Cranial grafting of stem cell-derived microvesicles improves cognition and reduces neuropathology in the irradiated brain

Novel Place Recognition.

One month after transplantation of MV, animals were habituated and tested on the novel place recognition (NPR) task (Fig. 1A). The total exploration of objects during the familiarization and test phases were found to be indistinguishable between all cohorts for this task. Group means and 95% confidence intervals (CIs) for the discrimination indices (DIs) were as follows: 0 Gy Control (Con, mean = 48.13, 95% CI = 32.48–63.78), 10 Gy irradiated (IRR, mean = 12.63, 95% CI = 8.348–33.61), and irradiated receiving MV transplantation (IRR+MV, mean = 39.32, 95% CI = 25.97–52.68). A significant overall group effect was found, F(2, 25) = 5.651, P = 0.009, for the DI to differ between groups. Following a 1-h retention interval between familiarization and test phases, IRR animals spent significantly less time exploring the novel place compared with Con (P = 0.01) and IRR+MV (P = 0.05) groups. In contrast, IRR+MV animals did not differ from Con animals. These data indicate that transplantation of MV improved spatial exploration behavior on the NPR task after irradiation compared with irradiated but nontransplanted animals.

Novel Object Recognition.

After NPR testing, animals were habituated and subject to a novel object recognition (NOR) task (Fig. 1B). As before, animals explored familiar and novel objects equally during this task. The group means and 95% CIs for the DI were as follows: Con (mean = 46.43, 95% CI = 36.24–56.61), IRR (mean = 1.25, 95% CI = –18.86–21.37), and IRR+MV (mean = 38.62, 95% CI = 18.77–58.47). Again, a significant overall group effect was found, F(2, 31) = 6.175, P = 0.005, for the DIs to differ between groups. Following a 5-min retention interval between familiarization and test phases, IRR animals spent significantly less time exploring the novel object compared with Con (P = 0.001) and IRR+MV (P = 0.01) groups. As in the prior task, the DIs for Con and IRR+MV groups were statistically indistinguishable. Thus, MV grafting improved novel object exploration behavior in irradiated animals.

Object in Place.

The third open arena test used to evaluate cognitive function was the object in place (OiP) task (Fig. 1C). Total exploration times for the objects followed similar trends as those observed for the previous tests. Rats having intact cortical function exhibit a preference for objects that had been moved to a novel location. The group means and CIs for the DI in each cohort were as follows: Con (mean = 24.08, 95% CI = 16.87–31.28), IRR (mean = –14.90, 95% CI = –27.91 to –1.89), and IRR+MV (mean = 38.26, 95% CI = 32.35–44.17). The overall group effect for DI between the three cohorts was again found to differ significantly, F(2, 31) = 52.66, P < 0.0001. Following familiarization, test phases showed that IRR animals spent significantly less time exploring the novel placement of objects compared with the Con (P = 0.001) and IRR+MV (P = 0.001) groups. Once again, the Con and IRR+MV groups did not differ statistically and showed a distinct preference for those objects placed at novel locations. These data again demonstrate that cranial grafting of MV improves exploration behavior after irradiation.

Temporal Order.

After OiP testing, rats were habituated and tested on the temporal order (TO) task (Fig. 1D). After familiarization with two sets of objects presented 4 h apart (sample phases 1 and 2) and 1 h after sample phase 2, rats were subjected to the test phase using one of the same objects presented in each of the earlier phases. Animals with intact perirhinal cortical function exhibit a preference for the object explored during sample phase 1 over the more recently explored, sample phase 2 object. Total exploration of the objects during the familiarization and test phases were reduced, albeit not significantly, after irradiation compared with Con and IRR+MV groups. Control and grafted cohorts were again not found to differ statistically. Although group differences did not reach significance for this task, overall trends indicate that transplantation of MV afforded some improvement in exploration behavior after irradiation.

In summary, for each open arena task, exploration ratios were normalized by the time spent at a familiar location or object by calculating the DI. In nearly every instance, a preference for novelty was found to be significantly higher for Con and IRR+MV groups in comparison with the IRR group, demonstrating the beneficial neurocognitive effects of MV transplantation (Fig. 1).

Contextual Fear Conditioning.

The training, cue, and context phases of the fear conditioning (FC) task were administered over a period of 3 d. Group means and 95% CIs for posttraining and context phase freezing (percent) were as follows: Posttraining Con (mean = 97.70, 95% CI = 95.0–100.4), IRR (mean = 94.56, 95% CI = 88.67–97.24), IRR+MV (mean = 92.96, 95% CI = 88.67–97.24), Context Con (mean = 48.63, 95% CI = 32.92–64.34), IRR (mean = 25.26, 95% CI = 10.86–39.66), and IRR+MV (mean = 52.15, 95% CI = 44.01–60.30). Using repeated measures (RM) ANOVA, a significant overall Group × Phase interaction effect was found for the percent time spent freezing during the FC task (Fig. 2), F(8, 104) = 2.196, P = 0.03. RM two-way ANOVA for the context phase revealed significant differences between the IRR and Con groups (P = 0.003) and between the IRR and IRR+MV groups (P = 0.001). Groups did not differ significantly in the freezing behavior across baseline, posttraining, precue, and postcue phases, indicating a selective deficit on the hippocampal-dependent contextual memory phase of the task. During the context test phase, post hoc tests confirmed that IRR animals spent significantly less time freezing compared with the Con (P = 0.02) and IRR+MV (P = 0.002) groups, whereas the Con and IRR+MV groups did not differ. Further, because all groups showed significant increases in freezing behavior after the tone–shock pairing (posttraining phase), irradiation did not impair motor or sensory function. The fact that cued memory was intact also demonstrates that the acquisition of the tone–shock pairing was not impaired and that the deficit was specific to the memory of the context in which the pairing had been learned.

Graft Location of MV.

The presence of grafted MV was confirmed at 2 d and 2 wk after surgery using rhodamine anti-green fluorescence protein (GFP) antibody immunofluorescence to amplify the GFP-CD63 signal (appearing as red puncta) and confocal Z-stack analysis of transplanted brains (Fig. 3). At 2 d posttransplantation, MV aggregates were found predominantly dorsal to the CA1 region of the hippocampus, in the vicinity of the needle track (Fig. 3A). Although smaller numbers of MV remained in the CA1 region of the hippocampus 2 wk postsurgery, many were found to disperse and to have migrated to the dentate gyrus (DG) region (Fig. 3 B and C). Immunostaining indicated that a certain fraction of the MV cargo was internalized into host brain cells. Dual immunofluorescence staining for MV cargo along with neuron-specific nuclear antigen (NeuN) or glial fibrillary acidic protein (GFAP) provided evidence, albeit circumstantial, indicating fusion and internalization of MV cargo by both neuronal and astrocytic cell types (Fig. 3 D and E, respectively).

Neuroinflammation.

Previously, we demonstrated that exposure to cranial irradiation or systemic chemotherapy increased neuroinflammation (5, 8). To assess the impact of hNSC-derived MV on inflammation in the brain, activated microglia (ED-1⁺ cells) were quantified using unbiased stereology throughout the distinct brain regions including the hippocampus, neocortex (layer II/III), and amygdala (Fig. 4). A significant overall group effect was found for the number of activated microglia in the hippocampal dentate hilus (DH), F(2, 6) = 14.32, P = 0.005, granular cell layer (GCL), F(2, 6) = 14.17, P = 0.005, and the CA1/3 subfields, F(2, 6) = 30.70, P = 0.0007, in addition to cortical layer II/III, F(2, 8) = 13.59, P = 0.003, and the amygdala, F(2, 8) = 18.00, P = 0.001, where radiation exposure induced 4–5-fold increases in ED-1⁺ cells, F(2, 6) = 21.47, P = 0.002 (Fig. 4D). However, MV grafting resulted in a significant reduction in the number of activated microglia throughout the hippocampus, reducing the yield of ED-1⁺ cells to below control levels (29%, 12%, and 45% lower than control levels for the DH, DG, and CA3/1 regions, respectively). Similar reductions in activated microglia were observed in the cortex (layer II/III) and amygdala. These data demonstrate that MV grafting could significantly suppress radiation-induced activation of microglia in regions of the brain both proximal and distal to the engraftment site.

Neuronal Morphometry.

Our past studies have shown that cranial irradiation could significantly disrupt components of neuronal architecture including dendritic length, volume, and complexity (13, 14). To determine the potential impact of MV grafting on neuronal structure, Golgi–Cox-impregnated sections were used in detailed morphometric analyses of GCL neurons in the DG.

Panoramic bright field images of Golgi–Cox-impregnated neurons in the DG reveal the marked impact of cranial irradiation and MV grafting. Compared with controls, neuronal complexity is severely compromised after radiation exposure, an adverse effect that is reversed by MV grafting (Fig. 5 A–C). Representative Golgi–Cox-stained neurons (Fig. 5 D–F) and Neurolucida tracings (Fig. 5 G–I) are shown for each cohort. Quantification of dendritic parameters on GCL neurons revealed the adverse effects of radiation and the beneficial effects of MV on neuronal structure (Fig. 5 J–L). Significant overall group effects were found for dendritic length, F(2, 9) = 9.022, P = 0.007, volume, F(2, 9) = 74.59, P = 0.0001, and complexity, F(2, 9) = 42.25, P = 0.0001, clearly demonstrating that cranial irradiation significantly compromised neuronal structure and that MV transplantation protected against this effect. Although MV transplantation did not restore dendritic volume to control levels, it did confer significant protection (P = 0.01). Transplantation was more effective in protecting other dendritic parameters, restoring dendritic length to 89% of control levels (P = 0.01) and restoring dendritic complexity to greater than control levels (P = 0.001). For each of these endpoints, data show that MV grafting preserved the structure of hippocampal GCL neurons following irradiation.

Animals and Irradiation.

All animal procedures are in accordance with NIH and were approved by the University of California Institutional Animal Care and Use Committee. Four-month-old male immunodeficient athymic nude (ATN) rats (Cr:NIH-Foxn1rnu, strain 316; Charles River) were maintained in sterile housing conditions (20 °C ± 1 °C; 70% ± 10% humidity; 12 h:12 h light and dark cycle) and had free access to sterilized diet and water. The ATN rats were divided into three experimental groups (n = 8–12 per group): 0 Gy receiving sham surgery (Con), 10 Gy head-only irradiation receiving sham surgery (IRR), and 10 Gy head-only irradiation receiving MV grafting (IRR+MV). For cranial irradiation, animals were anesthetized [2.5% (vol/vol) isoflurane/oxygen], placed ventrally on the treatment table (XRAD 320 irradiator; Precision X-Ray) without restraint, and positioned under a collimated (1.0 cm² diameter) beam for head-only irradiation delivered at a dose rate of 1.0 Gy/min.

hNSC Culture, MV Isolation, and Characterization.

The use of hNSCs was approved by the Institutional Human Stem Cell Research Oversight Committee. The validation, expansion, and characterization of hNSCs (ENStem-A; EMD Millipore) followed previously published procedures (4, 28). MV were isolated and purified from conditioned hNSC culture medium by ultracentrifugation (29) and characterized using a NanoSight 3000 (Malvern Instruments). The location of engrafted MV was determined using fluorescently labeled MV (CD63-GFP⁺).

Transplantation Surgery.

At 2 d postirradiation, each rat received bilateral, intrahippocampal transplantation of MV suspended in vehicle (hibernation buffer) using a 33-gauge microsyringe at an injection rate of 0.25 mL/min. Each hippocampus received four distinct injections of MV (∼1.0 × 10¹⁰ in 2 µL) per hemisphere using precise stereotaxic coordinates, as described previously (3). Sham surgery controls received equal volumes of sterile vehicle at the same stereotaxic coordinates. Animals were anesthetized using isoflurane/oxygen [5% (vol/vol) induction, 2.5% (vol/vol) maintenance; VetEquip].

Cognitive Testing.

To evaluate the outcome of MV transplantation on cognitive function, rats from each group were subjected to cognitive testing 1 mo after transplantation surgery. Cognitive testing was conducted over 3 wk and included four different open arena tasks followed by contextual and cued FC as described in detail previously (8, 30).

Tracking MV Engraftment.

The intracranial location of GFP⁺-engrafted MV was followed 2 d and 2 wk following surgical transplantation. Animals were processed for immunohistochemistry to determine the time-dependent distribution and intracellular location of fluorescently labeled MV cargo. For further details, refer to Cognitive Testing.

Neuroinflammation.

Following cognitive testing, animals were killed and perfused with 4% (wt/vol) paraformaldehyde/PBS. Animals subjected to behavioral testing were segregated equally for morphometric or immunohistochemical analyses (n = 4–8 per group), and brain tissues were processed for coronal sectioning using a cryostat. Immunostaining for activated microglia (ED-1⁺ cells) was carried out on serial sections (30 µm) as described previously (5). To facilitate color development, ABC Elite and Vector SG kits were used according to the manufacturer’s instructions (Vector Laboratories). Sections were mounted on gelatin-coated slides, air-dried, dehydrated, and counterstained with nuclear fast red (Vector Labs). To determine the number of activated microglia (ED-1⁺), the DH, DG, and CA3 and CA1 regions of the hippocampus, neocortex (layer II/III, 1,000 × 500 µm), and amygdala were analyzed by stereology.

Neuron Structure Analysis.

Extracted brains were subjected to Golgi–Cox impregnation and staining of neurons as per the manufacturer’s instructions (SuperGolgi kit, Bioenno Tech.), sectioned (150 µm), and counterstained by nuclear fast red. The morphometric analysis of mature neurons in the hippocampal DG was carried out as described previously (8).

Statistics.

Statistical analyses were carried out using GraphPad Prism (v6). One-way ANOVA was used to assess significance between groups. When overall group effects were found to be statistically significant, a Bonferroni multiple comparisons test was used to compare the individual groups. For analysis of FC data, RM two-way ANOVAs were performed. All analyses considered a value of P ≤ 0.05 to be statistically significant.

Characterization of hNSCs.

hNSCs were obtained from the ENStem-A cell line (Millipore), originally derived from human embryonic stem cell line H9. Cells were passaged (passage number 1) on poly–l-ornithine (20 μg/mL; Sigma-Aldrich) and laminin (5 μg/mL, Sigma-Aldrich)-coated flasks in EnStem-A neural expansion media (Millipore) containing neurobasal media supplemented with l-glutamine (2 mM; Invitrogen), bFGF (20 ng/mL; Millipore), and B27 and LIF (Leukemia Inhibitory Factor; Millipore). Monolayers of hNSCs were passaged every other day (1:2) and doubled every 28 h. Cells were maintained in a multipotent state at passage numbers between 5–10 during the course of MV isolation and confirmed by immunostaining against Sox2 and nestin as detailed previously (28).

MV Isolation and Characterization.

MV were isolated from the conditioned hNSC culture medium as described in detail (29). Briefly, large dead cells and cell debris were eliminated using successive centrifugations using increasing speeds (300 × g, 5 min; 2,000 × g, 10 min; 10,000 × g, 30 min; all at 4 °C). At each step the pellet was discarded and the supernatant carried forward. Vesicles were pelleted by ultracentrifugation using 100,000 × g (70 min, 4 °C). The pellet was then washed in sterile PBS to remove contaminating proteins and repelleted by centrifugation at 100,000 × g for an additional 70 min. MV quantity and size were determined using a NanoSight 3000 equipped with a 532-nm laser and diluted to deliver ∼1 × 10¹⁰ MV per injection site. The mean MV size was 191 nm.

Following engraftment, animals were euthanized (isoflurane) followed by intracardiac perfusion (4% PFA), and brain tissues were processed for coronal sectioning using a cryostat (Leica Microsystem). Serial sections (30 µm) were washed with Tris-buffered saline and blocked in 2% (wt/vol) bovine serum albumin (BSA) with 0.01% Triton-X 100 in Tris-buffered saline (Tris-A). For the detection of GFP alone, tissues were incubated overnight in a GFP primary antibody (chicken anti-GFP, 1:500 in 2% BSA and Tris-A; Aves Labs Inc.). The following day, sections were washed and labeled with rabbit anti-chicken rhodamine (1:250 in 1% BSA and Tris-A; EMD Millipore). After washing, sections were mounted onto gelatin-coated slides and counterstained with DAPI. For dual staining of GFP and NeuN or GFAP, the above methods were used along with mouse anti-NeuN primary antibody (1:500; EMD Millipore) or mouse anti-GFAP primary antibody (1:500; EMD Millipore) labeled with donkey anti-mouse Alexa Fluor 488 (1:500; Life Technologies).

Fluorescent Labeling of MV.

HEK293 cells were transfected using the lentivirus-based CD63-GFP plasmid (pCT-CD63-GFP, Cyto-Tracer, SBI) along with the lentivirus packaging plasmids (pLP1, pLP2, pL/VSVG, ViraPower Lentiviral Expression System; Invitrogen, Life Technologies). The supernatant with CD63-GFP lentiviral particles was harvested and GFP expression verified by fluorescence microscopy. hNSCs were transfected with the particles and cultured under puromycin selection. Optimal expression was observed 5–7 d posttransfection. Fluorescently labeled MV were purified and prepared for engraftment from hNSC culture medium as described above and transplanted as described in transplantation surgery methods.

Golgi–Cox Staining and Granule Cell Neuron Morphology.

Animals subjected to behavioral testing were segregated equally for morphometric or immunohistochemical analyses. Extracted brains were subjected to Golgi–Cox impregnation and staining of neurons as per the manufacturer’s instructions (SuperGolgi kit, Bioenno Tech.), sectioned (150 µM), and counterstained by nuclear fast red (Vector Labs). For morphometric analyses, apical and basal dendrites were traced on neurons within the DG and CA1 subfields of the hippocampus. All analyses were conducted blind from coded slides. Neurons (6–8 per region per animal; 4 total animals) were analyzed that satisfied the following criteria: (i) uniform Golgi staining throughout the dendritic tree, (ii) absence of truncated/severed dendrites, and (iii) minimal structural overlap with neighboring neurons.

Neurons satisfying the foregoing criteria were traced using a 100× oil immersion objective lens (Nikon), a computerized stage (BioPrecision 2; Ludl), and Neurolucida software (v11; Microbrightfield, Inc.). Morphological parameters that were quantified included total dendritic length, branch points, volume, dendritic complexity, and Sholl analysis, performed using the Neuroexplorer component of the Neurolucida program. Dendritic complexity was determined by the following equation: (Σ branch tip orders + number of branch tips) × (total dendritic length/total number of primary dendrites). Sholl analysis used a series of concentric circles (50 μm apart) to calculate the number of dendritic intersections as a function from distance to the soma.

ED-1 Stereology.

Stereologic quantification was used to determine the number of activated microglia (ED-1⁺) in the hippocampus after completion of behavior testing. Every 10th section through the entire hippocampus was processed for ED-1 immunohistochemistry. Stereological assessment was conducted using a Nikon microscope (Nikon TE2000-E) equipped with a MBF color digital camera, 100× (oil immersion, 1.30 N.A.; Zeiss) objective lens, three-axis motorized stage (Bio Precision 2; Ludl), and an optical fractionator probe (StereoInvestigator; MBF Bioscience). Systemic random sampling and image stack analysis modules were used to acquire batch images through the anterior–posterior planes of the hippocampus. The acquired images were uploaded onto an MBF Workstation (MBF Biosciences), and the number of activated microglia was quantified by counting ED-1⁺–positive cell bodies (black on nuclear fast red counterstained background) using the Optical Fractionator probe and systemic random sampling according to unbiased stereological principles. All analyses were conducted blind from coded slides (9–12 brain sections per animal, n = 3 animals per group). Sampling parameters (grid and counting frame size) were empirically determined to achieve low coefficients of error (Gunderson’s CE, ≤0.06 ± 0.01, n = 4) for each region of interest.