In the U.S., approximately 1.7 million people suffer traumatic brain injury each year, with many enduring long-term consequences and significant medical and rehabilitation costs. The primary injury causes physical damage to neurons, glia, fiber tracts and microvasculature, which is then followed by secondary injury, consisting of pathophysiological mechanisms including an immune response, inflammation, edema, excitotoxicity, oxidative damage, and cell death. Most attempts at intervention focus on protection, repair or regeneration, with regenerative medicine becoming an intensively studied area over the past decade. The use of stem cells has been studied in many disease and injury models, using stem cells from a variety of sources and applications. In this study, human adipose-derived mesenchymal stromal cells (MSCs) were administered at early (3 days) and delayed (14 days) time points after controlled cortical impact (CCI) injury in rats. Animals were routinely assessed for neurological and vestibulomotor deficits, and at 32 days post-injury, brain tissue was processed by flow cytometry and immunohistochemistry to analyze neuroinflammation. Treatment with HB-adMSC at either 3d or 14d after injury resulted in significant improvements in neurocognitive outcome and a change in neuroinflammation one month after injury.
Funding: SDO and CSC have received research support from Hope Bio as part of a sponsored research agreement between Hope Bio and the University of Texas Health Science System (UTHealth). KAR performed the work in this study as an employee of UTHealth, but has since accepted paid employment at Hope Bio. Hope Bio also provided support for this study in the form of salaries for HP and AD. The specific roles of these authors are articulated in the ‘author contributions’ section. Additionally, this study received funding from the Grace Reynolds Wall Research Fund and the Clare A. Glassell Family Pediatric Stem Cell Research Fund. The funders had no role in study design, data collection and analysis, or decision to publish. Hope Biosciences received and approved a draft of the manuscript prior to submission and contributed some minor corrections and additional details for the methods, specifically the ‘Isolation and Expansion of HB-adMSCs’ section.
Competing interests: The authors have read the journal's policy and the authors of this manuscript have the following competing interests: KAR, HP, and AD are paid employees of Hope Bio and have received salary support for their role in this study. SDO and CSC have both received research support from sponsored research agreements between the University of Texas Health Science Center at Houston and Hope Bio. SDO is a Guest Editor on the “Stem Cell Plasticity in Tissue Repair and Regeneration” Call for Papers for PLoS ONE. Hope Bio produces and markets HB-AdMSCs and HB-AdMSC-related products. There are no other patents, products in development or marketed products associated with this research to declare. This does not alter our adherence to PLOS ONE policies on sharing data and materials.
Traumatic brain injury (TBI) is a major cause of disability in the United States, causing immediate and long-lasting effects . Providing care and rehabilitation for TBI patients is a significant economic burden, estimated to be approximately $76.5 billion for both direct and indirect costs . Cognitive impairments, motor and sensory dysfunction, and emotional changes are among the most common issues affecting TBI patients and their families and communities. Although the physical, emotional and economic burdens are large and can be long-lasting, there has yet to be an effective therapy to mitigate some of these issues.
Following the primary mechanical insult of TBI, secondary injuries have been shown to increase neurological damage and perpetuate the chronicity of TBI . Increased blood-brain barrier (BBB) permeability , neuroinflammation , edema, neurodegeneration, oxidative stress and innate immune response are hallmarks of secondary injury, with onset as early as minutes after impact and some effects lasting for several years, supporting the idea that TBI develops as a chronic illness . Moments after impact, an intense inflammatory response develops and BBB integrity is compromised, allowing peripheral leukocytes to infiltrate the brain parenchyma. Injured cells release damage-associated molecular pattern molecules (DAMPs) , which are recognized by infiltrating immune cells and resident microglia, astrocytes and neurons, stimulating the release of pro-inflammatory cytokines, anti-inflammatory cytokines, reactive oxygen and nitrogen species. The complex cascade of events following the primary injury is referred to as secondary injury. Modulation of this signaling process would provide a therapeutic target to interrupt or mitigate the exacerbation of secondary injury, ultimately sparing neighboring cells and lessening the likelihood of poor outcome.
Microglia activation following TBI is associated with neuroinflammation. Activated microglia are capable of exhibiting pro- and anti-inflammatory phenotypes, based upon polarization states. M1 phenotype microglia release pro-inflammatory cytokines and oxidative mediators, while M2 phenotype microglia release anti-inflammatory cytokines and neurotrophic factors . Activation and polarization of microglia is transient, varying at many points after injury. The classical paradigm of M1 and M2 phenotypes may not accurately characterize microglial activation, which has been suggested in more recent publications . Understanding temporal patterns in activation and polarization of microglia following TBI may provide information that allows more specific targets for modulation of neuroinflammation.
The interest in cell therapy as a potential therapeutic intervention for TBI has gained momentum over the past several years. Mesenchymal stem cells (MSC) are multipotent, fibroblast-like cells that can be isolated from various tissues, including bone, adipose, muscle, teeth, pancreas, lung, liver, amniotic fluid, cord blood and umbilical cord tissues [10–13]. MSCs are appealing for therapeutic intervention because they are easily isolated from virtually any adult tissues, have been shown to be safe and non-tumorigenic[14, 15] and exhibit potent immunomodulatory properties [16–18]. Administration of MSCs has been studied in many experimental models and clinical trials including, TBI [5, 19–25], stroke [26, 27], and spinal cord injury [28–35]. In regenerative medicine, some stem cell therapies aim to replace damaged cells, however the primary benefit of MSCs is derived from paracrine immunomodulation and alteration of the injury environment. Our group, as well as others, has demonstrated that the majority of MSCs do not cross the BBB to the injury site but rather, they perform immunomodulation from the periphery [36–40]. The release of various trophic factors contributes to attenuation of neuroinflammation, promoting angiogenesis, neurogenesis, and reducing apoptosis. Proinflammatory signals such as lipopolysaccharide (LPS), tumor necrosis factor-α (TNF-α), and nitric oxide (NO), stimulate MSCs to secrete anti-inflammatory factors . Of the anti-inflammatory factors, prostaglandin E2 (PGE2) has been shown to be constitutively expressed by MSCs via the COX-2 pathway and acts as a potent immunomodulatory factor . MSCs derived from bone marrow, umbilical cord and adipose tissue are most commonly studied, with some differences noted between tissues of origin . In this study, we explored the potential for early and delayed administration of adipose-derived MSCs (adMSCs) as a treatment for TBI.
Adipose-derived stem cells, also referred to as adipose-derived mesenchymal stem cells, are appealing for therapeutic use because they are relatively easy to harvest and they are available in abundance. AdMSCs are harvested from liposuction waste tissue and undergo collagenase digestion and differential centrifugation to produce a stromal vascular fraction containing adipocyte progenitor cells [43, 44]. These cells differentiate into adipocytes, osteoblasts, myoblasts, chondroblasts and neural cells, and maintain their identifying characteristics through several cell passages (manuscript in review). AdMSCs have been used in animal models for hemorrhagic stroke , spinal cord injury  and cerebral ischemia , and have demonstrated decreased inflammation, decreased neurodegeneration, improvement in motor function and decreased immune response.
In this study, we investigated the effects of early and delayed initiation of treatment following TBI. Examination of delayed initiation of treatment following TBI is novel and can provide a clinically relevant timeline for potential autologous cell therapy. We used a controlled cortical impact (CCI) model to simulate moderate to severe TBI in adult rats, which were treated with human adMSCs at either 3 days or 14 days post-injury. Animals were routinely assessed for neurological and vestibulomotor function and, upon sacrifice at 32 days post-injury, brain tissues were analyzed for neuroinflammation via flow cytometry and immunohistochemistry.
All protocols involving the use of animals were following the National Institutes of Health Guide for the Care and Use of Laboratory Animals. All work performed was approved by the University of Texas Health Sciences Center at Houston Animal Welfare Committee (HSC-AWC 16–0046) following the NIH Guide for the Care and Use of Laboratory Animals. The study utilized inhaled isoflurane for short-term anesthesia during surgical procedures followed by a lethal dose as a primary method of euthanasia followed by exsanguination and decapitation. Rats were purchased from Envigo (Indianapolis, IN, USA) for use in this study and housed on a 12h light/dark cycle with ad libitum access to food and water.
Moderate/severe cortical contusion injury (CCI) was performed in this study using an electromagnetically driven controlled cortical impact device (Pittsburgh Precision Instrument, Inc.), as previously described . Briefly, male Sprague Dawley rats (Envigo Harlan, USA) weighing 250-275g were anesthetized with 4% isoflurane in oxygen and maintained using 1.5 L/min of 2–3% isoflurane throughout the procedure. The head of the rat was then immobilized on a stereotactic frame (Pittsburgh Precision Instrument, Inc.) and a 6–7 mm diameter craniectomy was performed on the right cranial vault. The center of the craniectomy was placed at the midpoint between bregma and lambda, 3 mm lateral to the midline, overlying the temporoparietal cortex. Animals received a single impact of 3.1 mm depth of deformation with an impact velocity of 5.6 m/s and a dwell time of 150ms using a 6mm diameter flat impactor tip to the parietal association cortex orthogonal to the surface at a 20° angle from the vertical plane.
HB-adMSCs were derived from one single donor. Fat extraction (~ 10cc) was performed by a licensed plastic surgeon via mini liposuction from the abdominal adipose tissues. Fat was centrifuged at 3,000 rpm for 5 mins, which then separated into three layers: oil (top), tumescent (middle) and blood (bottom) layer. The middle layer was collected and mixed with collagenase type 1 solution (HB-102, Hope Biosciences, Sugar Land, TX, 1mg/ml), 1 ml of fat for every 4 ml of collagenase, which was incubated at 37°C with gentle shaking (150 ± 30 rpm) for 1–2 hours. Then the mixture was centrifuged at 1,800 rpm for 5 mins. There were two layers after centrifugation, the bottom layer was collected and re-suspended with 10 ml of HB-103 isolation media (Hope Biosciences, Sugar Land, TX). The solution was then filtered with a 100 μm cell strainer to remove debris and subsequently centrifuged at 1,700 rpm for 5 mins. The cell pellet, aka stromal vascular fraction (SVF), was collected and seeded in HB-103 at 37°C, 5% CO2 for 16–32 hrs. Cell adhesion was checked using the inverted microscope and the flask was washed with PBS/DPBS to remove non-adherent cells. The growth media, Hope Biosciences’ proprietary HB-101, was added to the flask to achieve the first expansion passage (P0) and incubated at 37°C, 5% CO2 until reaching 95% confluence. Media was changed every 2 days. Once cells reached 95% confluence, the P0 cells were sub-cultured and expanded in HB-101 until passage 4. P4 cells were harvested in PBS solution and transported to UT on the same day, which was also the injection day, at room temperature. HB-adMSC have characteristic traits consistent with conventional MSC, including a phenotype negative for CD31, CD34, CD45, and HLA-DR and positive for CD29, CD44, CD73, and CD90 (S1 Table).
We assessed cognitive function at 21 days post injury to assess spatial memory and spatial learning in rats as previously described . Standard protocol was followed where a blinded investigator tested all the groups in a hidden platform MWM set-up by performing 4 trials per day with 4 minutes between each trial for 6 days for each rat. The latency to platform was measured as the time to find the platform (Adanac 3000 Digital, Marathon Watch Company Ltd., Ontario, Canada). Probe Trials (removal of platform) were done on day 7, 24 hours after the conclusion of platform testing. The EthoVision® XT 8.0 tool from Noldus Information Technology was used to record and analyze the probe trials.
Microglial isolation and analysis was performed as previously published . In brief, microglia were isolated by first extracting the the brain and then mechanical and enzymatic digestion was used to obtain a single-cell suspension using a Neural Cell Dissociation kit (Miltenyi Biotec). We utilized density centrifugation to remove a large amount of myelin, followed by microglial enrichment using CD11b/c microbeads in magnet activated sorting (MACS) kit (Miltenyi Biotec). The resulting CD11-enriched cells were then stained for CD45, CD11, P2Y12, CD32, CD86, CD200R, RT1B, CD163, and a Ghost™ viability dye (TONBO) with the addition of Cyto-Cal™counting beads (Thermo Fisher Scientific). This optimized multicolor immunofluorescence panel (OMIP) was designed to phenotype rat-derived microglia from the CNS, and allows for differentiation between macrophages and microglia, as well as phenotypic changes in microglia. Time points were selected as comparable to our previous work in MSCs to treat TBI.
Rats were anesthetized and brain was harvested following perfusion with PBS and 4% PFA 32 days after the injury. Coronal sections at 30 μm were cut in a vibrating blade microtome (Leica VT1000 S) and stained in suspension using standard staining protocol. Briefly, the sections were first rinsed with PBS with 0.01% Triton X-100 (PBST; T-9284, Sigma Aldrich) twice for a minute each followed by permeabilizing step for 20 minutes with PBS with 0.2% Triton X-100. Then they were blocked for 30 minutes in PBST containing 3% goat serum (005-000-121, Jackson Immuno Research, PA). The primary and secondary antibodies were both prepared in PBST with 2% Bovine Serum Albumin (A2153, Sigma Aldrich) and 1% goat serum incubated for overnight at 4°C and 2 hours at room temperature, respectively. The sections were washed thrice with PBST for 10 minutes each prior to and after incubation with secondary antibodies. Lastly, the sections were mounted onto slides and let dry at room temperature before they were cover-slipped with DAPI Fluoromount-G (0100, Southern Biotech, Birmingham, AL). Primary antibodies: regenerating neurons (Doublecortin, 1:1000, AB2253, Temecula, CA), mature neurons (NeuN, 1:500, Millipore Sigma, Billerica, MA), astrocytes (GFAP, 1:500, Millipore Sigma, Billerica, MA), activated microglia (IBA-1, 1:500, Wako-chem, Richmond, VA). Secondary antibodies: 1:1000, Alexa Fluor 488/green: A11073, 1:1000, Alexa Fluor 568/red: A11011, Invitrogen.
Neurogenesis was determined using the same brain slices used for microglia/macrophage characterization, but assessing the number of doublecortin positive cells (DCX+) in the subgranular zone of the dentate gyrus . Cell quantification was performed using fluorescent microscopy with a Leica BDX5100 20x objective. Two slices per brain, at least 30 μm apart, were analyzed for the total number of DCX+ cells present in the subgranular zone of the dentate gyrus. The number of DCX+ cells were normalized by the total length of the subgranular zone of the dentate gyrus on each brain slice analyzed.
Each day, starting 2 days prior to injury, neurological outcomes were assessed using a neuroscore system . The short functional neuroscore consists of five tests: (1) forelimb flexion test, (2) hind limb flexion test, (3) visually triggered placing test, (4) contact triggered placing test, and (5) hind paw grasping reflex test. Forelimb flexion was tested by lifting the rat by the tail and holding approximately 12 inches above the table surface, observing for flexion or extension of forelimbs. Flexion is abnormal and receives a score of 1. No flexion has a score of 0. The hind limb flexion test is done the same as the flexion test, scoring 1 for hind limb flexion or 0 for no flexion. Visual triggered placing tests were performed by lifting the rat by the tail and slowly lowering toward the table edge, up to approximately 10 cm from nose to table edge. Moving the rat towards the edge, observation of the presence or absence of extending forepaws was scored. Extension is a normal behavior with a score of 0, no extension has a score of 1. Contact triggered placing tests were performed by holding the rat with body in hand, parallel to table edge, with forelegs free. Slowly the rat was lowered to the table until the whiskers on one side touch the edge of the table. Forelimb extension on the same side as the touching whiskers is normal, scoring 0, while the absence of extension in response to tactile stimulation is abnormal and given a score of 1. The test was repeated for opposite side. Hindpaw grasping reflex was tested by holding the rat in hand, thumb and index finger around the chest, under the forelimbs. Gently touching the palm of each hind paw with right forefinger and observing whether the rat grasps the finger. The presence of grasp is normal and receives a score of 0, no grasp has a score of 1. Each test was performed on both left and right sides. All scores were tallied for a possible total of 21. A score of 0 indicated normal reflex function.
The balance beam apparatus consists of a beam 22.5” L x 1.5” H x .75” W and a barrier 10.5” H x 13” W. The beam was secured to a table and the barrier was attached to the beam so that 10.5” of the beam protrudes from the barrier away from the table over a padded safety box. Animals underwent two training sessions and one pre-assessment prior to injury (day 0) beginning on day –2. On day –2, the animal was placed on the balance beam for 60 second trials. Then the animal was removed from the beam for a 15 second resting period between each trial in order to disorient him from the beam. If the animal could not balance, it was allowed to fall from the beam into a padded box. Animals were trained until able to remain on the beam for three consecutive 60-second trials. Trials were scored numerically, 1–6. Each trial was scored as follows: (1) Balances with steady posture (grooms, climbs barrier), (2) Grasps sides of beam and/or has shaky movements, (3) Hugs the beam or slips or spins on beam, (4) Attempts to balance, but falls off after ten seconds, (5) Drapes over beam or hangs from beam and falls off in less than ten seconds, and (6) Falls off, makes no attempt to balance or hang from beam. On the day of the injury (day 0), the animal underwent a pre-assessment consisting of three trials. The animals were assessed on various days post injury, 3, 14 and 28 days–three trials each day [53–56].
The beam walk apparatus consists of a beam that is measured 40” L x 1” W. One end of the beam is stabilized by a stand (starting end) and the opposite end is attached to the goal box that is on a table. The goal box is a black box with a hinged lid, for accessing the animal. The box is 11” L x 7.25” H x 7.25” W. The beam leads to a doorway (4.25” square) in the goal box. Four pegs (0.75” H) are inserted at 9.5”, 18.5”, 28.5”, and 38.25” from the starting end. Peg placement alternates along the outer edges of the beam beginning on the right edge. A light source and white noise source are positioned on a cart at the starting end of the beam walk. Training on the beam walk begins on day –2. The animal was placed in the goal box for two minutes at the start of the training session and the pegs are removed from the beam. At the end of two minutes, the handler turned on the white noise and light and removed the animal from the goal box via the hinged lid. The animal was placed on the beam at the location of the peg closest to the goal box and allowed to walk to the goal box. As soon as the animal’s front feet crossed the threshold of the goal box, the light and noise were turned off. The animal was allowed to rest in the goal box for 30 seconds between each run. The animal was not allowed to fall from the beam; the handler assisted him if he started to fall off. This procedure was repeated twice at each peg location and from the starting position. After two training runs from each position, the pegs were inserted and one complete beam walk was done for practice. Three timed beam walk trials were then recorded to conclude training. The timer started once the animal was securely positioned at the starting point and the first step was taken in the direction of the goal box. On the day of the injury (day 0), the animal underwent a pre-assessment consisting of three timed trials. The animals were assessed on various days post-injury days 1, 2 and 3 with three trials each day [53, 54, 56].
The ability of HB-adMSC to change immune activity was characterized using a set of previously published in vitro assays [18, 41, 48]. In brief, a dilution of HB-adMSC was co-cultured with either lipopolysaccharide (LPS) (Sigma) or concanavalin A (conA) (Sigma) and incubated for 24 or 48 hrs with the resulting supernatant analyzed for TNF-α or IFN-γ by ELISA (BD), respectively. Additionally, the LPS-stimulated co-culture’s media was analyzed for PGE2 by ELISA (Cayman). Separately, MSC were stimulated with either TNF-α (50 ng/ml) or IFN-γ (50 ng/ml) and quantitative reverse-transcriptase PCR was used to evaluate expression of COX-2 (PTGS2), IDO1, TSG-6, and IL-1ra using commercial Taqman probes (Applied Biosystems).
We selected two different treatment strategies for HB-adMSC infusion to approximate two common cell therapy approaches. HB-adMSC were infused 3d post-injury to target sub-acute secondary injury mechanisms similar to our previous studies [48, 57], and consistent with the application of a previously manufactured allogeneic cell product. Alternatively, a separate group was treated with HB-adMSC 14d post-injury to approximate the time required to isolate, expand, characterize, and deliver an autologous cell product in the absence of previously banked cells.
The latency to platform was measured each day, for 6 days, beginning day 21 post-CCI (Fig 2). Animals are tested in two consecutive trials each day; the first trial is to learn the location of the platform (location) and the second trial is to remember the location (match). The amount of time required for the animal to locate the platform on the second trial is indicative of ability to remember location based upon the spatial cues surrounding the maze. Animals treated with HB-adMSCs at 3 days post-CCI displayed significantly shorter latencies on test days 2, 3, 4, 5 and 6, when compared to injured controls. Delayed treatment resulted in significantly shorter latencies on test days 2 and 6. There was also significant difference between treatment groups on test days 3, 4 and 5. At the end of the test period, both treatment groups exhibited significantly shorter latencies (****, p