Main

Bronchopulmonary dysplasia (BPD), a chronic lung disease that occurs in premature infants receiving prolonged ventilator support and oxygen supplementation, is associated with an increased risk of long-term neurodevelopmental impairments (1). Despite its limited predictive value, some clinical studies have also identified BPD as an independent risk factor for the development of neurofunctional deficits in premature infants, including cerebral palsy and developmental retardation, even in the absence of catastrophic brain injuries such as intraventricular hemorrhage and hypoxic-ischemic encephalopathy (1,2,3). No specific or effective treatment is currently available for the preterm infant brain. However, as indicated by evidence from Pham et al. (4), the close association between BPD and brain injury suggests that pulmo-protective therapies might also be neuroprotective in premature infants.

Recently, we reported that intratracheal xenotransplantation of human umbilical cord blood (UCB)-derived mesenchymal stem cells (MSCs) significantly attenuated hyperoxia-induced lung injuries in immune-competent newborn rats (5,6,7). Furthermore, in a phase I clinical trial, we showed that intratracheal transplantation of allogenic human UCB-derived MSCs in very preterm infants is safe and feasible (8). However, beyond its pulmo-protective properties, the potential neuroprotective effects of intratracheal MSCs transplantation have not yet been evaluated. In the present study, we used a newborn rat model of prolonged postnatal exposure to hyperoxia to develop neonatal hyperoxic lung and brain injuries, simulating clinical BPD and associated brain injury in very preterm infants. We tried to determine whether intratracheal MSC transplantation could simultaneously attenuate both lung and brain injuries, and whether the extent of such attenuation was correlated in these two tissues in hyperoxic newborn rats.

Results

Survival Rate, and Body and Brain Weight

Exposure to hyperoxia significantly reduced survival rate at postnatal day (P) 14 from 100% in the normoxia control (NC) group to 76% in the hyperoxia control (HC) group. The 87% survival rate observed in the hyperoxia with MSC transplantation (HM) group was not significantly different from the NC group.

Although birth weights did not differ significantly between the three study groups, body (24.69 ± 1.06 g) and brain weights (1.07 ± 0.02 g) of HC rats at P14 were significantly lower than those of NC rats (28.01 ± 0.42 and 1.16 ± 0.02 g, respectively). Body and brain weights at P14 (26.05 ± 0.78 and 1.10 ± 0.01 g, respectively) in the HM group were not significantly different from the NC group. However, the apparent increase in body and brain weight in the HM group compared to the HC group did not reach statistical significance ( Figure 1a , b ). The brain/body weight ratio (%) was not significantly different between NC, HC, and HM (NC: 4.12 ± 0.07, HC: 4.37 ± 0.20, HM: 4.34 ± 0.13) ( Figure 1c ).

Figure 1
figure 1

Body and brain weights. (a) Body weight, (b) brain weight, and (c) brain/ body weight ratio (%) at postnatal day (P) 14 of NC, HC, and HM rats exposed to 21% (normoxia) or 90% (hyperoxia) oxygen for 14 d. Data are given as mean ± SEM. *P < 0.05 vs. NC. HC, hyperoxia control; HM, hyperoxia with intratracheal transplantation of umbilical cord blood-derived mesenchymal stem cells; NC, normoxia control.

PowerPoint slide

Histological and Morphometric Analysis of P14 Rat Lungs

Representative light microscopy photomicrographs showing histopathological differences between the study groups are shown in Figure 2a . While small and uniform alveoli were observed in NC rat lungs, the HC group displayed fewer, larger, heterogeneously sized, and corrugated alveoli, indicative of impaired alveolarization. These hyperoxia-induced morphological changes and impairments in alveolar development were attenuated in the HM group. Morphometric analysis revealed that the mean linear intercept (MLI), an indication of alveolar size, was significantly higher in HC rats (72.5 ± 2.3 μm) compared to NC rats (52.7 ± 1.3 μm). This increased MLI was significantly attenuated in the HM group (63.5 ± 1.6 μm) ( Figure 2b ).

Figure 2
figure 2

Morphometric assessment of alveolarization. (a) Representative photomicrographs of lungs stained with hematoxylin and eosin (scale bar, 100 µm) at a magnification of 200 ×. (b) Morphometric measurements of mean linear intercepts (µm). White bar: normoxia control (NC); black bar: hyperoxia control (HC); gray bar: HM. Data are given as mean ± SEM. *P < 0.01 vs. NC. **P < 0.01 vs. HC.

PowerPoint slide

Inflammatory Cytokines, VEGF, and Superoxide Dismutase

Lung cytokines such as tumor necrosis factor (TNF)-α, interleukin (IL)-1α, IL-1β, and IL-6 were significantly increased and vascular endothelial growth factor (VEGF) was significantly decreased in the HC group compared to the NC group. These hyperoxia-induced effects were significantly diminished by MSC transplantation in HM rats ( Figure 3a , b ).

Figure 3
figure 3

Inflammatory cytokines in the lungs. Levels of (a) TNF-α, IL-1α, IL-1β, IL-6, and (b) vascular endothelial growth factor measured in lung tissue. White bars: normoxia control (NC); black bars: hyperoxia control (HC); gray bars: HM. Data are represented as mean ± SEM. *P < 0.01 vs. NC. **P < 0.05 vs. HC. †P < 0.01 vs. HC.

PowerPoint slide

In contrast, no significant differences in brain cytokine levels measured at P2, P6, and P14 (Supplementary Figure S1 online) and brain superoxide dismutase activity, measured at P14 (Supplementary Figure S2 online), were observed between the study groups.

TUNEL- and Ectodermal Dysplasia 1-Positive Brain Cells

Representative photomicrographs of terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL), ectodermal dysplasia (ED) 1 and ionized calcium binding adaptor molecule (Iba)-1-positive cells in the periventricular zone (PVZ), hippocampal dentate gyrus (DG), and cortex, are shown in Figure 4a c and Supplementary Figure S3a–c online, respectively. The number of TUNEL-positive cells in the brains of HC rats was significantly increased compared to the NC group. This hyperoxia-induced increase in TUNEL-positive cells in the DG was significantly mitigated by MSC transplantation, as seen in the HM group. In addition, the number of TUNEL-positive cells in the cortex did not significantly differ between the HM and NC groups ( Figure 4d f ). On the other hand, MSC transplantation did not cause any changes in the number of TUNEL-positive brain cells in the normoxic animals (Supplementary Figure S4a–c,e–g online).

Figure 4
figure 4

Dead cells and activated microglia in brain tissue. (a–c) Representative fluorescence micrographs of brain tissue stained with TUNEL (green), ED1 (red), and 4ʹ,6-diamidino-2-phenylindole (DAPI; blue) in the periventricular zone, hippocampal dentate gyrus, and cortex at P14. Images were taken at a magnification of 200× (scale bar, 100 µm). The number of (d–f) TUNEL-positive cells and (g–i) ED1-positive microglia per high-power field is shown in each brain region. White bars: normoxia control (NC); black bars: HC; gray bars: HM. Data are given as mean ± SEM. *P < 0.05 vs. NC. **P < 0.01 vs. NC. †P < 0.01 vs. HC.

PowerPoint slide

There were no significant differences in the number of ED1-positive cells ( Figure 4g i ), and Iba-1-positive cells (Supplementary Figure S3d–f online), indicating active and total microglial cells, respectively, between the study groups.

Brain Caspase-3 Levels and Oxidative Stress

Representative western blots for caspase-3 and carbonylated protein in brain homogenates at P2, P4, P6, P8, and P14, respectively are shown in Figure 5a . Caspase-3 levels ( Figure 5a , b ) in the brain were significantly increased in the HC group compared to the NC group at all time points, and these increases were significantly attenuated after MSC transplantation, as seen in the HM group ( Figure 5b ). Brain oxidative stress, assessed by carbonylated protein contents, was significantly increased at the early time points of P2 and P4, but not thereafter, at P6, P8, and P14 in the HC group compared to the NC group ( Figure 5c ).

Figure 5
figure 5

Western blots for Caspase-3 and carbonylated protein in brain homogenates at P2, P4, P6, P8, and 14. (a) Representative western blots of Caspase-3 (top), carbonylated protein (mid) and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (bottom; loading control). (b) Densitometric analysis of caspase-3 levels normalized to GAPDH. (c) Densitometric analysis of carbonylated protein levels normalized to GAPDH, given as percentage of normoxia control (NC) group at each corresponding time points. White bars: NC; black bars: HC; gray bars: HM. Data are mean ± SEM. *P < 0.05 vs. NC. **P < 0.01 vs. NC. †P < 0.05 vs. HC. (n = 6 per group).

PowerPoint slide

Brain Myelination

Brain myelination was assessed via immunofluorescent staining for myelin basic protein (MBP) at P14 ( Figure 6 ). MBP fluorescence intensity, indicative of the extent of brain myelination, was significantly reduced in the HC group compared to the NC group, but no difference was detected between the HM and NC groups. In addition, there was no difference in MBP fluorescence intensity between normoxic groups without MSC transplantation (NC) or with MSC transplantation (NM) (Supplementary Figure S4d,h online).

Figure 6
figure 6

Myelin basic protein (MBP)-positive nerve fibers and neural/glial antigen 2 (NG2)-positive cells in brain tissue. (a) Representative tile-scan confocal images of brain tissue stained for MBP (green) and with 4ʹ,6-diamidino-2-phenylindole (DAPI; blue) in white matter areas, including the lateral corpus callosum, cingulum, and external capsule, for each group at P14. (b) Magnified images of the rectangular boundary of the cingulate white matter area taken at a magnification of 100 × (scale bar, 200 µm). (c) Average myelin basic protein (MBP) fluorescence intensity in cingulate white matter. (d) Representative western blots of NG2 and MBP. (e) Densitometric analysis of NG2 levels normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH). (f) Densitometric analysis of MBP levels normalized to GAPDH. White bars: normoxia control (NC); black bars: HC; gray bars: HM. Data are given as mean ± SEM. *P < 0.05 vs. NC. **P < 0.01 vs. NC. †P < 0.05 vs. HC.

PowerPoint slide

In western blot analysis, level of proteoglycan nerve-glia antigen 2 (NG2), a marker for immature oligodendrocyte progenitor cells, and levels of MBP, a marker for mature oligodendrocyte were evaluated at P8 and P14, because brain myelination typically begins to develop at around P7 in neonatal rats (9). Levels of NG2 and MBP were significantly reduced at P8 and P14 in the HC group compared to the NC group. However, in the HM group, these reduced NG2 levels were restored at P8 and P14, and MBP levels were restored at P14 compared to the HC group ( Figure 6d f ).

Short-Term Neurofunctional Outcome

Developmental reflexes such as the righting reflex at P11–13 and negative geotaxis at P14 were done to assess short-term neurofunctional outcome (10,11). A significant delay in the maturation of righting reflex, as well as retardation of negative geotaxis, was observed in the HC group compared with the NC group. Although the HM group appeared to show some improvement on these measures, the effects did not reach statistical significance (Supplementary Figure S5 online).

Presence of Donor Cells

At P6 and P8, after intratracheal transplantation of MSCs, expression of human glyceraldehyde-3-phosphate dehydrogenase, indicative of the presence of human RNA derived from donor human MSCs, was detected only in the lung tissue, whereas not detected in the brain tissue of the HM group (Supplementary Figure S6 online).

Correlation Analysis of Lung and Brain Injuries

Lung MLI showed a significant inverse correlation with brain weight (R2 = 0.072, P = 0.036; Figure 7a ) and brain MBP fluorescence (R2 = 0.146, P = 0.034; Figure 7b ), a significant positive correlation with TUNEL-positive cell number in the PVZ (R2 = 0.397, P < 0.001; Figure 7c ) and DG (R2 = 0.362, P < 0.001; Figure 7d ), but not in the cortex (R2 = 0.095, P = 0.092; Figure 7e ), and no significant correlation with body weight (R2 = 0.041, P = 0.113; Figure 7f ). The inflammatory cytokines TNF-α, IL-1α, IL-1β, and IL-6 showed significant inverse correlations with brain weight (R2 = 0.197, P = 0.012; R2 = 0.359, P < 0.001; R2 = 0.281, P = 0.002; R2 = 0.165, P = 0.02, respectively; Figure 8a d ), and VEGF showed significant positive correlation with brain weight (R2 = 0.319, P = 0.001; Figure 8e ).

Figure 7
figure 7

Correlation analyses of lung alveolarization and brain characteristics. X-axes: (a–f) Mean linear intercept (μm). Y-axes: (a) brain weight (in grams) (R2 = 0.072, P = 0.036), (b) fluorescence intensity of myelin basic protein in cingulate white matter (R2 = 0.146, P = 0.034), (c–e) number of TUNEL-positive cells per high-power field in the periventricular zone, hippocampal dentate gyrus, and cortex, respectively (R2 = 0.397, P < 0.001; R2 = 0.362, P < 0.001; R2 = 0.095, P = 0.092, respectively), and (f) body weight (in grams) (R2 = 0.041, P < 0.113). White circles: normoxia control, black squares: HC, and gray triangles: HM.

PowerPoint slide

Figure 8
figure 8

Correlation analyses of lung characteristics and brain weight. Y-axes: (a–e) brain weight (in grams). X-axes: (a–d) lung inflammatory cytokine concentrations (pg/ml), including TNF-α, IL-1α, IL-1β, and IL-6, respectively (R2 = 0.183, P = 0.017; R2 = 0.289, P < 0.002; R2 = 0.241, P = 0.005; R2 = 0.179, P = 0.018, respectively), and (e) vascular endothelial growth factor concentration (pg/ml) (R2 = 0.319, P < 0.001). White circles: normoxia control, black squares: HC, and gray triangles: HM.

PowerPoint slide

Discussion

We previously observed that intratracheal transplantation of human UCB-derived MSCs improved hyperoxic lung injuries in newborn rats (5,6,7), and the clinical application of MSCs transplantation to the very premature infants is safe and feasible (8). In the present study, we asked whether intratracheal transplantation of MSCs could attenuate not only hyperoxic lung injury but also hyperoxic brain damage in newborn rats. In accordance with our previous studies (5,6,7), we observed that intratracheal transplantation of human UCB-derived MSCs significantly attenuated hyperoxia-induced lung injuries, including increased MLI, indicating impaired alveolarization, elevated levels of lung inflammatory cytokines, and reduced presence of VEGF. Simultaneously, MSC transplantation mitigated hyperoxic brain injuries, such as increased TUNEL positive cells in dentate gyrus as well as increased caspase-3 expression in whole brain and white matter injury evidenced by reduced MBP. To our knowledge, this is the first report of simultaneous neuroprotective effects of local intratracheal transplantation of MSCs for BPD in hyperoxic newborn rats. Taken together, these findings suggest that hyperoxic brain injury is closely connected to BPD, and that intratracheal transplantation of MSCs could go beyond mere attenuation of this condition to simultaneously improve associated brain injury and poor neurodevelopmental outcomes in very preterm infants.

The use of oxygen is critical for neonatal intensive care, and premature infants are especially vulnerable to oxygen toxicity due to the high possibility of exposure to supra-physiological oxygen concentrations during resuscitation and respiratory management, and a highly immature antioxidant defense mechanism. Besides the well-known toxic effects of hyperoxia on the developing lung, hyperoxia also leads to cell death in the developing white and grey matter of the brain, resulting in motor and cognitive deficits in premature infants (12,13,14,15). In addition to the lung injury known to occur after prolonged exposure to hyperoxia (90% oxygen for 14 d), our data revealed simultaneous hyperoxic brain injury, shown by a retardation in brain weight gain, transiently increased carbonylated protein, reduced MBP, and an increased TUNEL-positive cells and caspase-3 expression, without changes in inflammatory cytokine levels, the number of activated microglia, and superoxide dismutase activity. These observations indicate that the newborn rat model used in this study appropriately simulates the clinical conditions of BPD with associated brain injury in premature infants (4,16).

The mechanism by which local intratracheal transplantation of MSCs confers a neuroprotective effect, despite the absence of MSCs in the brain, remains unclear. In a multicenter clinical study (1,12), only very-low-birth-weight infants with BPD requiring prolonged oxygen supplementation were at higher risk of poor long-term neurocognitive and academic performance, compared to premature infants exhibiting less severe BPD. These findings suggest that only severe BPD constitutes an independent risk factor for associated brain injury and poor long-term neurodevelopmental outcome in very-low-birth-weight infants (1,17). Besides hyperoxia, infection and inflammation have also been implicated in preterm infant brain injury (18,19). In the present study, lung MLI, an indicator of alveolarization, showed a significant negative correlation with brain weight and MBP, and a significant positive correlation with the number of TUNEL-positive cells in the brain, measured at P14. Lung inflammatory cytokines and VEGF levels showed significant positive and negative correlations with brain weight, respectively, at P14. We have shown that the protective effects of MSC transplantation may primarily be mediated by a paracrine rather than a regenerative mechanism, and have highlighted the critical role of growth factors, including VEGF, in mediating the antioxidative, anti-inflammatory, and antiapoptotic paracrine effects of MSC transplantation (20). Taken together, these findings suggest that the neuroprotective effects of intratracheal MSC transplantation might be mediated by a multitude of factors, including the attenuation of hyperoxic lung injury itself, anti-inflammatory, antioxidative, and antiapoptotic paracrine effects, and growth factors, such as VEGF.

Hyperoxic injury in the developing brain is maturation-dependent, causing caspase-dependent apoptosis in immature and preoligodendrocytes, but not in mature oligodendrocytes (21). Furthermore, rats at P3 and P6, but not at P10, showed hypomyelination upon exposure to 80% oxygen (15). In the study by Vottier et al. (22), white matter myelin content and mature oligodendrocyte density at P10 were significantly reduced by postnatal hyperoxia for 7 d, but >2-fold increase in ED1- and TUNEL-positive cells at P3 tended to subside at P10, and significantly increased GFAP positive cells at P3 became normalized at P10. There were no significant differences in several neurotrophic factors, such as BDNF and VEGF, between the study groups. Overall, these findings suggest that hyperoxia induced brain injury occurs primarily during the most vulnerable developmental period, i.e., first postnatal week of life (14,15), comparable to human gestational age of 23–32 wk (23). In addition, the maturation of the antioxidant enzymes by P14 in rats could protect the newborn brain from oxidative stress and ensuing inflammation (24). In the present study, increased caspase-3 levels, which was persistently elevated between P2 and P14 in hyperoxic brains, was significantly attenuated at P8 and 14 by intratracheal MSC transplantation at P5. Caspase-3 is a key marker for apoptosis (25), but not specific because it is also involved in other metabolic processes (26). Therefore, performing simultaneous and serial evaluation of not only caspase-3 but also other apoptotic marker such as TUNEL would be more appropriate for the exact assessment of brain cell death. The lack of of additional serial assessment of other apoptotic cell markers is one of the limitations of the present study. Intratracheal MSC transplantation also ameliorated white matter injury in hyperoxic animals, preserving immature oligodendrocyte progenitor cells evident from increased NG2 level at P8 and 14, and, subsequently, preserving mature oligodendrocytes evident from increased MBP at P14.

Oxidative stress in the hyperoxic brain, marked by increased carbonylated protein, was only transiently increased at P2 and P4, before MSC transplantation, and returned to baseline levels thereafter, thus precluding demonstration of a direct antioxidative mechanism. Superoxide dismutase activity also did not differ between normoxic and hyperoxic brains. Given that there were no increases of inflammatory cytokines in hyperoxic brains, direct anti-inflammatory protection mechanisms could not be considered or evaluated. Moreover, MSCs transplanted were not detected in the brain. Therefore, we speculate that the neuroprotective mechanisms of intratracheally transplanted MSCs may be indirect as follows. First, intratracheally transplanted MSCs may indirectly affect the developing hyperoxic brain, via an improved systemic milieu associated with decreasing oxidative stress and inflammation in hyperoxic lungs. This is supported by evidence of a significant correlation of lung cytokine levels with brain weight as well as the number of brain TUNEL-positive cells, which may reflect that significant amelioration of hyperoxic brain injuries might associate with significant attenuation of systemic, and lung oxidative stress and ensuing inflammation by MSC transplantation. Second, paracrine factors, which are secreted from intratracheally transplanted MSCs, then entering systemic circulation, might play a direct protective role against increased cell death in the developing hyperoxic brain. The maturation of blood–brain barrier (BBB) is developmental and time dependent in neonatal rat (27). Furthermore, oxidative stress (28) and inflammation (29) have been shown to compromise the BBB function especially during the early developmental period. Wei et al. (30) have reported that the conditioned media of adipose stromal cells protect against hypoxia–ischemia-induced brain damage in P7 neonatal rats. They have showed that the factors within the conditioned media penetrated the BBB, which was disrupted by the inflammatory response after hypoxic–ischemic injury. Therefore, possible neuroprotective mechanism in our experiment could be through paracrine factors from intratracheally transplanted MSCs that enter into the systemic circulation and cross the BBB, which is relatively immature and probably disrupted by prolonged hyperoxia, in neonatal rat. However, to verify this, further studies are needed as the next step. Taken together, to clarify these neuroprotective mechanisms is necessary and future studies are needed. As some regions of the brain might be more vulnerable to hyperoxia than others, we measured TUNEL- and ED1-positive cells in brain through specific three regions, PVZ, DG, and parietal cortex. First, PVZ is known to be one of the most vulnerable regions to hyperoxic stress (31). Second, DG plays a critical role in learning and memory and neuroblasts reside, proliferate, and migrate (31) in its subgranular layer, and the cortex is neuron-rich gray matter, in which neurons process sensory information and generate fine motor skills (32). According to the previous reports (33), parietal cortex is also a region vulnerable to hyperoxia-induced cell death.

Besides attenuating abnormal histopathology, improving functional outcomes is crucial for the clinical application of MSCs in the premature infants. However, there is no consensus regarding the reliability of short-term neurodevelopmental effects of hyperoxic brain injury in the newborn rats up to P14. In the present study, hyperoxia (HC) retarded the maturation of righting reflex tested at P11-13 and slowed negative geotaxis tested at P14. Ostensible improvements in the HM group did not reach statistical significance. These findings suggest that short-term neurodevelopmental assessment is nonspecific, and may not be sensitive enough to detect the histopathologic improvements. Long-term neurodevelopmental studies, including the rotarod and water maze tests, will be necessary to assess the relevance of histopathology to behavior (34,35,36).

Optimizing MSC transplantation timing for hyperoxic lung and brain injuries is an important issue in clinical translation. Hyperoxic brain damage in rodents is age-dependent, with maximum vulnerability occurring during the first week of life (14,15). In our previous work (5), we showed that intratracheal transplantation of MSCs attenuated hyperoxic lung injury in newborn rats in a time-dependent manner, conferring significant protection in the early (P3), but not late (P10), phase of inflammation. These findings suggest that the therapeutic time window available for MSC transplantation may be narrow for BPD and hyperoxia-associated brain injury in preterm infants; further work will be necessary to define this critical window.

Successful clinical translation also requires understating of the optimal route of MSC transplantation. In the present study, MSCs were delivered intratracheally based on our previous preclinical studies showing >4-fold more potent therapeutic efficacy compared with systemic intraperitoneal (7) or intravenous transplantation (37). Intravenously transplanted MSCs do not cross the blood brain barrier, and are trafficked not only in the pulmonary capillaries but also in other organs, such as the liver, spleen, and kidney (38). Moreover, human umbilical mononuclear cells given intravenously did not attenuate excitotoxic white matter injury, and even aggravated white matter lesion following intraperitoneal administration (39). Overall, these findings indicate that local intratracheal route might be more efficient than the systemic intravenous route for MSC transplantation in an animal model of combining developing lung and brain injuries, simulating BPD and white matter damage in preterm infants.

In summary, by using a newborn rat model of prolonged exposure to hyperoxia in this study, we were able to induce both lung and brain injuries, simulating clinical BPD and associated brain injury in very premature infants. We also demonstrated that intratracheal transplantation of MSCs in hyperoxic infant rats significantly reduced not only lung injuries, but also associated brain damage and this attenuation was closely linked in the two tissues.

Methods

Cell Preparation

This study was approved by the Institutional Review Board of Samsung Medical Center and Medipost, Seoul, Korea. As described previously (5,6,7), MSCs were isolated from human UCB after neonatal delivery and informed consent was received from donors. UCB-derived MSCs were cultivated and confirmed for their characteristics using previously established protocols (5,6,7).

Animal Model

The Animal Care and Use Committee of the Samsung Biomedical Research Institute (Seoul, Korea) reviewed and approved the experimental protocol described herein. This study also followed the Institutional and National Institutes of Health guidelines for laboratory animal care. As described previously (5,6,7), timed pregnant Sprague-Dawley rats (Orient, Seoul, Korea) were housed in individual cages with free access to water and laboratory chow. Within 10 h after birth, the newborn rat pups were divided into three groups: normoxia control (NC; n = 14); hyperoxia control (HC; n = 17); and hyperoxia with intratracheal MSC transplantation (HM; n = 31). Normoxic rat pups were kept in room air, while hyperoxic rats were raised in hyperoxic chambers (90% oxygen) from birth until P14. To investigate the effect of MSC transplantation on normoxic control animals, a group of normoxic rats with intratracheal MSC transplantation (NM; n = 6) was added. In order to evaluate serial hyperoxic brain injury, time course experiments were done separately and animals were sacrificed as follows: the NC and HC groups at P2, P4, P6, P8, and P14, and HM group at P6, P8, and 14 respectively (n = 6 in each group). The details of hyperoxic exposure have been described previously (6,7).

Transplantation of Human UCB-Derived MSCs

For transplantation, fifth-passage human UCB-derived MSCs from a single donor were used. At P5, 5 × 105 MSCs in 0.05 ml normal saline (NS) were intratracheally transplanted into HM rats, while an equal volume of NS was intratracheally administered to NC and HC animals. No mortality occurred during the transplantation procedure. Details of this procedure, as well as transplantation of MSCs have been described previously (5,6,7).

Tissue Preparation

At P14, under deep pentobarbital anesthesia (60 mg/kg, i.p.), the lung and brain were harvested immediately following transcardiac perfusion with ice-cold NS. Rat brains were extracted wholly from olfactory bulbs to cerebellum, and the whole brain was weighed to an accuracy of two decimal places with an electronic scale guaranteed weight ranging from 0.20 to 400 g with 0.01 g readability (Scout Pro; OHAUS, Parsippany, NJ). Extracted lung tissue for morphometric analyses was inflated with NS and then immersion-fixed as previously (6,7). For histological analysis of both lung and brain in each rat (n = 6–16 per group), harvested tissues were fixed in 10% buffered formalin (40) before being embedded in paraffin wax the following day. Paraffin blocks of lung and brain tissues were then sliced into 4 µm sections. For biochemical observations (n = 8–15 per group), the lung and brain were snap frozen in liquid nitrogen and stored at −80 °C until use.

Lung Morphometry

The level of lung alveolarization was evaluated by MLI, representing mean interalveolar distance, using hematoxylin and eosin-stained lung sections, as described previously (5,6,7). MLI was measured as described by Cooney and Thurlbeck (41). A minimum of two sections per rat and six fields per section were examined randomly for each analysis.

Enzyme-linked immunosorbent assay

Following homogenization and centrifugation of frozen lung and brain tissues, the protein concentration in each supernatant was standardized across all samples. The levels of TNF-α, IL-1α, IL-1β, IL-6, and VEGF were measured using a MILLIPLEX MAP ELISA Kit according to the manufacturer’s protocol (EMD Millipore, Billerica, MA).

TUNEL Assay

Apoptosis was assayed via fluorescent TUNEL staining of brain sections (n= 6, 9, and 16 in NC, HC, and HM, respectively, and n = 6 in NM). Detailed materials and methods for counting TUNEL-positive cells are provided in the Supplementary Methods online.

Immunohistochemistry

To detect myelination, coronal brain sections were immunostained for MBP by incubation with an anti-MBP primary antibody (1:500; Abcam, Cambridge, UK) and an anti-mouse IgG secondary antibody (1:500; Invitrogen, Eugene, OR). To detect activated microglia, sections were immunostained for ED1 by incubation with an anti-monocyte/macrophage primary antibody (1:250; Millipore, Bedford, MA) and an anti-mouse IgG secondary antibody (1:500; Invitrogen). Total microglia, including activated and resting forms, were stained with Iba-1 antibodies. Brain sections were incubated with anti Iba-1 primary antibody (1:200; Abcam, Cambridge, UK) and anti-goat IgG secondary antibody (1:200; Invitrogen). Detailed materials and methods for observing MBP-positive nerve fibers, ED1-positive cells, and Iba-1-positive cells are provided in the Supplementary Methods online.

Western Blot

Carbonylated protein, caspase-3, Neural/glial antigen 2 (NG2), and MBP in whole brains were detected by western blot. Protein Carbonyl Western Blot kit (Cosmo Bio, Tokyo, Japan) was used according to the manufacturer protocol to detect carbonylated protein content, a biomarker of oxidative stress. To detect caspase-3, NG2 and MBP levels, the membranes were blocked and incubated with caspase-3 primary antibody (1:1,000; Cell Signaling Technology, Danvers, MA), NG2 primary antibody (1:1,000; Millipore, Bedford, MA), and MBP primary antibody (1:500; Abcam, Cambridge, UK). Housekeeping protein, glyceraldehyde-3-phosphate dehydrogenase (1:1,000; Santa Cruz Biotechnology, Santa Cruz, CA), level was measured as a loading control, followed by secondary antibody incubation with HRP-conjugated secondary antibody (1:1,000; DAKO, Glostrup, Denmark). Detailed materials and methods for electrophoresing protein, analyzing carbonylated protein and measuring levels of probed proteins are provided in the Supplementary Methods online.

Superoxide Dismutase Activity

Detailed materials and methods for measuring brain SOD activity are provided in the Supplementary Methods online.

Functional Behavioral Tests

On the P14, negative geotaxis test (35), which assesses the innate reflex rotation to face uphill when placed head-down on an inclined wooden platform was performed. Righting reflex (42) was assessed at P11–13. Detailed materials and methods for performing these functional behavioral tests are provided in the Supplementary Methods online.

Brain Donor Cell Detection

Presence of intratracheally transplanted MSCs was determined by performing RT-PCR for human DNA, which was derived from transplanted human MSCs; detailed material and methods are provided in the Supplementary Methods online.

Statistical Analysis

Data are expressed as mean ± SEM. Given that group sizes were small, statistical differences between groups were established using the Kruskal–Wallis test, followed by Nemenyi post hoc analysis. Correlations between lung and brain data were analyzed using Spearman’s or Pearson’s correlation coefficients, as appropriate. All data were analyzed using SAS 9.4 software (SAS Institute, Cary, NC) and P values less than 0.05 were considered statistically significant.

Statement of Financial Support

This work was supported by a Korea Research Foundation grant from the National Research Foundation of Korea (NRF) and the Ministry of Education, Science, and Technology (NRF-2014R1A1A2056427).

Disclosure

Samsung Medical Center and MEDIPOST Co., Ltd. own issued or filed patents for “Method of treating lung diseases using cells separated or proliferated from umbilical cord blood” in the names of the inventors, Y.S.C., W.S.P., and Y.S.Y.