disease and NGF signaling. J Neural Transm 2004; 111: 323–345.

    CAS  Article  Google Scholar 

  10. 10

    Lindvall O, Wahlberg LU . Encapsulated cell biodelivery of GDNF: a novel clinical strategy for neuroprotection and neuroregeneration in Parkinson's disease? Exp Neurol 2008; 209: 82–88.

    CAS  Article  Google Scholar 

  11. 11

    Hoffman D, Breakefield XO, Short MP, Aebischer P . Transplantation of a polymer-encapsulated cell line genetically engineered to release NGF. Exp Neurol 1993; 122: 100–106.

    CAS  Article  Google Scholar 

  12. 12

    Emerich DF, Winn SR, Harper J, Hammang JP, Baetge EE, Kordower JH . Implants of polymer-encapsulated human NGF-secreting cells in the nonhuman primate: rescue and sprouting of degenerating cholinergic basal forebrain neurons. J Comp Neurol 1994; 349: 148–164.

    CAS  Article  Google Scholar 

  13. 13

    Lindner MD, Kearns CE, Winn SR, Frydel B, Emerich DF . Effects of intraventricular encapsulated hNGF-secreting fibroblasts in aged rats. Cell Transplant 1996; 5: 205–223.

    CAS  Article  Google Scholar 

  14. 14

    Winn SR, Hammang JP, Emerich DF, Lee A, Palmiter RD, Baetge EE . Polymer-encapsulated cells genetically modified to secrete human nerve growth factor promote the survival of axotomized septal cholinergic neurons. Proc Natl Acad Sci USA 1994; 91: 2324–2328.

    CAS  Article  Google Scholar 

  15. 15

    Fjord-Larsen L, Kusk P, Tornoe J, Juliusson B, Torp M, Bjarkam CR et al. Long-term delivery of nerve growth factor by encapsulated cell biodelivery in the Gottingen minipig basal forebrain. Mol Ther 2010; 18: 2164–2172.

    CAS  Article  Google Scholar 

  16. 16

    Dunn KC, Aotaki-Keen AE, Putkey FR, Hjelmeland LM . ARPE-19, a human retinal pigment epithelial cell line with differentiated properties. Exp Eye Res 1996; 62: 155–169.

    CAS  Article  Google Scholar 

  17. 17

    Ivics Z, Hackett PB, Plasterk RH, Izsvak Z . Molecular reconstruction of sleeping beauty, a Tc1-like transposon from fish, and its transposition in human cells. Cell 1997; 91: 501–510.

    CAS  Article  Google Scholar 

  18. 18

    Ivics Z, Izsvak Z . The expanding universe of transposon technologies for gene and cell engineering. Mob DNA 2010; 1: 25.

    CAS  Article  Google Scholar 

  19. 19

    Hui EK, Wang PC, Lo SJ . Strategies for cloning unknown cellular flanking DNA sequences from foreign integrants. Cell Mol Life Sci 1998; 54: 1403–1411.

    CAS  Article  Google Scholar 

  20. 20

    Klein RL, Hirko AC, Meyers CA, Grimes JR, Muzyczka N, Meyer EM . NGF gene transfer to intrinsic basal forebrain neurons increases cholinergic cell size and protects from age-related, spatial memory deficits in middle-aged rats. Brain Res 2000; 875: 144–151.

    CAS  Article  Google Scholar 

  21. 21

    Bishop KM, Hofer EK, Mehta A, Ramirez A, Sun L, Tuszynski M et al. Therapeutic potential of CERE-110 (AAV2-NGF): targeted, stable, and sustained NGF delivery and trophic activity on rodent basal forebrain cholinergic neurons. Exp Neurol 2008; 211: 574–584.

    CAS  Article  Google Scholar 

  22. 22

    Kordower JH, Chen EY, Mufson EJ, Winn SR, Emerich DF . Intrastriatal implants of polymer encapsulated cells genetically modified to secrete human nerve growth factor: trophic effects upon cholinergic and noncholinergic striatal neurons. Neuroscience 1996; 72: 63–77.

    CAS  Article  Google Scholar 

  23. 23

    Frim DM, Uhler TA, Short MP, Ezzedine ZD, Klagsbrun M, Breakefield XO et al. Effects of biologically delivered NGF, BDNF and bFGF on striatal excitotoxic lesions. Neuroreport 1993; 4: 367–370.

    CAS  Article  Google Scholar 

  24. 24

    Emerich DF, Hammang JP, Baetge EE, Winn SR . Implantation of polymer-encapsulated human nerve growth factor-secreting fibroblasts attenuates the behavioral and neuropathological consequences of quinolinic acid injections into rodent striatum. Exp Neurol 1994; 130: 141–150.

    CAS  Article  Google Scholar 

  25. 25

    Emerich DF, Mooney DJ, Storrie H, Babu RS, Kordower JH . Injectable hydrogels providing sustained delivery of vascular endothelial growth factor are neuroprotective in a rat model of Huntington's disease. Neurotox Res 2010; 17: 66–74.

    CAS  Article  Google Scholar 

  26. 26

    Jørgensen JR, Emerich DF, Thanos C, Thompson LH, Torp M, Bintz B et al. Lentiviral delivery of Meteorin protects striatal neurons against excitotoxicity and reverses motor deficits in the quinolinic acid rat model. Neurobiol Dis 2011; 41: 160–168.

    Article  Google Scholar 

  27. 27

    Mullen RJ, Buck CR, Smith AM . NeuN, a neuronal specific nuclear protein in vertebrates. Development 1992; 116: 201–211.

    CAS  PubMed  Google Scholar 

  28. 28

    Grabundzija I, Irgang M, Mates L, Belay E, Matrai J, Gogol-Doring A et al. Comparative analysis of transposable element vector systems in human cells. Mol Ther 2010; 18: 1200–1209.

    CAS  Article  Google Scholar 

  29. 29

    Ting KK, Brew BJ, Guillemin GJ . Effect of quinolinic acid on human astrocytes morphology and functions: implications in Alzheimer's disease. J Neuroinflammation 2009; 6: 36.

    Article  Google Scholar 

  30. 30

    Rahman A, Ting K, Cullen KM, Braidy N, Brew BJ, Guillemin GJ . The excitotoxin quinolinic acid induces tau phosphorylation in human neurons. PLoS One 2009; 4: e6344.

    Article  Google Scholar 

  31. 31

    Montero CN, Hefti F . Rescue of lesioned septal cholinergic neurons by nerve growth factor: specificity and requirement for chronic treatment. J Neurosci 1988; 8: 2986–2999.

    CAS  Article  Google Scholar 

  32. 32

    Niewiadomska G, Komorowski S, Baksalerska-Pazera M . Amelioration of cholinergic neurons dysfunction in aged rats depends on the continuous supply of NGF. Neurobiol Aging 2002; 23: 601–613.

    CAS  Article  Google Scholar 

  33. 33

    Good PF, Werner P, Hsu A, Olanow CW, Perl DP . Evidence of neuronal oxidative damage in Alzheimer's disease. Am J Pathol 1996; 149: 21–28.

    CAS  PubMed  PubMed Central  Google Scholar 

  34. 34

    Butterfield DA, Perluigi M, Sultana R . Oxidative stress in Alzheimer's disease brain: new insights from redox proteomics. Eur J Pharmacol 2006; 545: 39–50.

    CAS  Article  Google Scholar 

  35. 35

    Heneka MT, O’Banion MK, Terwel D, Kummer MP . Neuroinflammatory processes in Alzheimer's disease. J Neural Transm 2010; 117: 919–947.

    CAS  Article  Google Scholar 

  36. 36

    Fjord-Larsen L, Johansen JL, Kusk P, Tornøe J, Gronborg M, Rosenblad C et al. Efficient in vivo protection of nigral dopaminergic neurons by lentiviral gene transfer of a modified Neurturin construct. Exp Neurol 2005; 195: 49–60.

    CAS  Article  Google Scholar 

  37. 37

    Demeure K, Quinton L, Gabelica V, De PE . Rational selection of the optimum MALDI matrix for top-down proteomics by in-source decay. Anal Chem 2007; 79: 8678–8685.

    CAS  Article  Google Scholar 

Download references


We wish to thank Dr Zsuzsanna Izsvák and Dr Zoltán Ivics for introducing us to the sleeping beauty transposon system as well as the generous assistance with NGF copy number determination. We thank Helle Nymark, Janni Larsen, Juliano Olsen, Marianne Kureer, Pia Knudsen, Hanne Fosmark and Helle T Wassmann for excellent technical assistance with cells and devices. Also, the technical assistance by Allan Kastrup with NGF purification and mass spectrometry analysis is highly appreciated.

Author information



Corresponding author

Correspondence to L Fjord-Larsen.

Ethics declarations

Competing interests

LF-L, PK, JT, MT and LUW are employed by NsGene A/S. This study was funded by NsGene A/S.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Fjord-Larsen, L., Kusk, P., Emerich, D. et al. Increased encapsulated cell biodelivery of nerve growth factor in the brain by transposon-mediated gene transfer. Gene Ther 19, 1010–1017 (2012).

Download citation


  • encapsulated cell biodelivery
  • sleeping beauty transposon
  • NGF
  • preclinical validation
  • cholinergic neurons
  • rat QA model

Further reading


Quick links