Short-, Mid-, and Long-Term Effect of Granulocyte Colony-Stimulating Factor/Stem Cell Factor and Fms-Related Tyrosine Kinase 3 Ligand Evaluated in an In Vivo Model of Hypoxic-Hyperoxic Ischemic Neonatal Brain Injury

Hematopoietic growth factors are considered to bear neuroprotective potential. We have previously shown that delayed treatment with granulocyte colony-stimulating factor (G-CSF)/stem cell factor (SCF) and Fms-related tyrosine kinase 3 ligand (FL) ameliorates excitotoxic neonatal brain injury. The effect of these substances in combined-stressor neonatal brain injury models more closely mimicking clinical conditions has not been investigated. The aim of this study was to assess the short-, mid-, and long-term neuroprotective potential of G-CSF/SCF and FL in a neonatal model of hypoxic-hyperoxic ischemic brain injury. Five-day-old (P5) CD-1 mice were subjected to unilateral common carotid artery ligation and subsequent alternating periods of hypoxia and hyperoxia for 65 minutes. Sixty hours after injury, pups were randomly assigned to intraperitoneal treatment with (i) G-CSF (200 μg/kg)/SCF (50 μg/kg), (ii) FL (100 μg/kg), or (iii) vehicle every 24 hours for three or five consecutive days. Histopathological and functional outcomes were evaluated on P10, P18, and P90. Baseline outcome parameters were established in sham-treated and healthy control animals. Gross brain injury did not significantly differ between treatment groups at any time point. On P10, caspase-3 activation and caspase-independent apoptosis were similar between treatment groups; cell proliferation and the number of BrdU-positive vessels did not differ on P18 or P90. Neurobehavioral assessment did not reveal significant differences between treatment groups in accelerod performance, open field behavior, or novel object recognition capacity on P90. Turning behavior was more frequently observed in G-CSF/SCF- and FL-treated animals. No sex-specific differences were detected in any outcome parameter evaluated. In hypoxic-hyperoxic ischemic neonatal brain injury, G-CSF/SCF and FL treatment does not convey neuroprotection. Prior to potential clinical use, meticulous assessment of these hematopoietic growth factors is mandated.


Introduction
Neonatal brain injury is a problem of great global concern [1,2]. Advances in perinatal care have improved survival rates of affected infants, but long-term morbidity is still substantial [3,4]. Etiopathogenesis is complex and incompletely understood, and various mechanisms are thought to contribute, including excitotoxicity, hypoxia-ischemia, inflammation, and additional factors such as oxidative stress/hyperoxia [5][6][7]. To date, causal pharmacologic therapies are not available for routine clinical use. The most auspicious treatment options are those that intervene on multiple levels and are able to not only inhibit cell death and promote cell survival, but also foster regeneration and functional reorganisation [8][9][10]. Interventions augmenting endogenous neuroprotective responses, among them cytokines and growth factors, hold particular promise [11][12][13][14]. Hematopoietic growth factors have been extensively studied and used in neonates for hematologic indications for several years [15]. In addition to their hematopoietic effects, they may have other potential benefits, including neuroprotection, neural tissue repair, and neurovasculogenesis [15][16][17]. The effects of these agents on 2 BioMed Research International neuronal death or survival, however, seem to be contextspecific [12], conveying neuroprotection in some neonatal brain injury models [18][19][20][21][22][23][24][25][26][27][28], while having no effect or even exacerbating damage in others [29,30].
Our research group has recently been able to show that systemic treatment with the hematopoietic growth factors/cytokines granulocyte colony-stimulating factor (G-CSF)/stem cell factor (SCF) and Fms-related tyrosine kinase 3 ligand (FL) starting 60 hours after insult protects against Nmethyl-D-aspartate receptor-mediated developmental excitotoxic brain damage by reducing injury extent and apoptotic cell death [31]. Whether these treatment regimina are effective and safe in other injury models, especially those encompassing multiple stressors, has not yet been investigated. Prior to clinical use, however, thorough assessment of treatment options in combined-stressor brain injury models is of utmost importance, as these may more adequately reflect and mimic the conditions neonates experience in intensive care units [32].
Furthermore, any neuromodulatory intervention taking place during potentially critical or sensitive periods of brain development possibly has not only short-, but also long-term impact [33]. The assessment of behavioral endpoints is thus crucial in order to evaluate whether morphological effects translate into differences in functional outcome [20].
Thus, the aim of the current study was to investigate short-, mid-, and long-term effects of delayed treatment with G-CSF/SCF and FL in a mouse model of combined hypoxichyperoxic ischemic neonatal brain injury. From our previous findings we hypothesised that G-CSF/SCF and FL reduce injury extent and apoptotic cell death, stimulate neuroproliferation, and improve neurobehavioral outcome following hypoxia-hyperoxia ischemia.
. . Animal Experiments. All animal studies were conducted in compliance with current EU legislation (Directive 2010/ 63/EU revising Directive 86/609/EEC) and Austrian law. Permission for this study was obtained by the ethics committee of the Austrian Federal Ministry of Education, Science and Research (BMWF-66.011/0131-II/10b/2008). CD-1 mice (Crl:CD1 (ICR), Charles River Laboratories, Sulzfeld, Germany) were bred and kept at the Central Laboratory Animal Facility, Medical University of Innsbruck, Austria, under standard housing conditions with a 12-hour lightdark cycle, a room temperature of 25 ∘ C, and a humidity of 75%. Five-day-old (P5) CD-1 mice of both sexes with a body weight (bw) of 3 ± 0.3 g were eligible for the study. For all experiments humane endpoints were considered as follows: Animals were controlled daily for pathological and behavioral changes, including exploratory activity, thirst, and drinking. Following the experimental procedures daily inspection of the animals was performed by trained personnel to detect signs of infection, pain, or discomfort. Abnormal feeding behavior, abnormal body performance, or isolation of animals was reported. If any sign was positive, reevaluation was performed the same day and if persistent a veterinary was asked to further evaluate the animal. Protocol included that, in case animals reached endpoint criteria, euthanasia would be performed immediately. Hypoxic-hyperoxic ischemic (HHI) brain injury was induced as described previously [34]. In brief, mouse pups were subjected to right common carotid artery ligation under general anesthesia (isoflurane in oxygen, 3.0 vol% induction/1.5 vol% maintenance; AbbVie, Vienna, Austria). After a 120-minute recovery period, they were exposed to 8% oxygen/nitrogen for 15 minutes, alternating with 100% oxygen for ten minutes for a total duration of 65 minutes in an incubated chamber kept at a constant ambient temperature of 34 ∘ C. Sixty hours after injury, pups were randomly assigned to one of the following treatment groups: (i) G-CSF (200 g/kg bw)/SCF (50 g/kg bw) in vehicle (1x PBS, phosphate-buffered saline), (ii) FL (100 g/kg bw) in vehicle, or (iii) vehicle only and received an intraperitoneal (i.p.) injection every 24 hours for three or five consecutive days, depending on timing of endpoint determination. Endpoints were assessed on P10, P18, and P90. Sex was visually determined by assessment of the anogenital distance and confirmed by polymerase chain reaction (P10, presence or absence of the SRY gene) or was determined by visual assessment alone (P18, P90) depending on mouse age as described previously [35]. Animals randomised to mid-(P18) and long-term endpoints (P90) received additional i.p. injections of BrdU solution (25 mg/kg bw) every 24 hours from P12 to P28. On P10, animals were sacrificed by decapitation; brains were harvested and immersionfixed in 4% formaldehyde for 72-120 hours. On P18 and P90, animals were deeply anaesthetised with phenobarbital (200 mg/kg i.v.) and intracardially perfused with 0.9% saline solution containing 1000 IU heparin per 100 ml. To ensure optimal fixation, the perfusion solution was switched to 4% paraformaldehyde after clearing of blood. Brains were harvested and immediately fixed in 4% paraformaldehyde for 24 hours. Following paraffin embedding, all brains were cut into 10-m-thick coronal sections. Sham-treated animals underwent anesthesia and surgical preparation of the right common carotid artery without ligation; healthy control animals did not undergo any form of treatment.
In addition, region-specific neuropathologic injury extent was determined in every section by two blinded observers according to a modified scoring system as follows: injury in cerebral cortex and adjacent white matter was scored 0-4 (0, no injury; 1, few small isolated groups of injured cells; 2, several larger groups of injured cells; 3, moderate confluent infarction; 4, extensive confluent infarction) and in hippocampus, striatum, and thalamus was scored 0-3 for mild, moderate, or severe atrophy or neuronal injury/infarction [29,36]. The average of the scores of the two investigators was used for statistical analysis.
. . Immunohistochemical Analyses. Immunohistochemical slides were pretreated as published [37]. In brief, paraffinembedded brain sections were deparaffinized and rehydrated through graded alcohols. Endogenous peroxidase activity was quenched by incubation with 2% hydrogen peroxide in methanol for 30 minutes. Heat-induced antigen retrieval was performed using citrate buffer. After nonspecific blocking with 1% bovine serum albumin (biotin-free) in TBS/0.05% Tween-20, sections were incubated overnight (4 ∘ C) with rabbit polyclonal anti-cleaved caspase-3 (Asp175) antibody No. PK-6100, Vector Laboratories) for 45 minutes at 25 ∘ C using diaminobenzidine (Invitrogen, Fisher Scientific Austria, Vienna, Austria) as a chromogen. After multiple rinses in distilled water, sections were dehydrated, cleared in xylene, and coverslipped. The Allen Mouse Brain Coronal Atlas (available from http://brain-map.org) was used as an anatomical reference for brain structures. Immunohistochemical quantification was conducted in different section planes depending on the respective regions of interest as follows: For counting of activated caspase-3 and AIF-positive nuclei, coronal levels 53 (Bregma 0.145 mm; striatum), 71 (Bregma -1.655 mm; thalamus), and 81 (Bregma -2.780 mm; hippocampus) were used as a reference. For BrdU, coronal levels 53 (Bregma 0.145 mm; subventricular zone, SVZ) and 81 (Bregma -2.780 mm; field CA1 of the hippocampus, CA1; granular layer of the hippocampal dentate gyrus, GrDG) were used as a reference. An Olympus IX83 microscope equipped with a DP27 digital color camera system and cellSens Dimension software (Olympus Austria, Vienna) was used to acquire images and count cells. ROIs were chosen ipsiand contralateral for cortical grey matter, white matter, hippocampus, striatum, and thalamus.
Manual cell counting for activated caspase 3 and AIFpositive nuclei was performed in four adjacent sections per slide by a blinded observer in grey matter, underlying white matter, hippocampus, striatum, and thalamus at the abovementioned stereotaxic coordinates at a 200x magnification within an area of 0.139 mm2 (one visual field). With regard to grey matter, cells were counted in cerebral cortex in 15 visual fields. In hippocampus, striatum, and thalamus, cells were counted in the entire displayed region. For quantification of BrdU-positive cells, the granular layer of dentate gyrus and CA1 region of the hippocampus were evaluated separately. BrdU-positive cells in the SVZ were counted in 3-4 cell diameters adjacent to the ependyma. Cell counts give absolute numbers of cells in the respective brain areas. For regions represented in more than one section plane (grey matter, white matter), mean values were calculated.
. . Behavioral Testing. Behavioral testing was conducted as described previously in detail [29]. At eleven weeks of age, animals were moved to a testing facility with a 12-hour lightdark cycle with lights on from 7 a.m. to 7 p.m. All tests were performed in accordance with circadian rhythms. Three different methods were used: (i) rotarod-accelerod to assess neuromotor performance, (ii) open field testing to observe general motor activity, exploratory behavior, and anxiety, and (iii) novel object recognition (NOR) to study learning and memory. Baseline behavioral parameters were assessed in healthy control animals for each task.
. . . Rotarod/Accelerod. For rotarod testing, a computeraided system (TSE Systems, Bad Homburg, Germany) with an accelerating rod setting (accelerod-acceleration from 5 to 60 rpm over a five-minute period) was used. Animals were trained three times a day with a test-free interval of one hour on five consecutive days. Testing was performed in five consecutive sessions on the sixth and seventh days. For statistical analysis, we compared the means of the time until task disruption (latency to fall off the rotarod apparatus) over all ten runs.

. . . Open Field.
For open field testing, animals were placed into the center of a dimly lit, opaque black box (50x50x40 cm). Animal movements were tracked by an automatic monitoring system (TSE Systems, Bad Homburg, Germany) for five minutes. Horizontal motor activity was evaluated by measuring travel distance. Anxiety was calculated as the percentage of time the animal stayed in the border areas of the box in relation to the total time spent in the maze (anxiety index, mean per animal). This procedure was repeated every 24 hours on four consecutive days in order to improve reliability and as training for the NOR [38].
Median value and interquartile range were calculated for each group. Vertical activity is given as mean rearing duration (i.e., standing on rear limbs) per treatment group ±SD. In addition to these variables, we recorded turning behavior with counter-clockwise movements and hyperactivity.
. . . Novel Object Recognition. Following open field, we performed NOR on consecutive days with a test-free interval of 48 hours. The NOR is designed as a two-trial, nonspatial, nonaversive memory test, consisting of a sample phase and a "choice" phase interrupted by a defined free interval (30 minutes, 5 hours, and 24 hours) during which the animals were returned to their cages. In the "choice" phase, one of the objects was replaced by a new one and the exploration was repeated for five minutes. They were placed at a distance (distance to centre) of 14 cm to the wall and 15 cm from each other. Novel object recognition was assessed by means of an automatic video tracking system (TSE Systems, Bad Homburg, Germany), using "three-point detection," which distinguishes between the head, rear end, and centre of gravity of the animal, allowing monitoring when the animal is exploring a specific object with its nose. An encounter was defined as sniffing or rearing on the object. The objects were designed in such a way that sitting on the object was impossible for the animals. Object recognition was defined as spending more time with the novel object, indicating ability to remember the familiar object. For better comparability in statistical analyses a discrimination index (DI) was calculated as follows: where t N is the time spent with the novel and t F is the time spent with the familiar object [39]. Positive values indicate a preference for the novel, and negative values indicate a preference for the familiar object.

Results
. . Study Population. Baseline histopathological injury was evaluated in sham-treated animals (P10, n=10; P18, n=8); baseline neurobehavioral outcome was established in healthy, untreated control animals (P90, n=5). For the evaluation of treatment effects in HHI, a total number of 149 animals were included in the study; 114 animals survived until respective endpoint analyses (male: n=61, female: n=53). Of the 35 deceased animals, one died during anesthesia, eight during or immediately after hypoxia, seven due to maternal cannibalism at any point, eight between postnatal days 6 (P6) and 10, eight between P10 and P18, and two were late deaths (>P18). One animal died during behavioral testing (brain not harvested). In animals who did receive treatment, occurrence ( 2 =0.163, p=0.922, Pearson Chi-Square, two degrees of freedom) and timing of death ( 2 =4.911, p=0.297, Pearson Chi-Square, four degrees of freedom) did not significantly differ between treatment groups. None of the animals reached humane endpoint criteria with the need to be euthanised.
. . Brain Injury Assessment. Baseline median gross brain injury in sham-treated animals was 0.00 (minimum 0.00, maximum 0.00) at all evaluated time points. In HHI, gross brain injury did not differ between treatment groups at any time point (P10: H=2.779, p=0.249, n=41; P18: H=2.256, p=0.324, n=39; P90: H=1.966, p=0.374, n=34; Kruskal-Wallis test, two degrees of freedom). Details can be found in Table 1, representative images of cresyl violet staining for gross brain injury assessment are shown in Figure 1. No sexspecific differences were observed (Mann-Whitney U test with Bonferroni correction for multiple tests; details are provided in Supplementary Table 1).
Baseline median neuropathological injury scores in sham-treated animals were 0.0 in all brain regions examined at all time points (all minimum 0.0, maximum 0.0; except for P10, white matter: maximum 0.5; P18, cerebral cortex: maximum 0.5).
In HHI, neuropathological injury scores did not significantly differ between treatment groups at any time point in cerebral cortex  Table 2.
No sex-specific differences were observed (Mann-Whitney U test with Bonferroni correction for multiple tests).
. . Caspase--Mediated Apoptosis. On P10, the number of activated caspase-3-positive cells did not significantly differ between treatment groups in any brain region in the ipsilateral hemisphere (grey matter: H=0.413, p=0.813; white    Table 3, and representative images of activated caspase-3 positive cells are shown in Figure 2. The number of activated caspase-3-positive cells in any of the regions examined was not significantly correlated with gross brain injury (Spearman rho correlation coefficient, all p>0.05, all data available on request). With regard to sex, no significant differences were detected between male and female experimental animals in all treatment groups (Mann-Whitney U test with Bonferroni correction for multiple tests; details are displayed in Supplementary Figure 1).    . . Caspase-Independent Apoptosis. On P10, the number of AIF-positive nuclei did not significantly differ between treatment groups in any brain region in the ipsilateral hemisphere ( Table 4, and representative images of AIFstaining are shown in Figure 3. No sex-specific differences were detected (Mann-Whitney U test with Bonferroni correction for multiple tests, all p>0.05).
. . Cell Proliferation. On P18, statistical analysis revealed an overall difference in the number of 5-bromo-2 -deoxyuridine-(BrdU-) positive cells in the granular layer of the hippocampal dentate gyrus in the ipsilateral hemisphere (H=6.038, p=0.049; n=17; Kruskal-Wallis test, two degrees of freedom). However, significance was lost in pairwise comparisons (1x PBS vs. G-CSF/SCF: p=0.050; 1x PBS vs. FL: p=0.738; Student-Newman-Keuls method). The number of BrdU-positive cells in other brain regions examined in the ipsilateral hemisphere did not significantly differ between Table 4: Region-specific apoptosis-inducing factor (AIF) positivity on postnatal day 10 (P10).

Region
Hemisphere  Tables 5 and 6, and data represent absolute cell numbers in the respective brain areas. Representative images of BrdU-positive cells are shown in Figure 4.
neurobehavioral endpoints with accelerod, open field, or novel object recognition testing is not possible in these animals, they were excluded from further behavioral analyses. With regard to accelerod performance, baseline time until task disruption in healthy control animals was 97.2 ± 46 sec (mean ± SD).
In   Figure 5: Open Field. Representative depiction of normal (a) and counter-clockwise movements and hyperactivity ("turning behavior") (b) in 90-day-old CD-1 mice subjected to hypoxic-hyperoxic ischemic brain injury on postnatal day 5 receiving daily treatment with 1x phosphatebuffered saline (PBS), granulocyte colony-stimulating factor (G-CSF)/stem cell factor (SCF), or Fms-related tyrosine kinase 3 ligand (FL) for five consecutive days starting 60 hours after injury. sex-specific differences were detected (n=5-7, Student's t-test with Bonferroni-adjusted level of significance).
In HHI, no overall differences in DI between treatment groups were observed after one (median DI (IQR); 1x PBS: . Details can be seen in Figure 6. No statistically significant sex-specific differences were detected (n=5-7, Student's t-test with Bonferroni-adjusted level of significance).

Discussion
The identification of novel treatment options for brain injury is one of the biggest challenges encountered in neonatal neuroscience [13,14,40,41]. Various cytokines and hematopoietic growth factors have been shown to have tissue-protective properties independent of their hematologic effects and have been investigated as therapeutic strategies in neonatal brain injury [42]. Several of them have yielded promising results in preclinical trials. However, due to the pleiotropic nature of many of these agents and their contextspecific mechanisms of action, meticulous assessment of their safety profiles is essential prior to potential routine clinical use [5,12]. Our research group has previously shown that the hematopoietic growth factors G-CSF/SCF and FL can exert both beneficial and injurious effects on neonatal brain injury, depending on type of injury and timing of treatment [29][30][31]. To the best of our knowledge their short-, mid-, and long-term effects in multiple-stressor neonatal brain injury models have not yet been systematically evaluated. : Long-term behavioral assessment following hypoxichyperoxic ischemic brain injury in CD-1 mice on postnatal day 90 by means of novel object recognition testing. Experiments were performed on three consecutive days. Mice were first placed in a dimly lit, opaque black box containing two identical objects for five minutes and were then returned to their cages. After a defined interval (one hour (1 h) on day 1, three hours (3 h) on day 2, and five hours (5 h) on day 3) they were put back in the apparatus, which now contained one familiar and one novel object. The time spent at each object was recorded and a discrimination index (DI) was calculated as follows: DI=((t N -t F ))/((t N +t F )), where t N is the time spent with the novel and t F is the time spent with the familiar object. In the study at hand, we used a combined hypoxichyperoxic ischemic brain injury model, thought to more adequately reflect the clinical situation in neonatal intensive care units [32,34]. Also, in this model, hypoxia-related mortality is regarded to be less pronounced, while brain injury can still be reliably induced [34]. In our study, the mortality rate was 23.5% and thus higher than previously described [34]. However, this finding did not reach statistical significance (p=0.090, Fisher's Exact test). Of relevance, mortality was not influenced by treatment type.
On a histomorphological basis, we did not observe statistically significant differences between treatment groups at any time point, either in gross brain injury, or in caspasedependent or -independent apoptosis as indicated by AIFpositivity. Another mechanism of action described for G-CSF/SCF is its trophic effect on neurons [28]. An increase in cell proliferation could be mediated by an enhanced release of CD34+ cells due to G-CSF/SCF application [30], which has been shown to increase angio-and neurogenesis and to improve morphological and functional recovery [43]. For this reason, we investigated whether G-CSF/SCF and FL treatment increases cell proliferation and the number of BrdU-positive vessels at the lesion site but did not observe a statistically significant difference. This is in discordance with previous neonatal rodent brain injury studies, reporting reduced damage extent, apoptotic cell death, and increased cell proliferation following both single and repetitive G-CSF treatments [18,19,21,23,24,28]. Also, in our study functional outcomes did not significantly differ between treatment groups-in contrast to previously published works reporting improvement in long-term cognitive and motor functions after G-CSF and SCF treatment [20,22,25,27,28]. However, all aforementioned experiments were conducted in newborn rat pups. As species and strain differences in brain injury susceptibility have been reported, a differential response to treatments in mice is entirely possible [44,45]. Of relevance, our own research group has shown no improvement in brain injury extent or neurobehavioral outcome following G-CSF treatment in newborn mice subjected to a modified Rice-Vannucci procedure [29]. In general, treatment effects might depend on type of injury, as all aforementioned studies used hypoxia-ischemia [18][19][20][21][22][23][24][25]29] or hypoxia only [27,28] for injury induction. In a mouse model of neonatal excitotoxic brain injury, we previously observed an exacerbation of grey and white matter damage by G-CSF/SCF when administered in the acute phase of injury [30]. When administered 60 hours after injury, however, G-CSF/SCF and FL treatments led to a significant reduction in lesion size and a decrease in caspase-3 activation in the same injury model [31]. The very same treatment regimina were applied in the current study. As we have shown that the administration of substances or even vehicle solutions in the acute phase of injury might lead to increased brain damage in newborn mice [29,30], an acute treatment strategy was not attempted again for animal-sparing reasons. This is also one of the main limitations of our study: We cannot rule out that the therapeutic window in combined hypoxic-hyperoxic ischemic neonatal brain injury might be narrow and an earlier intervention might be beneficial. However, bearing in mind our previous findings of an increased vulnerability in the acute phase of injury and the principles of the 3Rs in animal experimentation, a de novo assessment of acute treatment strategies in this animal model did not seem justifiable.
For the present study, a higher dose of G-CSF was selected. The dose of 200 g/g body weight was chosen, based on previous publications and our own studies, showing a dose response curve [29,30]. The effects of G-CSF dosage might be important, as Taguchi et al. reported impairments in behavior with increasing dosages of G-CSF following cerebral ischemia [43] The difference in the results might reflect the experimental model used, age and rodent used, the therapeutic time window, and the dose administered. As to why the treatment options conveying neuroprotection in the excitotoxic injury model [31] had no effect on hypoxic-hyperoxic ischemic brain injury, we can only speculate that the pathomechanisms leading to cell death following hypoxia-hyperoxia ischemia might be more complex, or injury extent more pronounced due to systemic hypoxia and hyperoxia and thus less susceptible to rescue treatments in general. Pathomechanisms involved in hypoxic-hyperoxic neonatal brain injury include alterations in apoptotic, inflammatory, and angiogenetic pathways, all of which are also implicated in hypoxic-ischemic brain injury [34] A direct comparison of these two injury models regarding pathogenesis and, particularly, contribution of individual factors would be of high relevance but is outstanding to date. Further studies evaluating this aspect in depth are required.
Interestingly, turning behavior as an indicator of severe brain damage was more frequently observed in G-CSF/SCFand FL-treated animals than in vehicle-treated controls. Even though the finding was not statistically significant, this might still be of relevance as turning behavior has been shown to correlate with histological damage and sensory motor deficit in rodents exposed to cerebral ischemia [46]. Cautious use of these substances might thus be warranted.
The importance of evaluating sex-specific differences has been proven by several studies, especially when considering the variable responses of male and female immature rodents to brain injury [36,47,48]. Hence, we also conducted a sex-specific evaluation, but no sex-specific differences were present.

Conclusion
To conclude, in this multistressor mouse model of hypoxiahyperoxia ischemia we cannot confirm the protective effect of G-CSF/SCF and FL on neonatal brain injury previously reported. As adverse effects cannot be ruled out, meticulous assessment of these hematopoietic growth factors in extensive preclinical studies is mandated prior to potential clinical use.

Data Availability
All data generated or analysed during this study are included in this published article (and its supplementary information files).

Disclosure
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Conflicts of Interest
The authors declare that they have no conflicts of interest.