Hypoxia Inducible Factor 1 alpha (HIF-1α) counteracts the acute death of cells transplanted into the injured spinal cord.

Abstract Cellular transplantation is in clinical testing for a number of central nervous system disorders, including spinal cord injury (SCI). One challenge is acute transplanted cell death. To prevent this death, there is a need to both establish when the death occurs and develop approaches to mitigate its effects. Here, using luciferase (luc) and green fluorescent protein (GFP) expressing Schwann cell (SC) transplants in the contused thoracic rat spinal cord 7 d postinjury, we establish via in vivo bioluminescent (IVIS) imaging and stereology that cell death occurs prior to 2–3 d postimplantation. We then test an alternative approach to the current paradigm of enhancing transplant survival by including multiple factors along with the cells. To stimulate multiple cellular adaptive pathways concurrently, we activate the hypoxia-inducible factor 1α (HIF-1α) transcriptional pathway. Retroviral expression of VP16-HIF-1α in SCs increased HIF-α by 5.9-fold and its target genes implicated in oxygen transport and delivery (VEGF, 2.2-fold) and cellular metabolism (enolase, 1.7-fold). In cell death assays in vitro, HIF-1α protected cells from H2O2-induced oxidative damage. It also provided some protection against camptothecin-induced DNA damage, but not thapsigargin-induced endoplasmic reticulum stress or tunicamycin-induced unfolded protein response. Following transplantation, VP16-HIF-1α increased SC survival by 34.3%. The increase in cell survival was detectable by stereology, but not by in vivo luciferase or ex vivo GFP IVIS imaging. The results support the hypothesis that activating adaptive cellular pathways enhances transplant survival and identifies an alternative pro-survival approach that, with optimization, could be amenable to clinical translation.


Introduction
The death of transplanted cells is a common feature of cell transplants. In the central nervous system, the majority of cells die soon after transplantation (Emgård et al., 2003;Bakshi et al., 2005;Hill et al., 2006Hill et al., , 2007. This undesirable consequence of transplantation, separate from immune-mediated rejection, poses a challenge to the therapeutic use of cellular transplants for neurologic repair. Development of approaches that counteract transplant death are needed to mitigate the deleterious effects of the acute cell death and maximize the clinical utility of cell transplantation. A necessary first step in developing interventions to counteract transplanted cell death is to accurately establish when post-transplantation (post-TP) the death occurs. In experimental models of spinal cord injury (SCI), 1-35% of cells remain after one week (Barakat et al., 2005;Karimi-Abdolrezaee et al., 2006;Hill et al., 2007), indicating that most transplant death occurs in the first week post-TP. Based on assessments of cell death markers, transplanted cell death peaks within 24 h (Hill et al., 2007). However, the exact time window of transplanted cell death remains to be established. This is due, in part, to the time-consuming nature of histologic quantification of transplanted cells and the fact that few methods currently exist to rapidly screen transplanted cell survival. Establishment of the time frame in which transplanted cells die is necessary to temporally target cell survival interventions. In vivo imaging of luminescence can detect expression of reporters , antibodies , and transplanted cells (Okada et al., 2005;Chen et al., 2006;Kim et al., 2006;Roet et al., 2012), including a reduction in cells over time (Okada et al., 2005;Roet et al., 2012). In the current study, we use in vivo bioluminescence imaging to establish the time window of transplanted cell death following engraftment into the injured rat spinal cord. We also test the efficacy of both in vivo luminescence imaging and ex vivo fluorescence imaging as alternatives to the use of stereology for assessment of transplant survival.
To counteract the potentially deleterious effects of acute transplanted cell death, interventions that promote transplant survival and are amenable to clinical translation are needed. Historically, transplant survival approaches have focused on targeting single factors (Nakao et al., 1994;Mundt-Petersen et al., 2000;Karlsson et al., 2002;Hill et al., 2010). To date, the presence of multiple potential cell death inducers (e.g., hypoxia, oxidative stress, excitotoxicity, lack of substrate/adhesion/growth factors) and the complex cross-talk between cell death pathways has limited the efficacy of this approach. An alternative approach that has proven efficacious, and which does not require identifying the factors responsible for the acute cell death, is the activation of survival pathways. In the injured spinal cord, inclusion of growth factors (Lu et al., 2012;Robinson and Lu, 2017) or enhancement of growth factor signaling  is effective. In other cell transplantation models, mildly stressing the cells to pre-condition them before engraftment is effective (Murry et al., 1986;Theus et al., 2008;Yu et al., 2013). Although beneficial in preclinical models, implementation as part of a clinical-grade product may prove challenging. Both growth factor signaling and preconditioning activate prosurvival transcriptional programs. Directly engaging transcription factors could provide an alternative means to engage survival pathways.
Among the transcription factors implicated in adaptive responses to cellular stress are members of the hypoxiainducible factor (HIF) family. The HIFs are DNA-binding transcription factors that consist of an oxygen-labile ␣-subunit and a constitutively expressed ␤-subunit. By altering the cell's metabolic program and gene expression, HIF-␣ signaling allows cells to sense and adapt to environmental stressors such as decreased oxygen and nutrients. HIFs regulate the expression of genes involved in adaptive transcriptional responses to improve cell survival by enhancing intracellular ATP and oxygen levels and decreasing the production of reactive oxygen species (ROS) in times of hypoxic stress (Semenza, 2007). The genes regulated by HIF-␣ include key pathways needed for cell survival, such as: glucose metabolism/ATP production, oxygen transport and delivery, and cell growth and fate (Schödel et al., 2013). The diversity of genes regulated by HIF-␣ shows that its protective effects are not limited to hypoxic environments. In a variety of injury models, elevation of HIF-1␣ in transplanted cells enhances transplanted cell survival (Theus et al., 2008;Wu et al., 2010). Moreover, pharmacological manipulators of HIFs exist, which makes clinical targeting of this pathway viable.
In this study, to examine the effects of HIF-1␣ on transplant survival, Schwann cells (SCs) were used. In preclinical SCI models, extensive examination has established that they promote axonal growth and remyelinate axons. A comprehensive review of SC preclinical studies exists, comparing their effects with other cellular therapies (Tetzlaff et al., 2011). Furthermore, a completed phase 1 clinical trial established their safety in humans with SCI (Anderson et al., 2017).
The current study establishes the time window post-TP in which cells die following engraftment into the subacutely injured spinal cord. It tests the hypothesis that transplanted cells die early after implantation due to inadequate activation of transcription factor pathways necessary for cells to adapt to stress and survive. Using viral expression of a non-hydryoxylatable version of HIF-1␣, the effect of HIF-1␣ on cytoprotection is assessed in vitro and in vivo following transplantation into the injured spinal cord. In an effort to identify methods to facilitate the screening of transplant survival, this study also compares three different techniques for assessment of transplant survival in rats: histologic quantification, ex vivo fluorescent imaging of transplanted spinal cords, and in vivo bioluminescent imaging.

Viral manipulation of SCs
Green fluorescent protein (GFP) and luciferase (luc) were used to identify the transplanted cells. SCs were transfected with lentiviruses (LVs). LVs were produced by the Miami Project Viral Vector Core. Viruses were serially applied to the cells at the beginning of each passage. Cells were exposed to the viruses for 18 -20 h. LV containing enhanced GFP (LV-GFP; MOI 23) was applied to P 1 cells and resulted in 94.5% of SCs expressing GFP. LV containing luciferase (LV-luc; MOI 40) was added to P 2 cells already expressing GFP. A variety of LV-luc constructs were generated to detect light and examine HIF expression in SCs. High levels of luciferase activity were needed to generate enough light to penetrate through the muscle and skin above the transplant site. As a result, cells with the greatest luciferase expression were used. These cells expressed a control construct for examining HIF stabilization (ODD-luc-AYIA; Smirnova et al., 2010). To overexpress HIF-1␣ in SCs, P 3 SCs expressing GFP and luc were transfected with a retrovirus expressing HIF in which HIF transcriptional activity is enhanced by the inclusion of the VP16 transactivation domain (TAD; RV-VP16-HIF; MOI 10) or a control virus (VP16; MOI 10; Kung et al., 2000;Aminova et al., 2005). The viruses were previously generated by transfection of 293T cells with pBAB-puro-HIF-1␣-VP16 or a control plasmid (pBABEpuro-VP16; Harvard Gene Therapy Initiative) and kindly provided by Dr. Rajiv Ratan. The RV-VP16-HIF virus results in the expression of a fusion protein encoding amino acids (aa) 1-529 of HIF-1␣ and 78 aa of the VP16 TAD. The VP16 TAD is a potent TAD located within the carboxyl terminus of herpes simplex virus type 1 transcription factor VP16. When the VP16 TAD is fused to a transcription factor, it amplifies its activity. By creating a fusion protein with aa 1-529 of HIF-1␣ that contains the DNA binding domains of HIF-1␣, but lacking the oxygen degradation domain, this virus produces a HIF-1␣ fusion protein with both enhanced stability and transcriptional activity. Retroviruses were added in the presence of polybrene (4 g/ml). Cells were subsequently expanded and frozen. For experiments, SCs between P 4 and P 8 were used.

RNA isolation and PCR
RNA was collected from SCs by adding TRI reagent (Zymo Research) to the SCs and then isolating the RNA using the Direct-zol RNA Miniprep (Zymo Research) according to the manufacturer's instructions. To confirm viral expression of HIF, RNA was isolated. RNA was quantified using the Nanodrop spectrophotometer (Thermo Fisher Scientific). cDNA was generated using Superscript III First Stand Synthesis System (Thermo Fisher Scientific) and PCR was subsequently performed using primers for HIF-1␣, VP16 and ␤-actin (as per Aminova et al., 2005) and PCR products identified following electrophoresis.

Protein isolation and western blotting
Nuclear and cytoplasmic proteins were isolated from cultured SCs using the NE-PER kit (Thermo Fisher Scientific), as per the manufacturer's instructions. Cells were collected in CER I buffer with protease inhibitors by mechanically scraping. During isolation of the nuclear fraction, NaCl (200 mM; Sigma-Aldrich) was added at 10 min to facilitate release of bound nuclear proteins and 1 l (Ն250 U) of benzonase nuclease (Sigma-Aldrich) was added at 20 min to digest nucleotides. Protein concentration was determined by DC assay, as per the manufacturer's instructions (Bio-Rad). For western blotting, the protein samples were quantified and either used immediately or stored at -80°C. Protein samples were heatdenatured (95°C, 5 min), mixed with 6x loading buffer with SDS (Boston BioProducts), and loaded onto a 4 -15% gradient precast SDS-PAGE gel (Bio-Rad). Using electrophoresis, proteins were separated (120 V, 1.5 h, RT, running buffer; Boston BioProducts), and subsequently transferred to a nitrocellulose membrane (100 V, 1.5 h, 4°C, transfer buffer; Boston BioProducts) with 20% methanol. Nuclear protein (50 g), was assessed for HIF-1␣ (1:1000, Novus NB 100-105) and HIF-2␣ (1:500, Novus NB 100-122). Nuclear protein (20 g) was assessed for VP16 (1:1000, ab4808, Abcam). Cytoplasmic protein (20 g) was assessed for VEGF (1:500, NB100-2381, Novus Biologicals) and enolase (1:1000, NB100-65252, Novus Biologicals). ␤-Actin (1:10,000, Sigma-Aldrich) or GFP (1: 1000, G6539, Sigma-Aldrich) was used as a loading control. Antibodies were diluted in Odyssey blocking buffer (LI-COR). Primary antibodies were incubated overnight at 4°C. LI-COR secondary antibodies (1:10,000; IRDye 800CW, IRDye 680LT) were incubated for 90 min at room temperature. Membranes were imaged and band intensities were quantified using the Odyssey Infrared Imaging System (LI-COR). For each antibody, a minimum of three separate samples per condition were run and analyzed from the same blot. For some experiments, SCs were treated for 24 h with 200 M deferoxamine (Sigma-Aldrich), a known HIF stabilizer. Samples from these cells were used as a positive control for localization of HIF proteins on HIF western blots.

In vitro cell death assays
To assess the effects of HIF on SC survival in vitro, LV-GFP-luc-VP16 HIF SCs (VP16-HIF SCs) or LV-GFPluc SCs (control) were plated onto 96-well plates and grown for 41-47 h. Medium was then removed, and fresh D-10ϩ3M was added along with various inducers of cell death. Hydrogen peroxide (H 2 O 2 ; 1-1000 M, Sigma-Aldrich) was added to dishes in which 10,000 SCs were plated to induce oxidative damage. SC survival was assessed 3 h later. To determine the impact of other activators of cell death, 25,000 cells per well were plated and SC survival was assessed 24 h later. Tunicamycin (0 -8 M, Sigma-Aldrich) was added to activate the unfolded protein response. Thapsigargin (0 -6 M, Sigma-Aldrich) was added to induce calcium release from intracellular stores. Camptothecin (0 -12 M, Sigma-Aldrich) was added to induce DNA damage. SC survival was assessed by MTS assay (Promega). Phenazine methosulfate (PMS) was added to tetrazolium compound: 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium, inner salt (MTS) immediately before addition to the cells, as per the manufacturer's instructions. The medium was replaced by MTS solution either 3 h (H 2 O 2 assays) or 24 h (all other assays) after treatment. Cells were incubated with MTS at 37°C, 6% CO 2 for 4 h before absorbance was read using the SpectraMax i3 cytometer/spectrophotometer (Molecular Devices) at 490 nm. Percentage of cell survival was calculated for each condition per plate relative to untreated cells of the same type (VP16-HIF or control SCs). Briefly, the percentage of cells surviving in each well was calculated [(well absorbance -blank absorbance)/average absorbance of untreated cells]. Next, for each treatment for a given cell type on a given plate, the mean for the technical replicates was determined to establish the percentage survival for each condition (i.e., each independent sample). Finally, percentage of cell survival per condition was calculated by averaging the results from each independent sample. Results represent the means of three to four independent assays in which the technical replicates (three to four/ replicate) were averaged. For all assays, 2 l of lysis solution (9% weight/volume Triton X-100; Thermo Fisher Scientific) was added to assess maximal death.

SCI and transplantation
Female Fischer 344 rats (Harlan, barrier 217, 9 -11 weeks of age) were anesthetized with isoflurane (2-3%). To expose the spinal cord, a laminectomy was performed to remove the dorsal process of thoracic vertebrae 9 (T9). The lateral processes of T8 and T10 were clamped and a 200-kdyn injury was induced using the Infinite Horizon impactor (Precision Systems and Instrumentation). The impact curve was checked for hit quality at the time of injury and saved. The injury site was inspected for a bruise before suturing the muscles closed in two separate anatomic layers. The skin was closed with wound clips. The temperature of the rats was monitored and maintained throughout the surgery using a thermoregulated heating pad. Buprenorphine (0.05 mg/kg) was given twice a day for the first 2 d postsurgery to alleviate pain. Lactated Ringer's (10 ml) was given one to two times per day for the first 2 d postsurgery to prevent dehydration. Gentamycin (5 mg/kg) was given once a day for the first 7 d postsurgery to prevent infections. Along with food and water ad libitum, wet food pellets were provided to help maintain the rat's weight. Rats were housed in pairs, or in groups of three, and cages were placed on thermoregulated heating pads (half on-half off) to assist with thermoregulation for the first week postsurgery.
Seven days after SCI, rats underwent a second surgery to transplant the cells. Rats were anesthetized with isoflurane (2-3%) or ketamine and xylazine (80 mg/kg:10 mg/kg). The anesthetic used was kept consistent for each experiment. The previous incision site was reopened and the laminectomy site re-exposed. The dorsal vertebral process of T8 was clamped to stabilize the rat and 6 l of cells was injected into the injury site. Cells were injected using a pulled glass capillary tube with silicone plug attached to a Hamilton syringe affixed to a nanosyringe pump (KD Scientific) attached to a stereotactic device. Cells were injected at a rate of 1 l/min at a depth of 1-1.25 mm. The capillary was left in place for an additional 3 min to allow the pressure to equilibrate before removal. The laminectomy site was then sutured closed in two anatomic layers and the skin was closed with wound clips. Surgical and postsurgical care (e.g., temperature monitoring, drugs) was administered as after the initial SCI surgery. All animal procedures were performed in accordance with the Weill Cornell Medicine animal care committee's regulations.

Optimization of in vivo bioluminescent imaging (IVIS)
To examine transplant survival in rats over time, and to assess whether IVIS imaging could be used to measure transplant survival more rapidly than stereological quantification, changes in light emission over the first 7 d following transplantation were quantified using the IVIS 100 (PerkinElmer). Initial experiments examined the sensitivity of the IVIS for detecting different numbers of transplanted cells and compared intraperitoneal injection of D-luciferin with intravenous injection following transplan-tation of 2 ϫ 10 6 GFP-luc SCs. Intravenous administration via the tail vein resulted in a more rapid increase in light production than intraperitoneal administration (data not shown). The time course of the decrease in light emitted from the transplants over time was compared to the number of transplanted cells at 3 d (n ϭ 4) and 7 d (n ϭ 5). A subset of these rats were quantified by stereology to determine the number of surviving SCs at 3 d (n ϭ 3) and 7 d (n ϭ 4). Spinal cords from two of the rats that underwent IVIS imaging were not available for histologic analysis.

Transplantation of HIF-1␣ SCs and assessments of transplant survival
To assess whether increasing HIF-1␣ in SCs promoted transplanted SC survival, the effect of overexpressing HIF was tested in a fully-blinded, randomized, in vivo SCI experiment. Both the treatment group assignment and the order of the cell types transplanted were assigned randomly before cell collection and transplantation. Twentysix rats underwent a T9, 200 kdyn IH SCI (one rat died post-SCI). The 25 remaining rats received SC transplants 7 d post-SCI and were euthanized at 7 d post-TP (14 dpi). Rats received transplants of 2 ϫ 10 6 SCs directly into the injury epicenter. The three transplant treatment groups were: GFP-luc SCs (control SCs, n ϭ 9), GFP-luc-VP16-SCs (VP16 SCs, n ϭ 8), GFP-luc-VP16-HIF-1␣ SCs (VP16-HIF SCs, n ϭ 8). Using up to three different methods, transplant survival was assessed in the same rats. Following transplantation, 12 of the 25 rats (n ϭ 4/group) underwent in vivo bioluminescent imaging using the IVIS Spectrum (PerkinElmer) to detect the transplanted cells based on their luciferase expression. In vivo bioluminescent imaging was performed on these 12 rats daily for the first 3 d post-TP, the time window in which we determined that transplanted cell death occurs. Seven days post-TP, after perfusion and isolation of the spinal cords (n ϭ 25), ex vivo GFP fluorescent imaging of the spinal cords was performed using the IVIS Spectrum. The spinal cords were then sectioned with a cryostat and the number of surviving SCs was quantified by Stereo Investigator (MBF Bioscience). By sequentially assessing the same spinal cords, we were able to compare non-biased stereology, which reliably detects differences in transplant survival, with alternative methods of transplant survival quantification. In this study, we compared the results of the stereological quantification of surviving SCs with ex vivo fluorescent imaging of light emission from GFP and in vivo bioluminescent imaging of luciferase activity within a single set of experimental rats. On histologic inspection of the 25 spinal cords, 21 rats were determined to have received good transplants (control SCs n ϭ 7; VP16 SCs n ϭ 6; VP16-HIF SCs n ϭ 8) and were included in the stereological and ex vivo imaging analyses. For in vivo bioluminescence, all imaged rats that were determined to have received good transplants were included in the analysis (control SCs n ϭ 3; VP16 SCs n ϭ 4; VP16-HIF SCs n ϭ 4).
Exclusion of rats from the analysis was based on assessment of the quality of the transplant, which was undertaken before unblinding of the experiment. The transplants were assessed histologically for tissue section completeness and location of the transplanted cells in both the rostral-caudal and dorsal-ventral axes relative to the lesion epicenter. The stereological results were then compared with the transplant surgery notes and notes on the histologic assessment of the transplants. Two rats with small transplants were excluded because of problems with the injection that were noted at the time of transplantation. Two additional rats were excluded because the transplant was not located within the lesion epicenter.

IVIS luminescent imaging
IVIS luminescent imaging was performed using either the IVIS 100 or IVIS Spectrum. Rats were anaesthetized with isoflurane (1-2%; Henry Schein). Baseline images were acquired before intravenous injection of D-luciferin (150 mg/kg, Gold Biotechnology) into the tail vein. IVIS imaging was performed daily as close as possible to the time of transplantation each day. Following administration of D-luciferin, IVIS images were acquired by collecting the amount of light emitted over 5 min (exposure: 5 min; binning: medium; F/stop: 1; field of view: 400 cm 2 ; emission filter: open). Rats were imaged continuously in 5-min intervals for up to 60 min to determine the time of maximum light emission. The image with maximum light was used to quantify the amount of light emitted. A region of interest (ROI) of 6.41 cm 2 was centered over the site of maximum light and both the average radiance (photons/ s/cm 2 /steradian) and the total flux (photons/s) were determined. Because total flux is derived from average radiance, results from both showed a similar profile. Average radiance is presented. In cases where light was not initially detected within 0 -30 min, rats received a second injection of D-luciferin and were re-imaged. If, following a second injection, light was not detected, the data were excluded from analysis for that time point only.

Ex vivo IVIS fluorescent imaging
The amount of fluorescent light emitted from the GFP transplants was quantified in the isolated fixed spinal cords using ex vivo imaging with the IVIS Spectrum. Spinal cords were placed on a Petri-dish with a black backing and inserted into the IVIS Spectrum. The number of photons emitted by the GFP ϩ SCs within the spinal cord was determined using the following IVIS settings: excitation/emission ϭ 500/540; exposure ϭ 0.5 s; binning ϭ 4, F-stop ϭ 2; field of view ϭ B. A photographic image of the spinal cord overlaid with the photon intensity count was generated. The ROI of GFP fluorescence was then auto-sized and the average radiant efficiency [(p/s/cm 2 / sr)/(W/cm 2 )] determined for each case. To account for variability in light intensity between images, the same settings were used to show visible photons on all the images (min ϭ 3.25 ϫ 10 8 ; max ϭ 10 ϫ 10 9 ).

Histology and quantification of SC survival and transplant volume
A 15-mm block of the spinal cord containing the injury epicenter was cut into longitudinal sections (20 m) on the cryostat. Four sets of serial sections were collected onto charged slides. Nuclei were labeled by incubating tissue sections with 1:1000 Hoechst (Sigma-Aldrich) in PBS (pH 7.4) for 1 h. Slides were air dried and coverslipped with Vectashield (Vector Laboratories) before imaging. To quantify the total number of transplanted GFP ϩ SCs and the GFP ϩ transplant volume, Stereo Investigator was used. The transplant was outlined at 10ϫ magnification. Using the Cavalieri function, transplant volume was quantified at 10ϫ magnification (grid size was 200 ϫ 200 m). Using the optical fractionator function, the number of GFP ϩ cells were quantified at 63ϫ magnification (grid box size was 150 ϫ 400 m and the sampling box was 50 ϫ 50 m).

Imaging
Tissue was examined, imaged, and quantified using either a Zeiss Axiovert 200M or a Zeiss AxioImager M2 equipped with Stereo Investigator. Virtual tissue sections were acquired at 10ϫ magnification. Similar Exposure and Gain settings were maintained for all image acquisition for a given experiment. Confocal images were acquired using a Zeiss LSM 510 META confocal microscope.

Chemical and biosafety
All viral work was performed using biosafety level 2 procedures and approved by the Weill Cornell Medicine Institutional Biosafety Committee. Individuals working with viruses and chemicals received institutional biosafety and chemical training and donned appropriate personal protective equipment when executing experiments with hazardous material.

Statistics
All statistics were performed using SPSS (version 22; IBM). An overview of statistical tests used can be found in Table 1. Specifics of the statistical tests used are included in the results.

Transplanted cell number does not change between 2 and 7 d post-TP
Transplanted cells die early after transplantation into the injured spinal cord (Barakat et al., 2005;Hill et al., 2006Hill et al., , 2007Karimi-Abdolrezaee et al., 2006). However, the exact time course of transplanted cell death remains to be established. To counteract transplanted cell death, it is necessary to establish the window in which the death occurs. Several studies have shown that the amount of light detected using bioluminescent imaging correlates with cell number transplanted and that the light decreases following the transplantation of cells (Okada et al., 2005;Takahashi et al., 2011;Roet et al., 2012;Nishimura et al., 2013). Using the IVIS imaging system, bioluminescent imaging was performed for either 7 d (n ϭ 4) or 3 d (n ϭ 3) to establish when post-TP transplanted cells die. Following intravenous injection of D-luciferin via the tail vein, it took 5-10 min for the amount of detected light to reach the maximal level, as measured by average radiance. The maximal amount of light produced by the luciferaseexpressing SCs decreased significantly over time post-TP (univariate ANOVA: F (6,36) ϭ 3.302, p ϭ 0.011 a ). The amount of light detected at day 2 was significantly lower than on day 1 (K matrix: 1 d vs 2 d, p ϭ 0.004 a ). After day 2, no further reduction in light production was detected. The amount of light detected was similar from 2 to 7 d To confirm that the number of surviving transplanted cells did not differ after the first few days, when the light levels in the IVIS had plateaued, the spinal cords from the IVIS imaged rats were examined histologically. The number of GFP ϩ SCs within the transplants were quantified at 3 d (n ϭ 3) and 7 d (n ϭ 4) post-TP. Stereological quantification confirmed that the number of surviving transplanted SCs did not differ between 3 d and 7 d post-TP (t-test, df: 5, p ϭ 0.8 b ; Fig. 1B). This fits with previous work in which necrosis and apoptosis of transplanted SCs was detected at 24 h but not 3 d post-TP (Hill et al., 2007). These experiments led us to conclude that the death of the cells occurs within a narrow time window immediately post-TP and concludes within 2-3 d.
Multiple factors are postulated to contribute to transplanted cell death (e.g., cell processing, transplant procedure, injury/transplant environment, transplant rejection). Examination of the factors that contribute to transplanted cell death indicate that the injury/transplant environment is the likely culprit (Hill et al., 2006(Hill et al., , 2007(Hill et al., , 2010. Cell processing and the transplant procedure account for Ͻ10% of the observed cell death (Hill et al., 2007). In a model of transplant rejection, immunosuppression does not block this early death (Hill et al., 2006). Transplant survival is enhanced by delaying engraftment for at least 7 d (McDonald et al., 1999;Hill et al., 2006), but is not further augmented by waiting for the chronic injury site to develop (Barakat et al., 2005). This suggests that the environment generated by the cell transplant is a key contributor to acute transplanted cell death, and likely why cell death is a common feature of all cell transplants.

Expression of VP16-HIF-1␣ in SCs increases HIF-1␣, but not HIF-2␣, and increases protein expression of HIF target genes
Although the exact cause of this death has not been established, the complexity and redundancy of the pathways involved in tissue damage after SCI support targeting coordinated responses instead of a single gene, protein, or pathway to improve transplant survival. Activation of transcriptional programs leads to alterations in targeted downstream cassettes of genes that act in concert to alleviate the relevant stress(es) and their byproducts. Harnessing the power of cellular transcriptional programs could abrogate the need to identify and target individual cell death inducers.
HIF-1␣ is a member of a family of transcription factors that regulate transcriptional responses involved in cellular metabolism and wound healing (Ruthenborg et al., 2014;Yang et al., 2014;Pugh, 2016). HIF target genes are involved in cell growth and fate, mitochondrial functioning, glycolysis and glucose metabolism, and oxygen transport and delivery (Semenza, 2012), all of which could benefit transplanted cell survival. In the current series of experiments, we sought to elevate HIF-1␣ in SCs and test whether it could promote the survival of the transplanted cells.

Elevation of HIF in SCs via retroviral expression of VP16-HIF enhances SC survival in response to oxidative stress but not ER stress or DNA damage
The transcriptome regulated by HIF contains a core set of genes activated in response to hypoxia (Benita et al., 2009), along with cell-specific transcriptional changes (Chi et al., 2006). This generates both specificity and diversity in the transcriptional response (Lendahl et al., 2009). Activation of HIF adaptive signaling is primarily associated with cytoprotection, as evidenced by its protective effect on transplanted cells when elevated either directly or indirectly by induction with hypoxia or pharmacological preconditioning (Chu et al., 2008;Theus et al., 2008;Wu et al., 2010). However, among HIF's target genes are BNIP3 and NIX, which are associated with mitochondrial damage, apoptosis and autophagy (Ney, 2015). Moreover, in a neuronal cell line in vitro, HIF expression augments glutamate-induced oxidative stressmediated cell death (Aminova et al., 2005). Thus, at least in some contexts, elevation of HIF-1␣ can lead to enhancement of cell death.
To determine whether expression of HIF-1␣ was cytoprotective for SCs against oxidative stress-induced cell death, the effects of overexpression of HIF-1␣ on cell survival was assessed in vitro by MTS assay. To model oxidative damage in vitro, SCs were exposed to H 2 O 2 (0 -1000 M) for 3 h. This dose range resulted in a dosedependent reduction in SC survival from 0 to 150 M (31.5 M), 45.2 Ϯ 4.58% of control SCs survived, whereas 62.8 Ϯ 4.64% of VP16-HIF SCs survived. Expression of HIF-1␣ increased SC survival by 39%. This is equivalent to the protection achieved by calpain inhibition (Hill et al., 2010). Therefore, expression of HIF-1␣ in SCs was sufficient to protect SCs against oxidative stressmediated cell death. These results were expected, given HIF-1␣'s established role in mitigating ROS-mediated damage (for review, see Thomas and Ashcroft, 2019), but were contrary to the results found in a neuronal cell line (Aminova et al., 2005).
Under some contexts, HIF expression can be deleterious (Aminova et al., 2005). The effects are both cell type (Vangeison et al., 2008) and injury inducer-dependent (Aminova et al., 2005). As the inducers of transplanted cell death remain to be identified, we further examined the survival of SCs by evaluating their survival in vitro. Several additional known inducers of cellular toxicity, including: DNA damage, ER stress, and protein unfolding were examined to establish the context in which HIF-1␣ leads to SC cytoprotection.
To assess the ability of HIF to protect SCs against DNA damage, SCs were exposed to camptothecin (0 -12 M) in vitro for 24 h (Fig. 3B). Camptothecin significantly reduced the viability of SCs across the range of concentrations tested (ANOVA: F (6,28) ϭ 8.5, p Ͻ 0.0001 i ). However, despite using a range of concentrations effective in other cells in vitro (Aminova et al., 2005), most of the SCs (80 -87%) remained viable. Only 0.375 M camptothecin treatment differed from the preceding concentration (K matrix: 0 vs 0.375 M, p Ͻ 0.0001, all other contrast p Ͼ 0.05; 0.375-12 M). The lack of a dose-response curve across a range of concentrations that are toxic for HT22 cells indicates that SCs are relatively resistant to DNA damage-induced death via camptothecin. Although the effect of camptothecin on SC viability was modest, HIF expression protected SCs against camptothecinmediated cell death (ANOVA: F (1,28) ϭ 13.7, p ϭ 0.001 j ). Viability of VP16-HIF SCs was significantly enhanced in response to application of 3 or 12 M camptothecin (post hoc: Fisher's LSD, 3 M, p ϭ 0.002, 12 M, p ϭ 0.014). At 3 M, HIF-1␣ increased cell survival by 17.7% (79.4 Ϯ 1.38% of control SCs survived compared to 93.5 Ϯ 7.31% of HIF-1␣ SCs). At 12 M, HIF-1␣ increased survival by 12.5% (83.9 Ϯ 1.35% of control SCs survived compared to 94.4 Ϯ 2.34% of HIF-1␣ SCs). Thus, although SCs are relatively resistant to DNA damageinduced cell death, elevating HIF is still sufficient to afford some protection to SCs against DNA damage-induced death.
ER stress was induced by exposure of SCs to thapsigargin (0 -6 M) for 24 h (Fig. 3C), which results in Ca ϩϩ release from intracellular stores. Thapsigargin significantly altered SC viability across a range of concentrations (ANOVA: F (5,24) ϭ 9.26, p Ͻ 0.0001 k ); however, across concentrations proven toxic for HT22 cells (Aminova et al., 2005), SCs did not show a dose-response curve and an LD50 was not obtained. Thapsigargin decreased cell viability to 88 -93% of control at high doses (Ͼ1.5 M). Unexpectedly, at low doses (0.375-0.75 M), thapsigargin increased cell viability to 107-117% of control (0 M). Although not tested, this is likely due to altered SC proliferation. Only application of 0.375 and 1.5 M thapsigargin resulted in changes in viability relative to the preceding concentration (K matrix: 0 vs 0.375 M, p ϭ 0.002; 0.75 vs 1.5 M, p ϭ 0.02). Based on the relatively small decrease in SC viability across a range of thapsigargin concentrations that are toxic to other cells, it appears that SCs were relatively resistant to ER stress induced by thapsigargin. Elevating HIF in SCs did not alter SC viability in response to thapsigargin (ANOVA: F (5,1) ϭ 1.5, p ϭ 0.225 l ). Larger alterations in SC survival in response to ER stress are needed to establish whether HIF-1␣ levels impact ER stress-mediated cell death in SCs.
Accumulation of misfolded or unfolded proteins can lead to cell death. To induce protein unfolding in SCs, tunicamycin (0 -8 M) was administered for 24 h (Fig. 3D). Administration of tunicamycin induces autophagy, which can either be protective, or, if autophagy is excessively activated or disrupted, lead to cell death (Ding et al., 2007). Tunicamycin administration resulted in a dosedependent decrease in SC viability (ANOVA: F (5,24) ϭ 82.7, p Ͻ 0.0001 m ), but only at the lowest concentrations tested. There was a significant decrease in viability from the preceding tunicamycin concentration at 0.5 and 1 M (K matrix: 0.5 vs 0 M, p Ͻ 0.005; 1 vs 0.5 M, p ϭ 0.04), Beyond 1 M, at concentrations that are toxic to HT22 cells (Aminova et al., 2005), 59 -63% of SCs continued to survive (Fig. 3D), indicating that SCs were relatively resistant to tunicamycin treatment. Expression of HIF-1␣ in VP16-HIF SCs did not counteract the reduction in SC viability in response to tunicamycin. Rather, VP16-HIF SCs had a small reduction in viability compared to control SCs (ANOVA: F (1,24) ϭ 9.7, p ϭ 0.005 n ). This significant difference was detected at only a single concentration (post hoc: Fisher's LSD, 1 M, p ϭ 0.04). At 1 M tunicamycin, HIF expression decreased SC viability by 13.3% (62.7 Ϯ 1.11% of control SCs survived compared to 54.5 Ϯ 2.69% of VP16-HIF SCs). Although a significant decrease in survival was observed when HIF-expressing SCs were treated with tunicamycin, the effect was restricted to a small change at a single dose of the five doses tested. Overall, SCs were relatively resistant to cell death induced by activation of the unfolded protein response. This likely reflects the importance of autophagy in the de-differentiation and reprograming SCs (Gomez-Sanchez et al., 2015).
Previous reports indicate that the effects of HIF-1␣ on cell survival are influenced by both the cell type and cell death mechanism (Aminova et al., 2005). In SCs, the in vitro assays indicate that elevation of HIF via retroviral expression of VP16-HIF-1␣ results in substantial protection against oxidative stress (38.9% increase). In other models of cell stress, the ability of elevated HIF-1␣ in SCs to protect SCs was more variable and of a limited magnitude (Ͻ20% increase or decrease). In SCs, HIF-1␣ resulted in modest protection (DNA damage), no alteration (ER stress), or a reduction (unfolded protein response) in survival. These results fit with the overall view that activation of HIF-1␣ adaptive pathways is cytoprotective, but that its effects are context dependent.

Elevation of HIF in SCs via retroviral expression of VP16-HIF promotes transplanted SC survival 7 d post-TP
Having established that expression of HIF-1␣ in SCs led to elevated levels of target genes in adaptive HIF pathways (Fig. 2), and that this protected SCs against oxidative stress-induced cell death in vitro (Fig. 3A), we tested whether elevating HIF in SCs enhances their survival following transplantation into the injured spinal cord.
HIF is associated with enhanced migration of cancer cells (Araos et al., 2018) and neural crest cells (Barriga et al., 2013). Larger transplants could arise from better transplant survival and/or enhanced transplant migration. No evidence for migration of the SCs out of the transplants and across the SC-astrocyte boundary were detected in this study. The presence of astrocytes and inhibitory molecules within the glial scar is known to prevent SC migration (Afshari et al., 2010).
In this study, we sought to test whether the elevation of HIF-1␣ and activation of its transcriptional programs could provide an alternative approach to the use of preconditioning or growth factor augmentation for promoting the survival of the transplanted cells. Elevating the expression of HIF-1␣ by 5.9-fold increased the expression of the HIF target genes VEGF and enolase by 2.2 and 1.7-fold, respectively. This level of HIF activity was associated with a 20 -35% increase in transplant survival. Thus, we demonstrate that overexpression of a single transcription factor is sufficient to protect SCs transplanted into the injured spinal cord. Further optimization is needed to obtain the level of protection afforded by HIF-1␣ in other transplant models either alone (Wu et al., 2010;Wakai et al., 2016) or with the inclusion of growth factors along with cell transplants into the injured spinal cord Lu et al., 2012;Robinson and Lu, 2017).

Ex vivo fluorescent imaging of spinal cords and in vivo luminescent imaging of rats did not detect differences in transplant survival
A barrier to the development and identification of new strategies that promote cell survival are the methods currently used to assess transplant survival accurately. As part of this study, we assessed whether ex vivo fluorescent imaging (Fig. 3H,I) or in vivo bioluminescence imaging (Fig. 3J,K) could detect a difference in transplant survival. Similar to the earlier experiment, IVIS imaging was sensitive enough to detect a difference in light emission over time (ANOVA: F (8.89,1.111) ϭ 11.955, p ϭ 0.006 q , sphericity corrected; post hoc: Bonferroni, d1 vs d2, p ϭ 0.01, d1 vs d3, p ϭ 0.031). Neither IVIS imaging of live animals, nor ex vivo imaging of isolated spinal cords, was sensitive enough to detect the increase in transplant survival observed histologically (ex vivo IVIS GFP: ANOVA: F (18,2) ϭ 2.448, p ϭ 0.115 r ; in vivo IVIS luciferase: ANOVA: F (8,2) ϭ 0.032, p ϭ 0.969 s ). Based on power analyses, a minimum of 14 -18 rats would be needed to detect a significant difference (power ϭ 0.8, ␣ ϭ 0.05) between the three groups using either luminescence or GFP as an outcome on the IVIS, respectively. Although ex vivo and in vivo imaging of the transplants is feasible; a large number of samples would be needed to reliably detect differences. Thus, these methods are unlikely to be a more rapid alternative to stereological quantification of transplanted cells.

Discussion
The results of the current studies establish that transplanted cells die before 2 d post-TP. The narrow window of cell death followed by the stabilization of the number of transplanted cells supports that acute, rather than prolonged, manipulation of survival pathways could be sufficient to counteract transplanted cell death. HIF-1␣ is involved in cellular adaptations to stress. Here, we demonstrate that the elevation of HIF-1␣ increases cell survival in both an in vitro model of transplanted cell injury and following transplantation of SCs into the subacutely injured spinal cord. We show that stabilization of HIF-1␣ elevates HIF transcriptional targets and is sufficient to protect SCs against oxidative stress and DNA damageinduced cell death, two major mechanisms of cell death following SCI (Ahuja et al., 2017). When HIF-1␣expressing SCs are transplanted into the injured spinal cord, where multiple inducers of cell death are present (Ahuja et al., 2017), more cells survive. These experiments support the utility of harnessing cellular adaptive responses to protect them from subsequent transplantinduced cellular stress. They also establish that targeting a transcription factor (HIF-1␣) and activation of its target pathways is cytoprotective for cells transplanted into the injured spinal cord. This identifies an alternative approach to those currently used which target individual cell death inducers or signaling pathways, or involve the inclusion of multiple growth factors along with the transplanted cells.
Importantly, strategies exist to transiently elevate HIF-1␣ transcriptional programs (e.g., hypoxic preconditioning or pharmacological pretreatment of transplanted cells), which, if effective, will enable the elevation of protective transcriptional programs using a pretreatment approach. Pretreatment of the cells could provide a more clinicallyfeasible method for enhancing transplant survival than current pro-survival approaches, which require inclusion of multiple proteins along with the cells.
Histologic quantification of transplant survival is laborious. The identification of optimal transplant conditions would greatly benefit from methodological advances that are both sensitive and facilitate reliable quantification of transplanted cell survival. Several new methods exist for quantifying transplant survival, including in vivo and ex vivo bioluminescence and fluorescence. Having established that HIF-1␣ increased transplant survival using stereological quantification, we assessed the ability of ex vivo fluorescent imaging and in vivo bioluminescent imaging to detect the improvement in survival. Neither approach detected the improvement in transplant survival. Thus, for rat spinal cord transplants, histologic quantification using non-biased stereology remains the most sensitive and reliable method for determining differences in transplant survival.

Transplanted cells die within the first 2-3 d when transplanted into the subacutely injured spinal cord
Several different cell types are currently in human testing for SCI repair (clinicaltrials.gov). Initial reports from human clinical trials indicate that cell transplants are safe (Tabakow et al., 2013;Shin et al., 2015;Anderson et al., 2017) but that the functional effects of cell transplants alone are limited. Ultimately, the effects of cell transplants will depend on both the cell type transplanted and the summation of the changes (both beneficial and potentially detrimental) that occur within the transplanted tissue. Because transplants persist long term, an often-overlooked concern which may impact transplant and tissue function is the acute death of the majority of transplanted cells (Barakat et al., 2005;Hill et al., 2006Hill et al., , 2007Karimi-Abdolrezaee et al., 2006).
To both maximize transplant efficacy and design superior anti-cell death interventions, establishing when this death occurs is needed. Here, using daily in vivo bioluminescence imaging and histologic confirmation, we narrow the time window of transplanted cell death to the first 2 d post-TP (Fig. 1). The current reduction in bioluminescence following SC transplantation parallels both the decrease in transplanted cell number (Barakat et al., 2005;Hill et al., 2007;Pearse et al., 2007) and bioluminescence determined in previous rodent spinal cord transplant experiments (Okada et al., 2005;Kumagai et al., 2009;Takahashi et al., 2011;Ozdemir et al., 2012;Roet et al., 2012;Nishimura et al., 2013;Iwai et al., 2014). Although our bioluminescence results are similar to previous reports, previous studies generally report a slightly more prolonged decrease in bioluminescence, where light levels plateau by 4 dpi (Okada et al., 2005;Ozdemir et al., 2012), 7 dpi (Nishimura et al., 2013), or 15 dpi (Roet et al., 2012) compared to the 2 d observed here. Our results with SC grafts in subacute rat contusion SCIs most resemble those of neural stem cell/progenitor cell grafts in subacute mouse contusion SCIs, where bioluminescent activity decreases by ϳ80% and plateaus by 4 d post-TP, the earliest time point presented (Okada et al., 2005;Nishimura et al., 2013). Differences in injury model, cell type, transplant location, mechanism of luciferase expression, and time of transplantation between the studies likely contribute to the variability in the results reported between this and previous studies. Collectively, this and previous studies assessing transplant survival/death establish that transplant death occurs in all cell transplants, and that the death occurs in a narrow, acute window immediately post-TP.
The largest contributor to transplant death is likely the transplant environment (Hill et al., 2006;Nishimura et al., 2013;Piltti et al., 2013). SCI results in complex biochemical and cellular changes that could contribute to the death of the transplanted cells, including: hypoxia, ischemia, oxidative and nitrosative stress, inflammation and immune mediators, decreased growth factors, and an altered extra-cellular matrix (Ahuja et al., 2017). This has led to the use of a multimodal transplant paradigm that includes a substrate, multiple growth factors, and a pharmacological inhibitor of cell death along with the cells (Lu et al., 2012;Robinson and Lu, 2017). Although effective, clinical translation of this approach will be difficult. The acute window of death determined here indicates that if interventions that protect the cells acutely can be identified, long-term inclusion of multiple factors is unlikely to be required.
In vitro, HIF-1␣'s effects on cell survival depend on the mechanism of cell death induction (Fig. 3A-D). This is similar to previous work in neurons Aminova et al., 2005). Although HIF-1␣ is generally considered protective, among its transcription targets are pro-death genes linked to apoptosis and autophagy (Chen et al., 2009) and, under some circumstances, HIF-1␣ augments cell death (Aminova et al., 2005;Vangeison et al., 2008). We observed that elevation of HIF-1␣ protects SCs against oxidative stress (Fig. 3A) and DNA damage (Fig. 3B), is slightly detrimental on activation of the unfolded protein response (UPS; Fig. 3D), and has no effect on ER stress (Fig. 3C). With the exception of the response to oxidative stress, the magnitude and breadth of the survival changes in SCs in response to the inducers of cell death tested was limited. This is in contrast to neurons, where HIF-1␣ affords substantial protection against DNA damage, ER stress, and UPS activation, but is pro-death in response to glutamateinduced oxidative stress (Aminova et al., 2005). It is wellestablished that cellular sensitivity to cell death and HIF-1␣'s effects are cell type and context-dependent (Chen et al., 2009). Compared to neurons (Aminova et al., 2005), SCs were relatively resistant to DNA damage, ER Ca ϩϩ release and induction of autophagy by initiation of the unfolded protein response. This could reflect a greater adaptability of SCs to harsh environments and/or the ability of SCs to modify their phenotype (e.g., SCs undergo substantial remodeling when they de-differentiate following peripheral nerve injury). This could account for the slightly higher survival rate of SCs (Barakat et al., 2005;Hill et al., 2007;Pearse et al., 2007) than neurons or neural progenitors (Barker et al., 1996;Karimi-Abdolrezaee et al., 2006) following transplantation. It is postulated that mild hypoxia leads to expression of adaptive HIF-1␣ responses, whereas severe or prolonged exposure leads to the expression of pro-death genes . The greater adaptability of SCs to harsh environments could also account for the differences between SCs and neurons exposed to oxidative stress. The specific cell death inducers that impact transplant survival remain to be established. Further studies to identify and test the mediators of transplanted cell survival/death are necessary to establish the context under which HIF-1␣ and other prosurvival interventions affect transplant survival in a cell and environment-specific manner.
Elevation of HIF-1␣ in SCs protects SCs against oxidative damage (Fig. 3A), to which SCs have been shown to be sensitive (Hill et al., 2010). In response to H 2 O 2 treatment, HIF-1␣ protected SCs to a similar level as inhibition of calpain-mediated cell death (Hill et al., 2010). In the vestibular system, SCs elevate HIF in response to oxidative stress, but it is not sufficient to prevent death (Riva et al., 2007), suggesting that the endogenous adaptive response of SCs to oxidative stress is insufficient to counteract the oxidative damage. Similarly, examination of the levels of HIF-1␣ in the spinal cord following transplantation indicate that SCs fail to initiate the HIF-1␣ adaptive response within the first 8 h after transplantation (unpublished data). Blocking lipid peroxidation, a consequence of oxidative damage, in vitro protects SCs but is insufficient to protect a variety of transplanted cells (Karlsson et al., 2002), including SCs (Hill et al., 2010). It is possible that counteracting oxidative stress is the wrong target for transplant protection. Alternatively, current treatments may not effectively reduce the generation of ROS and subsequent oxidative stress. As HIF-1␣ is a regulator of diverse biological pathways, including oxygen supply and utilization (Dengler et al., 2014), it may afford better protection against oxidative damage than previous approaches. Moreover, HIF not only activates transcriptional programs that lead to decreased ROS production (Semenza, 2011), but, in response to hypoxia, generation of ROS is necessary to increase HIF-1␣, making it a sensor of ROS (Chandel et al., 2000). The mechanism by which HIF is stabilized in response to ROS is not yet known, however, ROS are postulated to increase HIF-1␣ levels by interfering with HIF's hydroxylation by the HIF-prolyl hydroxylases at Pro 402 and Pro 564 and factor inhibiting HIF (FIH) at asparagine (Asn 803 ; Semenza, 2011). This implies that oxidative stress could augment HIF-1␣ stability and activity, further increasing its cellular functions. In the current study, the HIF construct used contains aa 1-529 of HIF. Thus, one of the hydroxylation sites, Pro 402 , is retained. It is therefore possible that the enhanced protective effects in response to oxidative stress arise from greater elevation in the levels of HIF in this condition.
HIF-1␣ increases the survival of SCs transplanted into the subacutely injured spinal cord (Fig. 4A-G). This result is similar to previous studies in which direct or indirect elevation of HIF-1␣ enhances transplant survival of cells transplanted into the damaged brain (Theus et al., 2008;Wu et al., 2010;Wakai et al., 2016), heart (Zhang et al., 2001), and pancreas (Stokes et al., 2013). Although expression of HIF-1␣ protected the cells, the increase in transplant survival with HIF-1␣ was smaller than in previous spinal cord transplant studies targeting trophic support, anoikis, and/or cell death signaling Hill et al., 2010;Patel et al., 2010;Lu et al., 2012;Robinson and Lu, 2017;Cerqueira et al., 2018). It is well established that prolonged overexpression of HIF in transplanted cells induces tumor formation (Kung et al., 2000). We chose to use low constitutive HIF expression to mitigate this potential problem. A greater increase in HIF-1␣ transcription may be required over the low-level HIF-1␣ induction of this study. Higher levels of HIF transcription are achievable with higher viral MOIs and pharmacological stabilization of HIF (unpublished data). Alternatively, activation of other transcription factors that mediate other key cellular programs (e.g., Nrf2 and NF␤) may provide greater survival benefits than HIF-1␣.
Elevation of HIF-1␣ can result in the doubling of transplant size (Wu et al., 2010;Wakai et al., 2016). This is substantially greater that the enhancement in survival achieved in the current study. Several factors could contribute to this difference in transplant survival including differences in the levels of HIF-1␣ in the cells post-TP. One limitation of this and previous studies is that although HIF-1␣ levels were measured in the cells before TP, they have not been measured post-TP. Cell type differences could also affect both the sensitivity of the cells to the transplant environment and the specifics of the transcriptional program(s) activated (Chi et al., 2006;Benita et al., 2009;Lendahl et al., 2009). Previous studies have examined stem cells or BMSCs, cell types that reside in lowoxygen niches. It is currently unclear how oxygen tension impacts adult de-differentiated SCs and which specific transcriptional programs are activated in SCs in response to elevation of HIF-1␣.
This study establishes that increasing HIF-1␣ is cytoprotective for transplanted SCs. It also demonstrates the feasibility of targeting HIF (and other transcription factors) as an approach to enhance transplanted cell survival. It supports the hypothesis that activating adaptive responses in transplanted cells can protect them following transplantation. This paves the way to test pharmacological interventions that temporarily elevate HIF (or other transcription factors) before transplantation to test whether overcoming the initial 2-d window in which the cells die is sufficient. Ongoing studies in the lab are testing the efficacy of clinically-feasible paradigms to elevate HIF-1␣ transiently and substantially.

Development of additional, rapid, reliable methods to quantify transplant survival is needed
Stereology proved to be the most sensitive method for detecting differences in transplant survival. In the injured spinal cord, IVIS imaging is able to detect a decrease in transplanted cells over time ( Fig. 1A; Okada et al., 2005;Takahashi et al., 2011;Roet et al., 2012;Nishimura et al., 2013). In cancer studies, it detects increases in tumor cells (Rehemtulla et al., 2000). Under the current conditions, neither in vivo bioluminescence imaging (Fig. 4J,K) nor ex vivo florescence imaging of spinal cords (Fig. 4H,I) was sensitive enough to detect the 35% improvement in transplant survival verified by stereology. Based on post hoc power analyses, a large number of animals would be required to detect a difference with either of these methods (Ն14). This decreases the utility of both in vivo bioluminescent imaging and ex vivo fluorescent imaging as primary screens for transplant survival in the injured rat spinal cord. In theory, bioluminescent and fluorescent imaging can detect as few as 1000 cells (Terrovitis et al., 2010); in practice, several factors limit light production/ detection. For spinal cord transplants, the location of the spinal cord deep within the vertebral column is particularly problematic due to tissue light absorption. Use of longer light wavelengths (i.e., far-red/near-infrared wavelengths) may circumvent this problem. Following SCI, several conditions within the transplant/injury site impact D-luciferin availability (e.g., altered vasculature) or enzymatic activity (e.g., reduced levels of required luciferin co-factors, ATP and O 2 ), which substantially impact light production and contribute to the variability in light detected between and within cases. This makes it difficult to ascertain whether different light levels detected arise from variability in the number of cells initially transplanted, delivery of D-luciferin to the cells, or differences in survival. Fluorescence-based probes overcome the limitations associated with D-luciferin administration. Far-red/near-infrared constructs for IVIS imaging have been developed (Shcherbo et al., 2007;Rumyantsev et al., 2016). They may facilitate in vivo imaging of cells following transplantation. One concern, however, is our failure to detect differences in transplant survival with ex vivo fluorescent imaging of GFP transplants where penetration of light through the tissue (i.e., bone, muscle, skin) is not required. This suggests additional technological advances are required for the use of fluorescence-based assays of transplant survival in whole tissues or animals. Current methods to quantify the number of surviving cells are either time-consuming (e.g., stereology), or require pulverization of the tissue (western blotting, qPCR) which prevents further examination, or either have too much variability or are not sensitive enough to detect anything but robust changes in survival (in vivo bioluminescent imaging and ex vivo fluorescent imaging). Additional rapid, sensitive approaches for screening transplant survival are needed to facilitate identification of prosurvival strategies. Until better assessments are developed, stereological quantification remains the most reliable method.

Summary and conclusions
To advance the field, and to maximize the therapeutic use and benefits of cellular transplants for human clinical use, there is a critical need to develop strategies that effectively promote and permit rapid assessment of transplant survival. We have identified the narrow time window in which transplanted cells die within the injured spinal cord, thus establishing the time window in which cytoprotection should be targeted to counteract transplanted cell death. We tested the effects of elevating HIF-1␣ in cells, and identify HIF-1␣ as a transcription factor that protects transplanted cells. Lastly, we tested three approaches to quantifying transplant survival, and demonstrate that stereology remains the most reliable until faster, more sensitive methods can be developed. We anticipate that interventions that specifically harness cellular adaptive responses before transplantation could obviate the need to add additional components to cells at the time of transplantation and thus aid in the adoption of this approach to current clinical transplant protocols.