Establishment of expandable induced neural stem cells (iNSCs) from adult rat astrocytes in vitro
Extracted astrocytes expressed astrocyte markers S100β (a calcium-binding protein in astrocytes) and Gfap, but not neuronal and/or NSC markers Sox2, Nestin, Doublecortin (Dcx; a microtubule-associated protein that is broadly expressed in neuroblasts and immature neurons), and β-tubulin III (Tuj1; a pan-neuronal marker) at the RNA and protein levels (Additional file 1: Figure S1).
In order to assess the ability of Zfp521 to reprogram adult rat astrocytes into iNSCs, we overexpressed Zfp521, Sox2, or the combination of both (2F) using a lentiviral delivery system (Fig. 1a). For controlled ectopic expression of the transgene, Dox-inducible lentiviruses were used. Sox2 was used as a control reprogramming factor because it has been used for both spinal cord- and brain-derived astrocytes [20, 29]. The cells were cultured in IM in the presence of Dox for 3 weeks and then in NSCM (Fig. 1b).
Overexpressions of Zfp521, Sox2, or their combination (2F) in rat astrocytes resulted in the down-regulation of the astrocyte markers S100β and Gfap and the up-regulation of NSC markers Sox2, Nestin, and Dcx at the RNA level as quantified at week 4 in vitro (Fig. 1c). Immunostaining results confirmed down-regulation of the astrocyte markers and up-regulation of the NSC markers (Fig. 1d, e). Transduction with an empty vector (astrocytes + empty vector in astrocyte medium [A] and astrocytes + empty vector in induction medium and DOX [AD]) did not result in up-regulation of the NSC markers Nestin and Dox (Additional file 2: Figure S2). A loaded vector in the absence of DOX (astrocytes + Zfp521 without DOX [AZ] or astrocytes + Zfp521 with DOX [AZD]) also did not lead to up-regulation of the NSC markers (Additional file 2: Figure S2).
A comparison of gene expression in the reprogrammed cells showed more up-regulation of Nestin and Dcx at the RNA and protein level when Zfp521 was overexpressed compared with Sox2 or 2TF overexpression (Fig. 1c, e). Therefore, we continued our experiments with only ZFP521.
Two weeks after Zfp521 transduction, several cell spheroids emerged. These spheroids were morphologically similar to spheres formed by wild-type NSCs (Fig. 2a).
The efficiency of reprogramming was estimated using the number of Dcx+ cells 4 weeks after transduction relative to the number of astrocytes that had been initially seeded in three independent experiments. The efficiency of reprogramming for adult rat-derived astrocytes with Zfp521 was 46.10 ± 2.9%. iNSCs generated by this method could be expanded onto laminin/poly-l-ornithine-coated dishes and showed morphological homogeneity (Fig. 2a).
qRT-PCR indicated that exogenous Zfp521 was not expressed in the absence of Dox (Fig. 2b) and the expression of astrocyte markers, S100β and Gfap, were down-regulated while Sox2, Nestin, Dcx, and Tuj1 were up-regulated at passage 5 in iNSCs (Fig. 2c). In addition, immunofluorescence analyses showed that iNSCs expressed endogenous Zfp521, Sox1, Pax6, Nestin, Dcx, and Tuj1 markers while astrocyte markers S100β and Gfap were down-regulated (Fig. 2d, e).
In order to perform additional characterization of the iNSCs, we monitored their growth curve at passage 10 (Fig. 3a). Assessment of the aging of these iNSCs by SA-β-Gal activity at passages 3 and 10 showed a non-significant difference in relative fluorescence unit (RFU) of the same number of cultured cells in both passages (Fig. 3b). We assessed apoptosis in the iNSCs by staining the cells with anti-Caspase 3 and Annexin V at passages 3 and 10. We observed no significant difference in apoptosis at both passages (Fig. 3c, d).
iNSCs were allowed to spontaneously differentiate into NSCM without growth factors to assess their multipotency and ability to generate the three main neural cell types (neurons, astrocytes, and oligodendrocytes) (Fig. 4a). Immunostaining of the cells at 3 weeks post-differentiation (3 WPD) showed that the iNSCs could differentiate into neurons (NF200 and Gaba-A receptor), astrocytes (Glast and Gfap), and oligodendrocytes (Olig2) (Fig. 4b).
Collectively, these results demonstrated that Zfp521 could reprogram adult rat astrocytes in vitro into stable self-renewing iNSCs.
Functional analyses of the adult rat contusion model of spinal cord injury (SCI) after in vivo transduction with Zfp521
We sought to determine whether the adult rat spinal cord astrocytes could be converted into iNSCs with the same factor and whether transduced cells or exogenous Zfp521 improved a degree of function in the rat SCI model. The SCI model was generated in the adult rat by a contusion at the T9–11 level (Additional file 3: Figure S3A) and resulted in paralysis of the hind limbs. Histological analyses of the spinal cords at 1 week post-injury showed high expression of Gfap+ cells around the formed cavity. Additional file 3: Figure S3B-D shows images of longitudinal and transverse sections of the injured spinal cord with hematoxylin and eosin (H&E) staining and immunostaining for Gfap. The expression of Gfap was lower in the rostral and caudal sections of the lesion site (Additional file 3: Figure S3C). Furthermore, scar tissue was detected by the expression of Gfap and fibronectin around the site of the injury (Additional file 3: Figure S3D).
Four groups of rats were treated 1 week post-injury: a mock group that received the empty vector (Mock, n = 6); a group that received the Zfp521 vector (Zfp, n = 13); a group that received Zfp astrocytes (2 × 105 cells, AST-Zfp, n = 8); and a group that received mock astrocytes (2 × 105 cells, AST-Mock, n = 5) (Fig. 5a). We used a Zfp-IRES-GFP construct (Zfp-GFP) under the control of an SFFV promoter to trace the insertion of the lentiviral vector (Fig. 5b). Transduction of astrocytes with this vector showed the generation of green aggregates at 1 WPT (Fig. 5c). These cells also expressed Dcx at 4 WPT (Fig. 5d).
The functional abilities of the rats were analyzed weekly for 7 weeks after the injury and graded according to the BBB locomotor rating scale [34]. Data analysis revealed significant improvements in the AST-Zfp group compared with the Mock and AST-Mock groups (Fig. 5e, p < 0.001), and in the Zfp group compared with the AST-Zfp group at 6 WPT (Fig. 5e, p < 0.001). Differences were also found between the intact group and all-treated rats at 6 WPT (Fig. 5e, p < 0.001).
BBB scores were further analyzed by calculating the subscores [34, 35, 37,38,39], which allows for the characterization of the individual aspects of locomotion, either alone or in combination. Because the subscore can only quantify characteristics of locomotion that are present once the animal can take a step, this measure allows a more targeted and expanded evaluation of stepping quality than the basic BBB score [40]. Figure 5f and g show an analysis of the BBB subscores at 6 WPT. The BBB subscores represent measures of paw position, toe clearance, trunk control, and tail position, independent of all other observable traits (Fig. 5f). These BBB subscores were significantly improved in the Zfp group compared with the Mock, AST-Mock, or AST-Zfp group as these rats stepped earlier and displayed a more normal stepping pattern (p < 0.05, Fig. 5g).
We analyzed the footprint parameters at 6 WPT to assess the locomotor activity of the hind limbs (Fig. 6a, b). The stride angle, step length, and toe spread in the Zfp-treated group showed values similar to the intact group and improved in comparison with the Mock group (Fig. 6c, p < 0.05). The step width parameter showed no improvement in the ZFP group. The foot length and paw area parameters were improved in the Zfp group compared with the Mock group (Fig. 6c, p < 0.05).
MEP recordings were performed 2 days before SCI, 2 days after SCI, and at 6 WPT to assess the functional integrity of the spinal cord (Fig. 7a). We stimulated the left motor cortex and recorded the right tibia muscle. In intact recordings (2 days before SCI), every excitation elicited two separate waves, N1 MEPs and N2 MEPs [41]. N1 MEPs were single transcranial electrical pulses that were evoked with a short latency and recorded in the tibia anterior muscles in all of the animals that we tested. In contrast, N2 MEPs are polyphasic component pulses that were evoked with a longer latency. Neither N1 nor N2 MEPs occured 2 days after SCI or appeared with a smaller amplitude (Fig. 7b). At 6 WPT, only N1 MEPs were detected in the Mock and ZFP groups. At 6 WPT, wave amplitude (Fig. 7b) increased significantly in the Zfp group and latency (Fig. 7c) decreased in comparison with the Mock group.
In vivo reprogramming of astrocytes in the adult contusion model of spinal cord injury (SCI)
We sought to identify possible mechanisms for the perceived functional improvement by examining the cell types targeted by the Zfp lentiviral system in our model. Immunohistofluorescence analysis of longitudinal sections indicated that GFP+ cells were present around the site of injection and scar at 1 WPT (Fig. 8a). The vast majority of GFP+ cells expressed the astrocyte-specific marker Gfap (83.3 ± 3.3%) (Fig. 8b, d). A small percentage of GFP+ cells stained positive for markers of oligodendrocyte precursors (O4), microglial cells (CD86), neurons (NeuN), and NSCs (Nestin) (Fig. 8c, d). We could not identify cells that co-expressed GFP and Dcx (neuroblasts) (Fig. 8c, d). These results demonstrated that our lentivirus primarily targeted spinal astrocytes under the regulation of the SFFV promoter.
We then examined Zfp-induced neurogenesis in the adult spinal cord using immunohistochemistry at 4 and 6 WPT (Fig. 9). At 4 WPT, the percentage of GFP and Gfap+ cells decreased and the percentage of GFP and Nestin+ or Dcx+ cells increased (Figs. 9a and 10a). However, at 6 WPT, the percentage of cells that co-expressed GFP and Nestin and Dcx decreased, and the percentage of Tuj1 and Map2 increased (Fig. 9b and 10a). These results are indicative of transdifferentiation of astrocytes into neurons.
Regenerating axons were shown by the expression of growth-associated protein 43 (Gap43), a protein that is expressed in growing axons during development and also after injury in adults. After 6 weeks, we observed Gap43+ and GFP+ cells around the scar areas (Fig. 10b).
Assessment of the spinal cord sections in the Mock and Zfp groups at 6 WPT (Fig. 11a) showed the cavity area, and there was significantly reduced Gfap fluorescence intensity in the Zfp group (Fig. 11b, c). The majority of GFP+ cells in the Mock group were Gfap+ astrocytes, whereas the majority of GFP+ cells in the Zfp group were Nestin+ and Map2+ cells (Fig. 11d, e).