Brain, Behavior, and Immunity 70 (2018) 118–130

Contents lists available at ScienceDirect

Brain, Behavior, and Immunity

journal homepage: www.elsevier.com/locate/ybrbi

Full-length Article

Interaction between hypothermia and delayed mesenchymal stem cell

therapy in neonatal hypoxic-ischemic brain injury
Josephine Herz a,⇑, Christian Köster a,b, Barbara S. Reinboth a, Mark Dzietko a, Wiebke Hansen c,
Hemmen Sabir a, Cindy van Velthoven d, Ivo Bendix a,b,1, Ursula Felderhoff-Müser a,⇑,1
a
Department of Pediatrics 1/Neonatology University Hospital Essen, University of Duisburg-Essen, Essen, Germany
b
Experimental Perinatal Neuroscience, University Hospital Essen, University of Duisburg-Essen, Essen, Germany
c
Institute of Medical Microbiology, University Hospital Essen, University of Duisburg-Essen, Essen, Germany
d
Department of Neurology and Neurological Sciences, Stanford University, Stanford, USA

a r t i c l e i n f o a b s t r a c t

Article history: Acute hypothermia treatment (HT) is the only clinically established intervention following neonatal
Received 18 September 2017 hypoxic-ischemic brain injury. However, almost half of all cooled infants still die or suffer from long-
Received in revised form 2 February 2018 lasting neurological impairments. Regenerative therapies, such as mesenchymal stem cells (MSC) appear
Accepted 13 February 2018
promising as adjuvant therapy. In the present study, we hypothesized that HT combined with delayed
Available online 14 February 2018
MSC therapy results in augmented protection, improving long-term neurological outcome. Postnatal
day 9 (P9) C57BL/6 mice were exposed to hypoxia–ischemia followed by 4 h HT. Murine bone
Keywords:
marrow-derived MSC (1
106 cells/animal) were administered intranasally at P12. Cytokine and growth
Neonatal hypoxia–ischemia
Perinatal asphyxia
factor levels were assessed by ELISA and LuminexÒ multiplex assay 24 h following MSC delivery. One
Therapeutic hypothermia week after HI, tissue injury and neuroinflammatory responses were determined by immunohistochem-
Mesenchymal stem cells istry and western blot. Long-term motor-cognitive outcome was assessed 5 weeks post injury. MSC
Long-term functional outcome responses to the brains’ environment were evaluated by gene expression analysis in MSC, co-cultured
Neuroinflammation with brain homogenates isolated at P12. Both, MSC and HT improved motor deficits, while cognitive func-
tion could only be restored by MSC. Compared to each single therapy, combined treatment led to
increased long-lasting motor-cognitive deficits and exacerbated brain injury, accompanied by enhanced
endothelial activation and peripheral immune cell infiltration. MSC co-cultured with brain extracts of HT-
treated animals revealed increased pro-inflammatory cytokine and decreased growth factor expression.
In vivo protein analysis showed higher pro-inflammatory cytokine levels after combined treatment com-
pared to single therapy. Furthermore, HI-induced increase in growth factors was normalized to control
levels by HT and MSC single therapy, while the combination induced a further decline below control
levels. Our results suggest that alteration of the brains’ microenvironment by acute HT modulates MSC
function resulting in a pro-inflammatory environment combined with alteration of the homeostatic
growth factor milieu in the neonatal hypoxic-ischemic brain. This study delineates potential unexpected
side effects of cell-based therapies as add-on therapy for acute hypothermia treatment.
Ó 2018 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction ing in cerebral palsy, epilepsy, visual impairment and motor-

Perinatal asphyxia, resulting in hypoxic-ischemic encephalopa-
thy (HIE), is one of the worldwide leading causes of death and
disability in children. In high-income countries 1 to 6 per 1000
newborns suffer from HIE during the perinatal period often result-
⇑ Corresponding authors at: Department of Pediatrics 1, University Hospital
Essen, University of Duisburg-Essen, Hufelandstr. 55, 45147 Essen, Essen, Germany.
E-mail addresses: josephine.herz@uk-essen.de (J. Herz), ursula.felderhoff@uk-
essen.de (U. Felderhoff-Müser).
1
Equally contributing senior authors.
cognitive deficits in later life (Eunson, 2015).
To date, therapeutic hypothermia (HT) is the only recom-
mended therapeutic intervention following perinatal asphyxia
(ILCOR guidelines 2015). Neuroprotective mechanisms involved
are inhibition of acute metabolic disturbance, inhibition of apop-
totic cell death and suppression of inflammation (Wassink et al.,
2014). Nevertheless, 40–50% of cooled infants still suffer from
major neurological problems and the therapeutic window of
opportunity (6 h) is challenging (Azzopardi et al., 2014). Experi-
mentally, we and others demonstrated short-term neuroprotection
by HT following HIE-induced brain injury, whereas long-term

https://doi.org/10.1016/j.bbi.2018.02.006
0889-1591/Ó 2018 The Authors. Published by Elsevier Inc.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
 J. Herz et al. / Brain, Behavior, and Immunity 70 (2018) 118–130
neurodevelopment is only partially restored, closely resembling
the clinical situation (Azzopardi et al., 2014; Burnsed et al., 2015;
Pappas et al., 2015; Reinboth et al., 2016).
Considering limited long-term neurodevelopmental improve-
ment following HT, combined treatment with regenerative cell-
based therapies attenuating sub-acute injury processes and pro-
moting repair appears promising. To increase endogenous repair,
bone marrow-derived mesenchymal stem cells (MSC) have gained
major interest in the past. Pioneer work in a murine model of
neonatal HIE by van Velthoven et al. revealed that MSC provide
long-term neuroprotection through promotion of endogenous cell
proliferation, inhibition of microglia activation and preservation
of white and grey matter (van Velthoven et al., 2010a). Recent evi-
dence also suggests that intranasal MSC administration is an effi-
cient and less invasive route to treat neonatal brain injury (van
Velthoven et al., 2010a). Furthermore, in rodents, the therapeutic
time window of intranasal MSC therapy has been reported to be
expandable for up to 10 days post HIE (Donega et al., 2013).
Delayed treatment options will provide time for logistics (e.g.
transferring patients from peripheral hospitals for treatment, con-
firmed diagnosis of HIE by clinical course, neurophysiology and
imaging). Furthermore, because of the short therapeutic window,
HT is supposed to target early pathophysiological processes. In
contrast, MSC have been shown to act on secondary injury pro-
cesses and promote endogenous repair which is initiated in the
post-acute phase (Ferriero, 2004; Gunn et al., 2017; Thornton
et al., 2012). Therefore, acute HT with delayed MSC therapy would
be an attractive therapy design.
Even though small clinical trials (e.g. NCT02612155,
NCT02881979), investigating the effect of stem cell therapy in
cooled asphyxiated newborns, are already underway, pre-clinical
studies have rarely addressed the benefit of the combined treat-
ment. Park et al. delivered human umbilical cord blood MSC intra-
ventricularly prior to HT, resulting in long-term protection from
HIE-induced brain tissue loss (Park et al., 2015). However, an acute
intracerebral cell administration is unlikely to be feasible in the
clinical setup following perinatal asphyxia. Experimental work
investigating HT in combination with a less invasive and delayed
stem cell therapy was lacking.
The present study investigated the hypothesis that immediate
HT combined with delayed intranasal MSC therapy promotes
endogenous repair mechanisms and inhibits secondary injury pro-
cesses overcoming current limitations of HT for the treatment of
neonatal HIE.
2. Materials and methods
2.1. Isolation and culture of murine bone marrow mesenchymal stem
cells
Bone marrow MSC were cultured according to previously pub-
lished protocols (van Velthoven et al., 2010a). Briefly, bone marrow
from femurs and tibias of 5–7 week old C57BL/6J mice (Harlan,
Germany) was flushed with Dulbeccos modified eagle medium
low glucose GlutaMAXTm (DMEM, Gibco, Germany) /20% fetal calf
serum (FCS, Biochrom, Germany). Cells were sub-cultured four to
six times prior to administration. For in vivo tracking analysis,
GFP-labelled MSC (Cyagen Biosciences Inc, USA) were thawed
and sub-cultured twice before intranasal delivery.
2.2. Phenotypic characterization of cultured murine bone marrow-
derived mesenchymal stem cells
Cultured MSC were stained with fluorescein isothiocyanate,
phycoerythrine, allophycocyanin, Alexa-Flour 780 or eF450-
119
conjugated anti-CD45 (BD Biosciences, Germany, clone: 30-F11),
anti-TER119 (eBiosciences, Germany, clone: TER-119), anti-
CD11b (BD Biosciences, clone: M1/70), anti-MHC II (eBiosciences,
clone: M5/114.15.2), anti-Sca-1 (eBiosciences, clone: D7), anti-
CD29 (eBiosciences, clone: eBioHMb-1(HMB1-1)), anti- CD44
(eBiosciences, clone: IM7) and anti-CD49e (eBiosciences, clone:
eBioHMa5-1 (HMa5-1)) antibodies for 20 min at 4 °C. Isotype-
matched control monoclonal antibodies were used to set appropri-
ate gates for positive staining considering the level of background
staining (Supplementary Fig. S1A). Flow cytometric analyses were
performed on a BD LSRII equipped with FACS Diva Software (both
BD Biosciences) which was used for acquisition and data analysis.
Osteogenic and adipogenic differentiation potential (Supple-
mentary Fig. S1B) of cultured MSC was evaluated by seeding 5 x
104 MSC per well in a 24-well culture plate (duplicates) with
MSC growth medium (DMEM low glucose GlutaMAXTm,
Gibco)/20% fetal calf serum (FCS, Biochrom). After 48 – 72 h
(100% confluence) one of each duplicate samples was exposed to
MSC osteogenic or adipogenic differentiation medium (PromoCell,
Germany). MSC growth medium was used for the remaining wells
as negative control. Cells were incubated for 18–21 days with
media exchange every third day. Osteocyte differentiation was
analyzed via Alizarin Red (2%) staining for 45 min. Differentiation
into adipocytes was confirmed by Oil-Red (0.3%) staining for 15
min (Supplementary Fig. S1B).
Chondrogenic differentiation potential was determined by
seeding 4x105 cells in 15 ml Falcon tubes in duplicate to enable
formation of spheroids. After 24–48 h one sample of duplicates
was exposed to chondrogenic differentiation medium (PromoCell)
while MSC growth medium was used as negative control in the
remaining tube. Medium was exchanged every 3 days. After incu-
bation for 18–21 days cell pellets were carefully removed and fro-
zen on dry ice. Cell pellets were cut into 10 mm cryostat sections
and differentiation into chondrocytes was evaluated by staining
with Alcian Blue (0.25%) for 5 h (Supplementary Fig. S1B).
2.3. Animal care and group allocation
Experiments were performed in accordance to the Animal
Research: Reporting of In Vivo Experiments (ARRIVE) guidelines
with government approval by the State Agency for Nature, Envi-
ronment and Consumer Protection North Rhine-Westphalia.
C57BL/6J mice were bred in house and kept under a 12 h light/dark
cycle with food and water ad libitum. Animals subjected to behav-
ioral testing were weaned at postnatal day 21 (P21) and housed in
groups of 4–5 animals per cage. Bodyweight was recorded at P10,
P11, P12, P16 and once a week after weaning.
A total of 305 P9 pups (n = 151 female and n = 154 male)
derived from 41 litters were enrolled. Forty-seven mice underwent
sham surgery. Out of the 258 animals exposed to hypoxia-ischemia
18 animals (7%) died immediately after or during hypoxia. Temper-
ature during intervention was monitored in sentinel pups (n = 42
in total, Supplementary Fig. S2A), being excluded from data analy-
sis to minimize confounding effects by stressful temperature mea-
surements. Remaining animals were randomly assigned to 4
treatment groups by an independent scientist not involved in data
acquisition: (1) hypoxia ischemia (HI)/normothermia (NT, n = 18
(5 weeks survival), n = 16 (1 week survival), n = 8 (4 days survival),
n = 8 (3 days survival)), (2) HI/hypothermia (HT, n = 20 (5 weeks
survival), n = 14 (1 week survival), n = 8 (4 days survival), n = 8
(3 days survival)), (3) HI/NT/MSC (MSC, n = 17 (5 weeks survival),
n = 16 (1 week survival), n = 16 (4 days survival)) (4) HI/HT/MSC
(HT + MSC, n = 18 (5 weeks survival), n = 15 (1 week survival),
n = 16 (4 days survival)). In total 1 NT and 1 HT + MSC died within
the first week after HI; 2 HT and 2 HT + MSC animals died between
120 J. Herz et al. / Brain, Behavior, and Immunity 70 (2018) 118–130
1 and 5 weeks post injury induction. Experimenters were blinded
to all interventions and data analysis.
2.4. Neonatal hypoxia-ischemia
Hypoxic-ischemic (HI) brain injury was induced as previously
described (Reinboth et al., 2016). Briefly, the right common carotid
artery was occluded through cauterization (high temperature cau-
ter, 1200 °C, Bovie, USA) under isoflurane anesthesia (1.5–4 vol%,
total duration of surgery: 5–7 min) followed by one hour hypoxia
(10% O2) in an air-tight oxygen chamber (OxyCycler, Biospherix,
USA) after one hour recovery with their dams. According to our
previous study where we identified a physiological body core (i.e.
nesting) temperature of 35 °C for P9 mice (Reinboth et al., 2016)
animals were placed on a warming mat (Harvard Apparatus,
USA) to maintain nesting temperature during hypoxia. Sham-
operated animals received anesthesia and neck incision only.
2.5. Therapeutic hypothermia and MSC-treatment
According to our previous study, HT was applied immediately
following HI for 4 h with a target temperature of Trectal 32 °C
(Reinboth et al., 2016). Mice were placed on a custom-made plate
with temperature control by water circulation. Controls (NT) were
placed on a warming mat to maintain physiological body core tem-
perature (Trectal 35 °C) conforming to our previous work (Reinboth
et al., 2016). Three days after HI (Fig. 1A), a total of 1x106 MSC or
vehicle (0.9% sodium chloride) were administered intranasally as
previously described (Donega et al., 2013; van Velthoven et al.,
2010a). This treatment regime aimed to extend the therapeutic
time window and was selected based on a previous study reporting
pronounced neuroprotection after delivery of 1x106 cells at 3 days
post HI (Donega et al., 2013).
2.6. Long-term functional outcome
Behavioral assessment was performed according to our previ-
ously described protocols (Reinboth et al., 2016). Shortly, explo-
ration and impulsive behavior was assessed in the elevated plus
maze (Lister, 1987; Walf and Frye, 2007). Mice were placed on
the central platform and behavior was recorded for 5 min. The
time spent in the open arms and the time of head-dipping in open
and center regions was measured. The open field test was used to
evaluate spontaneous motor- and rearing activity and to assess
anxiety/exploration-related behavior (Milner and Crabbe, 2008).
Animals were placed into the center of an open field arena and
movements were recorded by the tracking system for 5 min.
Besides measurement of mean velocities, anxiety/exploration-
related behavior was determined as the percentage of time the
animal stayed in the central area of the box in relation to the total
time spent in the arena. Rearing activity was calculated as the
time the animals reared in the border region relative to the total
time the mice spent in this region. The novel object recognition
task is a non-spatial, non-aversive memory test which relies on
the observation, that animals preferentially explore novel objects
over those that are familiar (Bevins and Besheer, 2006). In the ini-
tial familiarization trial, animals were placed in the open field
arena, with 2 identical objects in two facing corners. The animals
were returned to their cages for an inter-trial interval of 30 min.
In the following retention/test trial, animals were exposed to
one familiar object and one novel object replacing the second
familiar object in the arena. Novel object activity was recorded
for 5 min.
From P21 onwards, animals were transferred to an inversed
12 h light/dark cycle and behavioral testing was initiated at P44
(Reinboth et al., 2016). It started with one day of elevated plus
maze followed by one day open field and one day novel object
recognition test. Data were recorded using an automatic tracking
system (Video-Mot2, TSE Systems, Germany) and exported for sta-
tistical analysis. All experimental procedures were carried out in
the dark phase in a dimly lit (red light) and a low noise environ-
ment (behavioral unit).
2.7. Tissue preparation, histology and immunohistochemistry
One week after HI, mice were deeply anesthetized with chloral-
hydrate (200 mg/kg body weight) and transcardially perfused with
ice-cold phosphate buffered saline (PBS). Brains were removed and
snap frozen on dry ice. Tissue injury was assessed and scored on
cresyl violet stained 20 mm cryostat sections as previously
described (Reinboth et al., 2016; Sheldon et al., 1998) with minor
modifications. Briefly, 9 regions were scored: the anterior, middle
and posterior cortex, CA1, CA2, CA3 and dentate gyrus of the hip-
pocampus, the striatum and the thalamus. Each region was given
a rating from 0 to 3 (0- no detectable cell loss, 1- small focal areas
of neuronal cell loss, 2- columnar damage in the cortex or moder-
ate to severe cell loss in the other regions, 3- cystic infarction and
gliosis). The sum score from different regions was calculated for
each animal resulting in a total maximum score of 27.
Cellular injury and neuroinflammatory responses were assessed
on cryostat sections from the striatal (+0.2 to +0.3 mm from
bregma) and the hippocampal (1.9 to 2.0 mm from bregma)
level via immunohistochemical detection of neurons (NeuN), oligo-
dendrocytes (Olig2), leukocytes (CD45) and vessels (CD31) accord-
ing to previously published protocols (Herz et al., 2015; Reinboth
et al., 2016). Briefly, brain tissue sections (20 mm) were thawed
and dried at 37 °C followed by fixation in ice-cold 4%
paraformaldehyde (for NeuN, Olig2) or aceton/methanol (1:1 v/v
%, for CD45, CD31) for 10 min at 4 °C. Unspecific antibody binding
was blocked by incubation with 1% bovine serum albumin (BSA,
Sigma-Aldrich), 0.3% cold fish skin gelatin (Sigma Aldrich), 0.05%
Tween 20 (Sigma Aldrich) in Tris-buffered saline (TBS, for NeuN,
Olig2) or 5% donkey, 5% normal goat serum, 2% BSA, 0.2% Tween
20 in PBS (for CD45, CD31) for 1 h at room temperature. Sections
were incubated with the following primary antibodies: rabbit
anti-mouse Olig2, (1:100, Millipore, Germany); rabbit anti-mouse
NeuN, (1:500, Millipore), rat anti-mouse CD45 (1:20, BD Pharmin-
gen, Germany), rat-anti-mouse CD31 (1:500, BD Pharmingen) in
blocking solution at 4 °C overnight. Antibody binding was
visualized by incubation with appropriate secondary antibodies
(anti-rabbit or anti-rat Alexa Fluor 488, both: 1:500, Invitrogen,
Germany) for 2 h at room temperature. Nuclei were counterstained
with 40 ,6-Diamidin-2-phenylindol (Dapi, 100 ng/ml; Molecular
Probes, USA).
2.8. Quantification of neuronal, oligodendrocyte and vessel density;
vessel diameters and peripheral leukocyte infiltration
Defined non-overlapping regions of interest (ROI, each 1,89,000
mm2 (CD45, CD31) and 45,900 mm2 (NeuN, Olig2)) were visualized
by fluorescence microscopy (20
and 40
objective; Axioplan;
Zeiss, Germany) connected to a CCD camera (Axiocam ICc1, Zeiss).
For representative images, low and high magnification images
were acquired with confocal microscopy (A1plus, Eclipse Ti, with
NIS Elements AR software, Nikon, Germany). At the level of the
striatum 6 defined non-overlapping ROI were analyzed in each
hemisphere (3 ROI in the cortex and 3 ROI in the striatum). At
the level of the hippocampus 10 ROI (3 ROI in the cortex, 4 ROI
in the hippocampus and 3 ROI in the thalamus) were evaluated
in each hemisphere. Oligodendrocytes and vessels were counted
manually on acquired images. Since single cell counting could
not be applied in densely packed regions of neurons or strong
 J. Herz et al. / Brain, Behavior, and Immunity 70 (2018) 118–130 121

Fig. 1. Improved long-term functional outcome by acute hypothermia and delayed intranasal MSC treatment is reduced after combination of both therapies. (A) Experimental
paradigm and readouts. (B–D) Behavioral testing was initiated on P44 in young adult animals that were either sham-operated or exposed to HI followed by 4 h NT or HT on
P9. MSC were administered intranasally at P12. (B) In the Elevated plus maze the time spent in open arms and time of head-dipping in open and center regions was measured.
(C) Rearing activity was quantified in the Open field maze by calculation of the time animals reared in the border region relative to total time mice spent in this region.
(D) Cognitive function was assessed by Novel object recognition task where animals were exposed to one familiar object and one novel object after a training trial with
two identical objects. Percentage of time mice spent with the new object of total object time was analyzed. n = 16–18/group, *p < 0.05, **p < 0.01, ***p < 0.001,
NT = normothermia, HT = hypothermia, MSC = mesenchymal stem cells.

leukocyte infiltration (e.g. hippocampus), images were converted were measured with Image J for 12–15 randomly selected vessels
into binary images followed by measurement of NeuN and CD45 per animal derived from three non-overlapping hippocampal ROI
positive areas using Image J (NIH, USA). Ipsilateral vessel diameters (CA1, CA2, DG covering a total area of 0.567 mm2).
122 J. Herz et al. / Brain, Behavior, and Immunity 70 (2018) 118–130
2.9. Western blot
For western blot analysis sections of 200 mm thickness of the
ipsilateral hemisphere within the range of 0.3 to 0 mm from
bregma (striatal level) and 2.0 to 2.3 mm from bregma (hip-
pocampal level) were dissected and homogenized in ice-cold lysis
buffer (RIPA, Sigma-Aldrich) containing protease and phosphatase
inhibitors (cOmplete, Roche) and 100 mM PMSF (Sigma-Aldrich).
The supernatant was collected and processed as previously
described (Reinboth et al., 2016). Briefly, after determination of
protein concentration using the Pierce bicinchoninic acid assay
(Thermo Scientific, USA), protein lysates were separated on gradi-
ent sodiumdodecylsulphate polyacrylamide gels (Mini-PROTEAN
TGX Precast Gels, Any kDa, BioRad, Germany) and transferred to
nitrocellulose membranes (0.2 mm, Amersham, USA) at 4 °C over-
night. Equal loading of 20 mg/lane and transfer of proteins was con-
firmed by staining of membranes with Ponceau S solution (Sigma-
Aldrich). Nonspecific binding was blocked by incubation in 5% non-
fat milk powder, 0.1% Tween in TBS (TBST) followed by incubation
with the primary antibodies, i.e. rabbit anti-myelin basic protein
(MBP, 1:10.000, Covance, USA), mouse anti-microtubuli associated
protein-2 (MAP-2, 1:1000, Sigma-Aldrich), rabbit anti-ionized cal-
cium binding adaptor molecule-1 (Iba-1, 1:1000, Wako, Japan),
goat anti-vascular cell adhesion molecule-1 (VCAM-1, 1:10.000,
R&D Systems, USA) and rabbit anti-glutaraldehyde-3-phosphate
dehydrogenase (GAPDH, 1:1000, Santa Cruz, USA) in blocking solu-
tion at 4 °C overnight. Membranes were incubated with appropri-
ate peroxidase-conjugated secondary antibodies (all 1:2000 except
of anti-goat horseradish peroxidase (1:10.000), all Dako, Denmark)
in blocking solution at room temperature for 1 h followed by
Chemiluminescent detection with the enhanced chemilumines-
cence prime western blotting detection reagent (Amersham, GE
Healthcare Life Science, USA). For visualization and densitometric
analysis the ChemiDocXRS+ imaging system and ImageLab soft-
ware (Bio-Rad, Germany) were used.
2.10. MSC tracking analysis
According to Donega et al. MSC homing was analyzed at 14–16
h following intranasal MSC delivery 3 days after HI (Donega et al.,
2014). The amount of GFP-labelled MSC in brains of NT and HT-
treated animals was quantified by flow cytometry. Isolation of sin-
gle cell suspension for flow cytometry analysis was performed as
previously described (Herz et al., 2014). Hemispheres were dis-
sected, cerebellum removed followed by homogenization through
a 70 lm cell strainer (BD Biosciences, Germany) during continuous
rinsing with 15 ml of cold HEPES-buffered RPMI1640. Two hemi-
spheres were pooled per sample to avoid potential cell loss
through tissue processing. Samples were centrifuged at 400
g
for 10 min at 18 °C. The supernatants were discarded and the pel-
lets were re-suspended in 37% Percoll followed by centrifugation at
2800
g for 20 min. Myelin was removed and the remaining cell
pellet was washed twice in PBS. Total cell counts were determined
using BD TrueCount beads (BD Biosciences). Viable cells were
quantified by adding 1 mg/ml propidium iodide immediately prior
to measurement. Data acquisition and analysis were performed
on a BD FACS LSRII equipped with FACS Diva software (BD
Biosciences).
2.11. MSC co-culture with brain extracts and gene expression analysis
MSC were cultured with brain extracts as previously described
(van Velthoven et al., 2010b). Tissues of ipsilateral hemispheres
(+0.5 to 2.3 mm from bregma) of mice from all treatment groups
were dissected on ice followed by homogenization in knock-out
DMEM (Gibco). Supernatants were stored at -80 °C until use. One
million MSC/well were pre-cultured in 6-well plates for 24 h in
DMEM/20% FCS followed by culture in DMEM knock out medium
supplemented with isolated brain extract (1 mg protein/ml med-
ium). After 48 h total RNA was isolated with the RNeasy Micro
Kit (Qiagen Germany) according to the manufactures recommen-
dations. First strand complementary DNA was synthesized using
1 mg of total RNA and TaqMan reverse transcription reagents
(Applied Biosystems/Thermo Fisher Scientific USA). PCR amplifica-
tion was performed in 96 well optical reaction plates for 40 cycles
with each cycle at 94 °C for 15 s and 60 °C for 1 min using the Ste-
pOnePlus Real Time PCR system (Applied Biosystems/Thermo
Fisher Scientific). PCR products of tumor necrosis factor alpha
(tnf-alpha, Mm00443258_m1), interleukin (IL)-1beta (il-1beta,
Mm00434228_m1)), il-6 (Mm00446190_m1); il-12
(Mm00434169_m1), il-10 (Mm01288386_m1), brain derived neu-
rotrophic factor (bdnf, Mm01334043_m1), epidermal growth factor
(egf, Mm00438696_m1), insulin like growth factor (igf,
Mm00439560_m1), vascular endothelial growth factor (vegf,
Mm00437306_m1) and beta-2-micorglobulin (Mm00437762_m1)
were quantified using assay on demand primers and fluorogenic
reporter oligonucleotide probes (Applied Biosystems/Thermo
Fisher Scientific). CT values were normalized for the housekeeping
gene beta-2-microglobulin [DCT = CT (target gene)-CT (beta-2-
microglobulin)] and related to the mean of control samples (MSC
co-cultured with sham brain extracts) using the DDCT formula
[DDCT = DCT (sample) -DCT (control)]. Fold change values were
calculated.
2.12. Assessment of cytokines and growth factors
Abundance of cytokines and growth factors was assessed in
brain lysates, plasma samples and supernatants of MSC co-
cultures. For analysis of brain tissues, mice were perfused, hemi-
spheres dissected (excluding cerebellum), snap frozen on dry ice
and stored at 80 °C until further processing. Tissues were homog-
enized in ice-cold Cell Lysis Buffer 2 (R&D Systems, Germany).
Blood was collected from the right atrium prior to perfusion and
collected in ethylenediaminetetraacetate (EDTA) coated tubes.
Samples were centrifuged at 2000
g for 20 min and the super-
natant was frozen at 80 °C until further analysis. Cell culture
supernatants, brain lysates and plasma samples were assayed for
the presence of TNFalpha, IL-1beta, IL-6, IL-12, IL-10 and VEGF per-
forming a LuminexÒ multiplex screening assay (R&D Systems)
according to the manufacturers’ instructions. Samples were ana-
lyzed by Luminex 200 technology using Luminex IS software
(Luminex Corporation, USA). In brain lysates concentrations of IL-
12, in plasma samples concentrations of IL-1beta, IL-6, IL-12 and
in culture supernatants concentrations of IL-1beta, IL-12, IL-6 and
VEGF were out of the detection range of the multiplex assay. Pro-
tein abundance for BDNF, EGF and IGF was determined by ELISA
(R&D Systems) according to manufacturers’ instructions.
2.13. Statistical analysis
All results are presented as scatter blots with bars presenting
mean values unless stated otherwise. Sample size was determined
a priori using G⁄Power (version 3.1). An a-level of 0.05 and a
power of 0.8 were required. Based on our previous report
(Reinboth et al., 2016), a Cohen’s d ES of 0.5 and a mortality of
15% was assumed yielding a final sample size of at least 14 animals
per group for sub-acute brain injury and long-term functional out-
come. For statistical analysis, the GraphPad Prism 6.0 software
package (GraphPad Software, USA) was used. After checking for
normal distribution either ordinal 1-way ANOVA with post hoc
Bonferroni multiple comparisons test or Kruskal-Wallis with
Dunn’s multiple comparisons test were applied. For comparison
 J. Herz et al. / Brain, Behavior, and Immunity 70 (2018) 118–130 123

of two groups either Students’ t-test or Mann Whitney U test was treatment leading to a significantly reduced novel object activity
used. A p value of <0.05 was considered significant. compared to sham controls (Fig. 1D).

3.2. Combining acute hypothermia with delayed MSC treatment

3. Results diminishes early neuroprotection

3.1. Improved long-term neurological function by acute hypothermia
and delayed intranasal MSC treatment is decreased following
combined treatment
Using our established HI/HT protocol for neonatal C57BL/6 mice
we confirmed that the previously determined nesting temperature
of 35 °C and the target temperature difference of 3 °C (Reinboth
et al., 2016) was maintained during experimental modelling
(Supplementary Fig. S2A). HI animals revealed significantly
reduced body weight gain compared to sham-operated animals
early after HI independent of treatment (Supplementary
Fig. S2B). Confounding effects on neurobehavioral assessment by
different weight gain could be excluded because no significant
differences were observed across all experimental groups 5 weeks
after HI (Supplementary Fig. S2B).
Exploratory and impulsive behavior, analyzed by elevated plus
maze, demonstrated that both single therapies (HT and MSC treat-
ment) reduced HI-induced alterations with no significant differ-
ences compared to sham-operated control mice (Fig. 1B).
However, combined treatment reversed protective effects by HT
and MSC leading to significantly increased time intervals mice
spent in the open arms compared to sham mice (Fig. 1B). A compa-
rable regulation was observed for the time of head-dipping in the
open and center regions of the maze (Fig. 1B). General motor activ-
ity assessed in the open field maze was not significantly modulated
by either treatment (Supplementary Fig. S3A). Similar to observa-
tions made in the elevated plus maze HT + MSC animals spent
longer times in the center region of the open field maze (Supple-
mentary Fig. 3B). HT significantly improved vertical activity, as a
measure of motor function, which was diminished after combina-
tion with MSC therapy (Fig. 1C). Cognitive function, evaluated by
novel object recognition was reduced in NT mice and was not
recovered by HT. However, MSC significantly improved long-term
cognitive function (Fig. 1D), being reversed when HT preceded
To assess whether the observed detrimental interaction was ini-
tiated early after combining both treatments, we analyzed
histopathological, cellular and structural grey and white matter
injury at one week post HI.
Total and local hippocampal injury was significantly reduced by
both, HT and MSC (Fig. 2). After combining both treatments, pro-
tective effects were decreased resulting in significantly increased
injury scores of HT + MSC compared to single MSC therapy for total
and local injury within the hippocampus (Fig. 2). In cortex and
striatum, HT- and MSC-induced significant protection was
reversed after combined treatment (Supplementary Fig. S4A, B).
In the thalamus, all administered therapies revealed similar pro-
tection (Supplementary Fig. S4C).
For further insight into target structures with respect to sub-
acute grey and white matter injury, western blot analyses for
MAP-2 and MBP expression at the level of the striatum and the hip-
pocampus were performed. NT animals demonstrated a marked
loss of both axonal and myelin structures in ipsilateral hemi-
spheres (Fig. 3). Whereas HT significantly improved expression of
MAP-2 expression, MSC treatment did not have an effect on axonal
degeneration (Fig. 3). In contrast, myelination assessed by quantifi-
cation of MBP expression was significantly improved by MSC treat-
ment but not by HT (Fig. 3). Interestingly, combination of HT and
MSC resulted in a significant improvement of MAP-2 expression
at the hippocampal level, whereas enhanced myelination by MSC
was abolished in animals treated with HT before MSC administra-
tion (Fig. 3).
To analyze whether differential effects on structural damage
might be attributed to cellular loss or differentiation deficits, we
quantified neuronal and oligodendrocyte density via immunohis-
tochemistry for NeuN and Olig2. The most pronounced reduction
of neuronal density was observed in the hippocampus and stria-
tum, which was ameliorated by single HT and MSC treatment
(Fig. 4A, B). However, after combining both therapies the single

Fig. 2. Combining acute hypothermia with delayed MSC treatment diminishes protective effects on sub-acute histological brain injury. Histological brain injury was
determined on cresyl violet stained 20 mm cryostat sections of P16 mice that were exposed to HI followed by 4 h NT or HT on P9. MSC were administered intranasally at P12.
(A) Representative images of injured hemispheres for each experimental group are shown (scale bar: 1 mm). (B) Pictures display higher magnification images (scale bar:
100 mm) of squares depicted in (A) revealing severe cell loss throughout the pyramidal cell layer of the hippocampus in NT and HT + MSC animals compared to small focal
areas of cell loss in HT and MSC single treatments (arrows). (C) Injury scores were assessed in different brain regions (cortex, hippocampus, striatum, thalamus) resulting in a
sum score quantified for each animal (left column). Quantification of regional injury scores in the hippocampus is depicted in the right column. Bars present median values.
n = 14–16/group, *p < 0.05, **p < 0.01, ***p < 0.001, NT = normothermia, HT = hypothermia, MSC = mesenchymal stem cells.
124 J. Herz et al. / Brain, Behavior, and Immunity 70 (2018) 118–130

Fig. 3. Differential effects of acute hypothermia and delayed MSC therapy on grey and white matter injury. P9 mice were exposed to HI followed by 4 h NT or HT. MSC were
administered intranasally at P12. Protein lysates were isolated from ipsilateral hemispheres at the hippocampal (A, B) and the striatal level (C, D) at P16. Protein abundance of
MAP-2 and MBP was quantified by western blot. Data were normalized to the reference protein GAPDH and sham controls. n = 12–14/group, *p < 0.05, **p < 0.01, ***p < 0.001,
NT = normothermia, HT = hypothermia, MSC = mesenchymal stem cells.

therapy-effects were diminished resulting in no significant differ-
ences compared to NT animals (Fig. 4A, B). While neuronal density
is decreased, HI induced a significant increase in oligodendrocytes
in most severely affected brain regions (i.e. striatum, hippocam-
pus) which was not modulated by HT, MSC or combined treatment
(Fig. 4C, D).
3.3. Anti-inflammatory effects of acute hypothermia and delayed MSC
treatment are reversed after combined treatment
To dissect the underlying mechanisms of opposite effects by
the combined treatment, we analyzed sub-acute neuroinflamma-
tory responses. HT, MSC and the combination of both demon-
strated a marked down-regulation of microglia activation
assessed by Iba-1 protein expression (Fig. 5A). As these results
may not explain the detrimental interaction of both therapies,
we analyzed peripheral leukocyte infiltration and endothelial acti-
vation. Whereas HT and MSC inhibited HI-induced infiltration of
leukocytes, animals exposed to the combined treatment showed
an increased leukocyte infiltration with no significant differences
compared to normothermic controls (Fig. 5B). Similar to these
results, HI-induced up-regulation of the endothelial adhesion
molecule VCAM-1 in NT animals was significantly reduced by HT
and MSC (Fig. 5C), which was reversed after combining both ther-
apies (Fig. 5C). Whether increased VCAM-1 expression was caused
by general modulation of vascular integrity was examined by eval-
uation of blood vessel density and diameters. HI-induced vessel
loss and dilation was improved by all therapeutic interventions
(Fig. 5D).
3.4. MSC home to HI-injured brains independent of treatment
To investigate whether observed effects might have been
caused by impaired MSC homing in HT-treated animals we
quantified GFP-labelled MSC in brains of NT and HT-treated ani-
mals 14–16 h after intranasal delivery. These analyses revealed a
huge inter-individual variation, which was, however, independent
of treatment (Supplementary Fig. S5A,B) suggesting a similar
migratory behavior of MSC in NT and HT animals. Of note, no
GFP-positive cells were detected in contralateral hemispheres
(Supplementary Fig. S5A). Overall cell number was low implicating
that MSC may also have been distributed to other organ systems
 J. Herz et al. / Brain, Behavior, and Immunity 70 (2018) 118–130 125

Fig. 4. Protection from regional neuronal loss by acute hypothermia and delayed MSC therapy is decreased after combining treatments. Regional neuronal and
oligodendrocyte densities were determined in cortex, striatum, hippocampus and thalamus of P16 mice exposed to HI followed by 4 h NT or HT on P9. MSC were
administered intranasally at P12. (A, B) Neuronal density was assessed by immunohistochemistry quantifying the NeuN-positive area. (C, D) Oligodendrocyte density was
quantified by counting Olig2-positive cells. Representative images in (A) and (C) show low and higher magnification images of the hippocampus. Scale bars: low
magnification: 250 mm, high magnification: 25 mm. n = 14–16/group, *p < 0.05, **p < 0.01, ***p < 0.001, NT = normothermia, HT = hypothermia, MSC = mesenchymal stem
cells.

thereby potentially mediating systemic effects. Therefore, we mea- TNFalpha and IL-10 while slightly increased levels of IL-4 in NT ani-
sured blood plasma cytokine levels for all experimental groups mals were significantly reduced by single MSC therapy and the
demonstrating no significant alteration by injury or therapy for combined treatment with HT (Supplementary Fig. S5C). However,
126 J. Herz et al. / Brain, Behavior, and Immunity 70 (2018) 118–130

Fig. 5. Anti-inflammatory effects of hypothermia and MSC on endothelial activation and leukocyte infiltration are reversed after combination therapy. P9 mice were exposed
to HI followed by 4 h NT or HT. MSC were administered intranasally at P12. Analysis was performed at P16. (A) Microglia activation was assessed via western blot quantifying
Iba-1 protein levels. (B) Leukocyte infiltration was evaluated by immunohistochemistry for CD45. Representative images show low and high magnification images of the
hippocampus. Scale bars: low magnification: 250 mm, high magnification: 25 mm. The CD45 positively stained area was quantified. (C) VCAM-1 expression was analyzed in
protein lysates via western blot. For (A) and (C) data were normalized to the reference protein GAPDH and sham controls. (D) Blood vessel density and diameter were
determined via immunohistochemistry. Mean and SD values of uninjured contralateral tissues are indicated by solid and dashed lines, respectively. n = 12–14/group for
(A) and (C), n = 14–16/group for (B) and (D), *p < 0.05, ***p < 0.001, NT = normothermia, HT = hypothermia, MSC = mesenchymal stem cells.

no differences were detected between MSC and HT + MSC treated
animals, thus not explaining interaction effects for HI-induced
brain injury.
3.5. Modulation of MSC phenotype after culture with brain extracts of
hypothermia-treated animals
According to a similar homing of MSC to injured hemispheres in
NT and HT treated animals we hypothesized that MSC change their
phenotype due to the changed in vivo microenvironment in cooled
animals. Therefore, we analyzed gene expression in MSC for cytoki-
nes and growth factors in MSC, co-cultured with conditioned med-
ium containing brain extracts obtained from P12 mice exposed to
sham operation, HI + NT or HI + HT on P9 (Fig. 6A). In response to
culture with HT-brain homogenate, MSC up-regulated pro-
inflammatory cytokine expression (e.g. IL-1beta and IL-6) while
expression of the anti-inflammatory cytokine IL-10 was reduced
compared to co-culture with brain extracts derived from NT-
treated control mice (Fig. 6B). Furthermore, NT-brain homogenates
led to increased growth factor expression (e.g. BDNF and EGF)
which was significantly reduced after incubation with brain
extracts obtained from HT-treated animals (Fig. 6B). Analysis of
cell culture supernatants to determine protein secretion by MSC
partially confirmed mRNA analysis demonstrating reduced levels
of IL-10 and IGF in HT-brain homogenate co-cultures compared
to NT-brain co-cultures (Fig. 6C). For few analyses (e.g. BDNF and
TNFalpha) no significant differences were observed, most likely
due to the simultaneous time point of analysis, i.e. transcriptional
regulation preceding complex processes of protein translation and
intracellular targeting for extracellular secretion. Medium contain-
 J. Herz et al. / Brain, Behavior, and Immunity 70 (2018) 118–130 127

Fig. 6. Increased pro-inflammatory cytokine expression and reduced growth factor expression by MSC exposed to brain extracts from hypothermia-treated animals. MSC
were cultured for 48 h with brain extracts obtained from P12 mice that were exposed to HI followed by 4 h NT or HT on P9 (A). mRNA expression (B) and protein analysis in
culture supernatants (C) for cytokines and growth factors was assessed by real time PCR on cell lysates (B) and ELISA/LuminexÒ in culture supernatants (C) in MSC co-cultures
with conditioned medium containing isolated brain extracts of normothermia (NT brain) or hypothermia (HT brain) treated animals (A). mRNA data were normalized to the
reference gene beta-2-mircoglobulin and to MSC co-cultured with brain homogenates of sham controls (n = 6, presented as dashed line, B). Dashed lines in (C) indicate
the mean value of MSC co-cultured with brain homogenates of sham controls (n = 4). n = 8/group for (B) and n = 6/group for (C). Data points reveal biological replicates/group,
*p < 0.05, **p < 0.01, ***p < 0.001 NT = normothermia, HT = hypothermia.

ing brain extracts at culture concentration (1 mg/ml) without cells treatment induced a further decline below control levels
demonstrated cytokine and growth factor concentrations below (Fig. 7B). Of note, VEGF is down-regulated by HI and improved
the detection limit, thus not confounding results of these analyses. by HT, MSC but also by the combination (Fig. 7B).

3.6. Combination of hypothermia and delayed MSC therapy increases 4. Discussion

pro-inflammatory cytokines and reduces neurotrophic factor levels in
HI-injured brains
To verify in vitro results cytokine and growth factor levels were
analyzed in brain tissue lysates 24 h after intranasal administra-
tion of MSC or vehicle in NT and HT-treated animals. These analy-
ses partially confirmed in vitro data, e.g. IL-6 and IL-1beta levels
were increased by the combination of HT and MSC compared to
single therapies (Fig. 7A). For IL-10 slightly increased levels in sin-
gle treatments were reduced by HT + MSC though not reaching sig-
nificance (Fig. 7A). Growth factor concentrations (except for VEGF)
were increased to abnormal high levels by HI but normalized to
control levels by HT and MSC single therapy, while the combined
The present study shows that acute hypothermia (HT) following
neonatal hypoxia–ischemia (HI) results in protection of neuronal
structures and long-term motor outcome, while white matter
injury and long-term cognitive function are not improved. In con-
trast, a single delayed therapy with mesenchymal stem cells (MSC)
reduced HI-induced hypomyelination and memory deficits. Sur-
prisingly, the combined treatment reversed protective effects of
each single therapy. Inhibitory effects on endothelial activation
and associated peripheral leukocyte infiltration by both monother-
apies were abolished after combination. As a potential mechanism
we provide evidence that modulation of the cerebral microenvi-
ronment by HT alters the MSC phenotype, resulting in increased
128 J. Herz et al. / Brain, Behavior, and Immunity 70 (2018) 118–130

Fig. 7. Modulation of cytokine and growth factor levels by acute hypothermia and delayed intranasal MSC delivery 4 days after HI. P9 mice were exposed to HI followed by
4 h NT or HT. MSC were administered intranasally at P12. Cytokine (A) and growth factor (B) analysis was performed 24 h after MSC delivery. Protein lysates of ischemic
hemispheres were analyzed by LuminexÒ (TNFalpha, IL-1beta, IL-6, IL-10, VEGF) and ELISA (BDNF, EGF, IGF). Dashed lines indicate mean values of sham-operated control
mice (n = 7). n = 8/group, *p < 0.05, **p < 0.01, ***p < 0.001 NT = normothermia, HT = hypothermia, MSC = mesenchymal stem cells.

expression of pro-inflammatory cytokines and reduced expression
of growth factors, leading to impaired restoration of long-term
neurobehavioral development.
Despite an increasing number of currently initiated clinical trials
of cell based therapies treating neonatal brain injury, only few pre-
clinical reports focused on the combined treatment with HT (Fleiss
et al., 2014). Intraventricular delivery of human umbilical cord
blood MSC prior to HT revealed significant long-term protection
from brain atrophy in a rodent model of neonatal HI (Park et al.,
2015). However, with regard to clinical translation less invasive
delivery routes with a prolonged therapeutic window would be pre-
ferred. Therefore, we applied MSC in the sub-acute phase through
the nasal route, reflecting a clinically applicable treatment design.
Unexpectedly, we observed a negative interplay of both therapies.
Confirming previous reports (Burnsed et al., 2015; Reinboth
et al., 2016; Sabir et al., 2012), we show that HT promotes short-
term protection, particularly from neuronal/axonal injury, result-
ing in long-lasting improvement of motor deficits. However, myeli-
nation was not restored by HT leading to impaired long-term
cognitive abilities. MSC only, but not HT or HT + MSC therapy,
improved myelination and memory function. These results under-
line the importance of white matter development for long-term
neurological outcome after perinatal brain injury. Nevertheless,
our results seem to contrast a previous report revealing that myeli-
nation is enhanced by HT, which might be explained by different
methods of analyses (immunohistochemistry vs. western blot)
and by the fact that protection was regionally restricted to the cau-
date nucleus (Koo et al., 2017).
Since in the investigated therapy design combined treatment
with MSC and HT revealed detrimental effects compared to single
therapies, we evaluated the underlying mechanisms instead of
investigating further time points. MSC homing may have been
impaired due to anti-inflammatory signals elicited by HT, resulting
in reduced abundance of pro-inflammatory chemo-attractants for
MSC migration to the injury site. However, the amount of accumu-
lated MSC in injured hemispheres of NT and HT animals was sim-
ilar. Furthermore, our results revealed that both single treatment
effects were reversed, i.e. HT-induced protection is decreased after
additive MSC therapy. These results support the hypothesis that
MSC reach the injury region in HT mice, but their neuroprotective
capacity seems to be altered by acute HT preceding delayed MSC
therapy.
Protective effects on microglia were similar for HT, MSC and the
combined treatment. Furthermore, HI-induced downregulation of
VEGF is improved by all treatments at P13, closely resembling
effects on vessel density and integrity observed at P16. Therefore,
detrimental interaction effects may be rather attributed to modu-
lation of endothelial activation and associated leukocyte infiltra-
tion. Inhibition of HI-induced up-regulation of vascular cell
adhesion molecule expression (as a marker of endothelial activa-
tion) and associated peripheral leukocyte infiltration by single
therapies was absent in the combined treatment. These results
suggest that MSC exposed to the anti-inflammatory environment
induced by HT, may increase inflammation. This is supported by
our in vitro analyses, demonstrating that MSC exposed to HT brain
extracts increase expression of pro-inflammatory cytokines (e.g.
IL-6 and IL-1beta) while reducing expression of the anti-
inflammatory cytokine IL-10. Similar results were obtained
in vivo revealing increased IL-6 and IL-1beta concentrations in
the combined treatment compared to single therapies early after
 J. Herz et al. / Brain, Behavior, and Immunity 70 (2018) 118–130
MSC administration. This pro-inflammatory environment most
likely contributed to induction of vascular cell adhesion molecule
expression (O’Carroll et al., 2015) resulting in increased peripheral
leukocyte infiltration.
In addition to immunomodulation, MSC act through promotion
of regeneration via growth factor release. Importantly, MSC
expressed more BDNF and EGF following contact with an HI envi-
ronment in vitro, supporting previous reports that MSC promote a
pro-regenerative environment in the HI brain (Donega et al., 2014;
van Velthoven et al., 2010b). However, after exposure to HT brain
extracts BDNF expression significantly decreased suggesting that
MSC’s pro-regenerative function is impaired in the brain of HT-
treated mice. Of note, in vivo analysis of BDNF, EGF and IGF demon-
strated an abnormal increase in HI-injured brains, which was
downregulated to control levels by single therapies while the com-
bined treatment induced a further decline below control levels.
These results support the hypothesis that a certain threshold con-
centration of these factors is particularly important for outcome,
with mild to moderate changes in growth factor signaling having
potentially positive effects, but extremes being unfavorable
(Binder et al., 2001; Chavko et al., 2002; Conover, 2016). Taken
together, our results suggest that HT modulates MSC function
inducing a pro-inflammatory environment combined with alter-
ation of the homeostatic growth factor milieu which may account
for the observed reduction in therapeutic efficacy on myelination
and long-term cognitive function.
Our study has strengths and limitations. Strengths of the
current study are large animal numbers and assessment of
long-lasting motor and also cognitive outcome, not previously
investigated. Multimodal approaches on potential underlying
mechanisms were used to help to define more appropriate treat-
ment designs. We administered bone marrow derived MSC from
the same species with well characterized treatment effects, a wide
therapeutic window, the possibility of a non-invasive application
mode combined with a reduced risk for adverse effects driven by
a potential mismatch between human and murine systems. A
potential limitation regarding comparability to previous work
and current initiated clinical trials is the use of bone marrow
MSC instead of cells derived from the umbilical cord, which are
supposed to be more primitive than MSC isolated from adult tissue
sources. However, their therapeutic capacity is still under debate
(Fleiss et al., 2014). The detected cell number of infiltrated MSC
was low and revealed a huge inter-individual variation not allow-
ing reliable immunohistochemical analysis to determine spatial
distribution of cells. Even though this was independent of NT or
HT treatment, further systematic kinetic analyses are needed to
determine the spatiotemporal distribution of MSC after intranasal
delivery. Another potential limitation might be that stereology
methods have not been applied. However, quantification was per-
formed according to our previous study (Reinboth et al., 2016)
where we established the injury model and immunohistochemical
analyses. When comparing different quantification methods (serial
section and total region analysis vs. methods described here) a
close correlation was observed between both methods (not shown)
while the inter-individual, i.e. biological variability was high, as
frequently reported in this injury model (Engels et al., 2017;
Patel et al., 2015; Rice et al., 1981; Sabir et al., 2012). Therefore,
we choose to include higher biological replicates (i.e. sample size)
rather than increasing technical replicates. Even though we applied
a broad set of behavior tests covering locomotion, anxiety and
impulsive related behavior as well as cognitive function, further
cognitive tests, e.g. addressing learning behavior might be critical
to confirm our findings.
Our results confirm efficacy but also limitations of HT and MSC
monotherapy for treatment of neonatal HI. In contrast to our
hypothesis, the combination of acute HT with delayed MSC therapy
129
deteriorated long-term protection from neurological deficits com-
pared to single therapies. We provide evidence that the detrimen-
tal interaction may be most likely caused by an alteration of the
MSC phenotype in the HT-treated brain resulting in increased
inflammatory responses and reduced expression of pro-
regenerative factors. Even though we confirmed MSC migration
into injured HI brains in rodents, further investigations especially
in higher developed organisms comparable to humans are needed
to elucidate the possible disposition of intranasally applied cells
and their distribution within the brain to demonstrate safety and
efficacy of this approach. Furthermore, several risk factors con-
tributing to neonatal encephalopathy should be considered. For
instance, in cases of infection-sensitized HI loss of protection by
HT (Osredkar et al., 2014) most likely caused by overwhelming
inflammation might be overcome by stem cell therapy. Neverthe-
less, our unexpected results call for caution for precipitately
designed clinical trials for adjunctive cell-based therapies in com-
bination with hypothermia, since careful research is required to
define the optimum timing and application mode.
5. Source of funding
This work was supported by the German Research Council (FE
518/5-1, UFM), an IFORES grant by the medical faculty (JH) and
grants by the C.D. and the Karl-Heinz-Frenzen foundation
(UFM;IB).
6. Disclosure
None
Acknowledgment
We thank R. Herrmann for assistance and discussion of real
time PCR data, S. Lupus for technical assistance in LuminexÒ
assays, J. Göthert and his team providing the opportunity to
perform flow cytometry analysis with BD FACS LSRII.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at https://doi.org/10.1016/j.bbi.2018.02.006.
References
Azzopardi, D., Strohm, B., Marlow, N., Brocklehurst, P., Deierl, A., Eddama, O.,
Goodwin, J., Halliday, H.L., Juszczak, E., Kapellou, O., Levene, M., Linsell, L., Omar,
O., Thoresen, M., Tusor, N., Whitelaw, A., Edwards, A.D., Group, T.S., 2014. Effects
of hypothermia for perinatal asphyxia on childhood outcomes. N. Engl. J. Med.
371, 140–149
Bevins, R.A., Besheer, J., 2006. Object recognition in rats and mice: a one-trial non-
matching-to-sample learning task to study ’recognition memory’. Nat. Protoc. 1,
1306–1311.
Binder, D.K., Croll, S.D., Gall, C.M., Scharfman, H.E., 2001. BDNF and epilepsy: too
much of a good thing? Trends Neurosci. 24, 47–53.
Burnsed, J.C., Chavez-Valdez, R., Hossain, M.S., Kesavan, K., Martin, L.J., Zhang, J.,
Northington, F.J., 2015. Hypoxia-ischemia and therapeutic hypothermia in the
neonatal mouse brain–a longitudinal study. PLoS One 10, e0118889.
Chavko, M., Nadi, N.S., Keyser, D.O., 2002. Activation of BDNF mRNA and protein
after seizures in hyperbaric oxygen: implications for sensitization to seizures in
re-exposures. Neurochem. Res. 27, 1649–1653.
Conover, C.A., 2016. Discrepancies in insulin-like growth factor signaling? No, not
really. Growth Horm. IGF Res. 30–31, 42–44.
Donega, V., Nijboer, C.H., van Tilborg, G., Dijkhuizen, R.M., Kavelaars, A., Heijnen, C.J.,
2014. Intranasally administered mesenchymal stem cells promote a
regenerative niche for repair of neonatal ischemic brain injury. Exp. Neurol.
261, 53–64.
Donega, V., van Velthoven, C.T., Nijboer, C.H., van Bel, F., Kas, M.J., Kavelaars, A.,
Heijnen, C.J., 2013. Intranasal mesenchymal stem cell treatment for neonatal
brain damage: long-term cognitive and sensorimotor improvement. PLoS One 8,
e51253.
130 J. Herz et al. / Brain, Behavior, and Immunity 70 (2018) 118–130
Engels, J., Elting, N., Braun, L., Bendix, I., Herz, J., Felderhoff-Muser, U., Dzietko, M.,
2017. Sildenafil Enhances Quantity of Immature Neurons and Promotes
Functional Recovery in the Developing Ischemic Mouse Brain. Dev. Neurosci.
39, 287–297.
Eunson, P., 2015. The long-term health, social, and financial burden of hypoxic-
ischaemic encephalopathy. Dev. Med. Child Neurol. 57 (Suppl 3), 48–50.
Ferriero, D.M., 2004. Neonatal brain injury. N. Engl. J. Med. 351, 1985–1995.
Fleiss, B., Guillot, P.V., Titomanlio, L., Baud, O., Hagberg, H., Gressens, P., 2014. Stem
cell therapy for neonatal brain injury. Clin. Perinatol. 41, 133–148.
Gunn, A.J., Laptook, A.R., Robertson, N.J., Barks, J.D., Thoresen, M., Wassink, G.,
Bennet, L., 2017. Therapeutic hypothermia translates from ancient history in to
practice. Pediatr. Res. 81, 202–209.
Herz, J., Hagen, S.I., Bergmuller, E., Sabellek, P., Gothert, J.R., Buer, J., Hansen, W.,
Hermann, D.M., Doeppner, T.R., 2014. Exacerbation of ischemic brain injury in
hypercholesterolemic mice is associated with pronounced changes in
peripheral and cerebral immune responses. Neurobiol. Dis. 62, 456–468.
Herz, J., Sabellek, P., Lane, T.E., Gunzer, M., Hermann, D.M., Doeppner, T.R., 2015.
Role of Neutrophils in Exacerbation of Brain Injury After Focal Cerebral Ischemia
in Hyperlipidemic Mice. Stroke 46, 2916–2925.
Koo, E., Sheldon, R.A., Lee, B.S., Vexler, Z.S., Ferriero, D.M., 2017. Effects of
therapeutic hypothermia on white matter injury from murine neonatal
hypoxia-ischemia. Pediatr. Res.
Lister, R.G., 1987. The use of a plus-maze to measure anxiety in the mouse.
Psychopharmacology 92, 180–185.
Milner, L.C., Crabbe, J.C., 2008. Three murine anxiety models: results from multiple
inbred strain comparisons. Genes Brain Behav. 7, 496–505.
O’Carroll, S.J., Kho, D.T., Wiltshire, R., Nelson, V., Rotimi, O., Johnson, R., Angel, C.E.,
Graham, E.S., 2015. Pro-inflammatory TNFalpha and IL-1beta differentially
regulate the inflammatory phenotype of brain microvascular endothelial cells. J.
Neuroinflammation 12, 131.
Osredkar, D., Thoresen, M., Maes, E., Flatebo, T., Elstad, M., Sabir, H., 2014.
Hypothermia is not neuroprotective after infection-sensitized neonatal
hypoxic-ischemic brain injury. Resuscitation 85, 567–572.
Pappas, A., Shankaran, S., McDonald, S.A., Vohr, B.R., Hintz, S.R., Ehrenkranz, R.A.,
Tyson, J.E., Yolton, K., Das, A., Bara, R., Hammond, J., Higgins, R.D., 2015.
Hypothermia Extended Follow-up Subcommittee of the Eunice Kennedy
Shriver, N.N.R.N. Cognitive outcomes after neonatal encephalopathy.
Pediatrics 135, e624–634.
Park, W.S., Sung, S.I., Ahn, S.Y., Yoo, H.S., Sung, D.K., Im, G.H., Choi, S.J., Chang, Y.S.,
2015. Hypothermia augments neuroprotective activity of mesenchymal stem
cells for neonatal hypoxic-ischemic encephalopathy. PLoS One 10, e0120893.
Patel, S.D., Pierce, L., Ciardiello, A., Hutton, A., Paskewitz, S., Aronowitz, E., Voss, H.U.,
Moore, H., Vannucci, S.J., 2015. Therapeutic hypothermia and hypoxia-ischemia
in the term-equivalent neonatal rat: characterization of a translational
preclinical model. Pediatr. Res. 78, 264–271.
Reinboth, B.S., Koster, C., Abberger, H., Prager, S., Bendix, I., Felderhoff-Muser, U.,
Herz, J., 2016. Endogenous hypothermic response to hypoxia reduces brain
injury: Implications for modeling hypoxic-ischemic encephalopathy and
therapeutic hypothermia in neonatal mice. Exp. Neurol. 283, 264–275.
Rice 3rd, J.E., Vannucci, R.C., Brierley, J.B., 1981. The influence of immaturity on
hypoxic-ischemic brain damage in the rat. Ann. Neurol. 9, 131–141.
Sabir, H., Scull-Brown, E., Liu, X., Thoresen, M., 2012. Immediate hypothermia is not
neuroprotective after severe hypoxia-ischemia and is deleterious when delayed
by 12 hours in neonatal rats. Stroke 43, 3364–3370.
Sheldon, R.A., Sedik, C., Ferriero, D.M., 1998. Strain-related brain injury in neonatal
mice subjected to hypoxia-ischemia. Brain Res. 810, 114–122.
Thornton, C., Rousset, C.I., Kichev, A., Miyakuni, Y., Vontell, R., Baburamani, A.A.,
Fleiss, B., Gressens, P., Hagberg, H., 2012. Molecular mechanisms of neonatal
brain injury. Neurol. Res. Int. 2012, 506320.
van Velthoven, C.T., Kavelaars, A., van Bel, F., Heijnen, C.J., 2010a. Mesenchymal
stem cell treatment after neonatal hypoxic-ischemic brain injury improves
behavioral outcome and induces neuronal and oligodendrocyte regeneration.
Brain Behav. Immun. 24, 387–393.
van Velthoven, C.T., Kavelaars, A., van Bel, F., Heijnen, C.J., 2010b. Repeated
mesenchymal stem cell treatment after neonatal hypoxia-ischemia has distinct
effects on formation and maturation of new neurons and oligodendrocytes
leading to restoration of damage, corticospinal motor tract activity, and
sensorimotor function. J. Neurosci. 30, 9603–9611.
Walf, A.A., Frye, C.A., 2007. The use of the elevated plus maze as an assay of anxiety-
related behavior in rodents. Nat. Protocols 2, 322–328.
Wassink, G., Gunn, E.R., Drury, P.P., Bennet, L., Gunn, A.J., 2014. The mechanisms and
treatment of asphyxial encephalopathy. Front. Neurosci. 8, 40.