Neonatal erythropoietin mitigates impaired gait, social interaction and diffusion tensor imaging abnormalities in a rat model of prenatal brain injury

Abstract

Children who are born preterm are at risk for encephalopathy of prematurity, a leading cause of cerebral palsy, cognitive delay and behavioral disorders. Current interventions are limited and none have been shown to reverse cognitive and behavioral impairments, a primary determinant of poor quality of life for these children. Moreover, the mechanisms of perinatal brain injury that result in functional deficits and imaging abnormalities in the mature brain are poorly defined, limiting the potential to target interventions to those who may benefit most. To determine whether impairments are reversible after a prenatal insult, we investigated a spectrum of functional deficits and diffusion tensor imaging (DTI) abnormalities in young adult animals. We hypothesized that prenatal transient systemic hypoxia-ischemia (TSHI) would induce multiple functional deficits concomitant with reduced microstructural white and gray matter integrity, and tested whether these abnormalities could be ameliorated using postnatal erythropoietin (EPO), an emerging neurorestorative intervention. On embryonic day 18 uterine arteries were transiently occluded for 60 minutes via laparotomy. Shams underwent anesthesia and laparotomy for 60 min. Pups were born and TSHI pups were randomized to receive EPO or vehicle via intraperitoneal injection on postnatal days 1 to 5. Gait, social interaction, olfaction and open field testing was performed from postnatal day 25–35 before brains underwent ex vivo DTI to measure fractional anisotropy, axial diffusivity and radial diffusivity. Prenatal TSHI injury causes hyperactivity, impaired gait and poor social interaction in young adult rats that mimic the spectrum of deficits observed in children born preterm. Collectively, these data show for the first time in a model of encephalopathy of prematurity that postnatal EPO treatment mitigates impairments in social interaction, in addition to gait deficits. EPO also normalizes TSHI-induced microstructural abnormalities in fractional anisotropy and radial diffusivity in multiple regions, consistent with improved structural integrity and recovery of myelination. Taken together, these results show behavioral and memory deficits from perinatal brain injury are reversible. Furthermore, resolution of DTI abnormalities may predict responsiveness to emerging interventions, and serve as a biomarker of CNS injury and recovery.

INTRODUCTION

Globally, almost 15 million infants are born preterm each year (Blencowe et al. 2012), making preterm birth a leading cause of neurological and neuropsychiatric disability. Children born extremely preterm (<28 weeks) are particularly at risk, and prone to a spectrum of deficits including cerebral palsy, epilepsy, hyperactivity and inattention, and impaired social interaction, memory and executive function (Anderson 2014). Affecting up to 50% of survivors (Nosarti et al. 2012, Anderson 2014), cognitive and behavioral deficits are the most common, often the most debilitating, and present the greatest hurdle to adult independence. As survival from preterm birth into adulthood improves, the need to optimize neuropsychiatric outcomes for preterm survivors is increasingly apparent (Nosarti, Reichenberg et al. 2012). Further, a lack of resource-efficient preclinical models that adequately replicate neuropsychiatric impairments in children born preterm is a major barrier to translation of potential therapies.

The neurological deficits from preterm birth arise from a complex amalgam of developing central nervous system (CNS) abnormalities that involve white matter (WM), subplate and gray matter (GM) injuries (Robinson 2005, Kostovic and Judas 2006, Kinney 2009, Volpe 2009, Kostovic et al. 2014). Despite the shift observed from cystic periventricular leukomalacia to diffuse WM gliosis on imaging in preterm survivors at major academic medical centers (Hamrick et al. 2004, Rutherford et al. 2010, de Vries and Volpe 2013), a concomitant improvement in neuropsychiatric outcomes has not occurred. Collectively, these data emphasize that cognitive and behavioral difficulties likely reflect complex, multiregional injury extending beyond white matter alone (Kostovic and Judas 2006, Robinson et al. 2006, Ligam et al. 2009, Volpe 2009, Kostovic et al. 2014, Ceschin et al. 2015).

For many preterm infants, CNS injury begins in utero with maternal-placental-fetal axis dysfunction (Leviton et al. 2010, Dammann and Leviton 2014, Shevell et al. 2014, Johnson and Marlow 2016). To facilitate translation to clinical scenarios, we developed a model of CNS injury from prematurity that: 1) replicates the intrauterine insult and capitalizes on the hemichorial, discoid placental unit in rats that is similar to humans (Jantzie and Robinson 2015); 2) recapitulates the neuronal, oligodendroglial and subplate developmental injury (Robinson et al. 2005, Mazur et al. 2010, Jantzie et al. 2013, Jantzie et al. 2015, Jantzie et al. 2016); and as shown here produces a sustained spectrum of functional deficits in mature animals. We hypothesized that this model would be an effective platform to study and support testing of neuro-repair strategies for CNS injury from preterm birth.

An example of such an intervention is erythropoietin (EPO) (McPherson and Juul 2010, Wu et al. 2012, Jantzie, Miller et al. 2013, Jantzie et al. 2014, Leuchter et al. 2014, Ohls et al. 2014, Fauchere et al. 2015, Jantzie, Corbett et al. 2015, O’Gorman et al. 2015, Jantzie, Winer et al. 2016). Without ligand present, unbound EPO receptors trigger cell death (Knabe et al. 2004, Knabe et al. 2005, Mazur, Miller et al. 2010, Ott et al. 2015). After perinatal brain injury (PBI), neural cell EPOR expression increases without concomitant EPO ligand expression, and exogenous EPO restores balanced EPOR signaling (Spandou et al. 2004, Mazur, Miller et al. 2010, Ott, Martens et al. 2015). Exogenous EPO crosses the blood-brain barrier to enhance recovery after CNS injury (Brines et al. 2000, Xenocostas et al. 2005, Mazur, Miller et al. 2010, Gonzalez et al. 2013, Jantzie, Miller et al. 2013). Thus, to improve the evaluation of emerging interventions for deficits from preterm birth, we used prenatal injury and neonatal EPO to test whether changes in mature, sophisticated outcomes, including social interaction could be detected in a rat model. Indeed, we show using translatable diffusion tensor imaging (DTI) metrics and behavioral assessment that postnatal EPO treatment restores impaired complex gait deficits and social interaction, and normalizes DTI abnormalities and aberrant microstructure in white matter and deep gray matter that are associated with CNS injury from preterm birth.

METHODS

All procedures were performed in accordance with NIH Guide for Care and Use of Laboratory Animals with approval by Institutional Animal Care and Use Committees at University of New Mexico Health Sciences Center and Boston Children’s Hospital. Rats were housed in standard colony rooms with lighting on from 7 am to 7 pm, and food and water available ad libitum.

Transient Systemic Hypoxia-Ischemia

Embryonic day (E) 18 pregnant Sprague-Dawley dams underwent isoflurane anesthesia and laparotomy, as previously described (Robinson, Petelenz et al. 2005, Mazur, Miller et al. 2010). Like humans, rodent oligodendrogenesis begins prenatally, and differentiation and myelination continue postnatally. In rats, pioneering subplate neurons exit the thalamus by embryonic day 16 (E18) and arrive in the cortex by postnatal day 2 (P2) (Kanold and Luhmann 2010). For prenatal transient systemic hypoxia-ischemia (TSHI), uterine arteries were transiently occluded for 60 minutes, while shams underwent laparotomy only for 60 min. This approach accurately recapitulates TSHI injury through intact maternal-placental-fetal units, and capitalizes on changes individual fetal microenvironments. Pups were born at term (E22) and reared with dams until postnatal day 21 (P21), when they were weaned and housed in single sex groups (2–3 per cage). Prenatal TSHI produces a reproducible insult with consistent fetal mortality and postnatal growth and neuropathological, biochemical and functional outcomes (Robinson, Petelenz et al. 2005, Mazur, Miller et al. 2010, Jantzie, Miller et al. 2013, Jantzie et al. 2014, Jantzie, Getsy et al. 2014, Jantzie, Corbett et al. 2015, Jantzie, Winer et al. 2016). For all experiments, both sexes, and rats generated from at least two separate litters were used. The average litter size for all experiments were similar, with sham litters an average of 10.4 ± 0.5 pups and TSHI litters at 11.6 ± 0.6 pups (p>0.05). The average pup size also did not differ between groups with sham weighing 7.3 ± 0.05g on P1, TSHI pups weighing 7.1 ± 0.1g and TSHI+EPO pups weighing 7.2±0.1g at P1, consistent with our previously published reports of no significant differences between Sham and TSHI litters or pups, and the cumulative effects of in utero TSHI on litter size, postnatal survival and offspring body weight (Jantzie, Corbett et al. 2014).

Postnatal EPO treatment

Using an established, clinically-relevant dosing regimen (Mazur, Miller et al. 2010, Jantzie, Getsy et al. 2014, Jantzie, Corbett et al. 2015, Jantzie, Winer et al. 2016), P1 pups from all litters were individually randomized to receive either EPO (2000 U/kg, R&D Systems, Minneapolis, MN) or vehicle (sterile saline) intraperitoneally once daily from P1 to P5, a dosing regimen comparable to those used in human neonatal neuro-reparative trials. Rodents are born at a time equivalent to the human third trimester, with P9 approximately equivalent to term in human gestation (Jantzie and Robinson 2015). Thus, EPO or vehicle administration at P1 to P5 is approximately equivalent to 30 to 35 weeks gestation in humans. Previous studies showed this dosing regimen improved stride length and seizure threshold, and demonstrated no unexpected findings with EPO-treated sham animals (Mazur, Miller et al. 2010).

Behavioral Testing

Three rat cohorts (each cohort consisting of a sham and a prenatal TSHI litter) were sequentially tested for gait, and open field (OF) testing at P25-30. Specifically, Gait was tested at P25-P26, and then Open field was tested at P28-30. Two separate cohorts (4 litters) underwent social interaction testing at P30-P32.

Gait Analysis

Computerized gait analysis was performed on P25-P26 (Jantzie, Corbett et al. 2014, Jantzie, Corbett et al. 2015). Briefly, digital video of each rat running on a backlit transparent treadmill set at 30 cm/s was acquired with a high-speed camera and analyzed using Digigait software (Mouse Specifics, Framingham, MA). Digigait software analyses identifies individual paw prints and allows calculation of multiple gait metrics and kinematic measurements based on the position, area and timing of each paw step. Similarly, posture, cadence, stance duration, well as braking, swing and propulsion phases of each step are measured. Propulsion was defined as the maximum paw contact on belt to just before the swing phase (push off and paw acceleration). Ataxia coefficients were calculated as the deviation of the minimum and maximum stride length from the mean stride length (Puram et al. 2011). As TSHI induces a global injury, data from right and left limbs were combined. Forelimb and hindlimb gait patterns were analyzed separately.

Open Field

A circular open field arena (100 cm diameter) was placed in a quiet, well-lit room (130 lumens), and was marked to divide the arena into three equally-spaced, concentric circles labeled the center, neutral and peripheral zones. At P28-P30, each rat was initially placed against the wall of the testing arena, and allowed to explore for 15 minutes. Anymaze^™ video-tracking software measured the distance traveled, velocity, and time spent in each zone, and entries from one zone to the adjacent zone.

Social Interaction and Olfaction

A standard paradigm was used to identify impaired social interaction in rats at P30-P32 (Schneider and Przewlocki 2005, Foley et al. 2014, Zugno et al. 2016). Briefly, one hour prior to testing, each rat was isolated in a clean cage. For social interaction testing, two rats of the same sex and treatment group, but from different litters, were placed in a dimly lit (30 lumens) circular testing arena (100 cm diameter) and recorded for 10 min using Anymaze^™ video-tracking software. Each pair was counted as one social unit. Two observers blinded to the treatment group independently reviewed the trials and scored periods of social interaction (trailing, sniffing, grooming, playing, etc.). The inter-rater reliability was 0.94. Following assessment of social interactions, intact olfaction of each rat was confirmed by exposure to sequentially to scented-cotton swabs (water, adult male cage swab, and food odor) affixed to the top of a clean cage for 1 minute per swab. An observer blinded to treatment group recorded the time that the rat sniffed within 2 cm of the swab consistent with prior reports (Schneider and Przewlocki 2005, Zugno, Canever et al. 2016).

Ex vivo Magnetic Resonance Imaging

Ex Vivo MRI was performed on a Bruker 4.7 T BioSpec 47/40 Ultra-Shielded Refrigerated nuclear magnetic resonance imaging system, as previously described (Jantzie et al. 2015, Robinson et al. 2016, Robinson et al. 2016). The MRI was equipped with a 72 mm I.D. quadrature RF coil and a small-bore (12 cm I.D.) gradient set with a maximum gradient strength of 50 Gauss/cm. Briefly, at P35-40 rats were deeply anesthetized with sodium pentobarbital and perfused with 4% paraformaldehyde. Brains were removed from the skull and post-fixed in 4% paraformaldehyde for 1 week, and embedded in 2% agarose containing 3 mM sodium azide for immediate ex vivo MR imaging. Fixation adequately preserves brain microstructures and allows the generation of MR images of superior contrast and resolution, which is especially beneficial for evaluating white matter fibers and for assessing the brain’s microstructural integrity (Robinson, Berglass et al. 2016, Robinson, Winer et al. 2016). T2 multi-slice multi-echo (MSME) structural images were obtained with a TR of 2514.7 ms and a TE of 12 ms. Echo-planar diffusion tensor imaging (EP-DTI) of twelve contiguous coronal 1 mm slices were obtained with a FOV (field-of-view) of 3.00 cm, a TR of 3000 ms, TE of 40 ms, MTX of 256, and b-value of 2000 mm²/s with 30 gradient directions. Brain regions of interest (ROI) including the sensory barrel cortex, hippocampus, striatum, thalamus, and white matter (Suppl Fig 3) were analyzed using Bruker’s Paravision 5.1 imaging software. Fractional anisotropy (FA), axial (λ1) and radial ((λ2+λ3)/2) diffusivity eigenvectors were measured and calculated. Directionally encoded color maps were created and are presented as a composite image of primary eigenvectors overlaying T2 morphology. The morphology α (0.5) and eigenvector α (1.0) were equal for all images, with an overall composition α of 0.4 ×10⁴.

Immunohistochemistry

Frozen coronal sections (20μm) were cut on a cryostat, consistent with prior published methodology (Jantzie, Corbett et al. 2014, Jantzie, Corbett et al. 2015). Briefly, sections were washed and then incubated sequentially with 0.3% hydrogen peroxide and blocking solution containing 10% goat serum in phosphate-buffered saline. Primary antibodies against neurofilament (NF, SMI-312, Covance, San Diego, CA, 1:500) or CHOP ((C/EBP homologous protein, 1:250, Cell Signaling, Danvers, MA) in blocking solution containing 0.1 % Triton were then applied and incubated overnight at 4°C. The next morning, sections were rinsed, incubated with species appropriate biotinylated secondary antibodies, followed by Vectastain (Vector Labs, Burlingame, CA) and 3′3′-diaminobenzidine (DAB). Sections were then dehydrated in a grades series of alcohols, cleared in xylenes and coverslipped with Permount (Millipore Sigma, St. Louis, MO). Appropriate negative controls (no primary antibody) were run in parallel and as per the procedure described above. Using bright-field illumination, representative images were photographed on a Leica microscope.

Black Gold Staining

Myelin was stained using a Black Gold-II kit per manufacturers’ specifications, with slight modifications (Millipore Sigma, St. Louis, MO). Briefly, 20 μm slide mounted sections were rinsed with MilliQ water and incubated with a pre-warmed 0.3% Black Gold Solution for 20 minutes at 60°C. After rinsing, sections were then incubated with 1% sodium thiosulfate for 1.5 minutes at 60°C. Sections were then washed, counterstained with cresyl violet, rinsed and dehydrated using a graded series of ethanols, and coverslipped with Permount. With brightfield illumination, representative images were photographed on a Leica microscope.

Multiplex Electrochemiluminescent Immunnoassay (MECI)

Using an electrochemiluminescent immunoassay platform (MesoScale Discovery, Gaithersburg, MD), brain levels of CHOP, a transcription factor and marker of endoplasmic reticulum (ER) stress, were assayed in P5 sham, TSHI, and TSHI+EPO rat pups. Briefly, brain lysate (200 μg) was loaded onto a 96-well plate in duplicate per manufacturer’s specifications, and consistent with our prior publications (Yellowhair et al. 2017). Plates were read on a Quickplex SQ 120 Imager (MesoScale Discovery, Gaithersburg, MD).

Statistical Analysis

For all analyses, data are represented as mean ± standard error of the mean (SEM), with p<0.05 considered statistically significant. For analysis of sham, TSHI-veh and TSHI-EPO groups for gait, open field, social interaction and MRI, all parametric variables were tested for normal distribution with the Shapiro-Wilk test, with Levene’s test to confirm homogeneity of variances. Two-way ANOVA was performed with Bonferroni’s post-hoc correction for multiple comparisons using SPSS21 (IBM, Armonk, NY). For functional assays, both sexes were analyzed together, and then each sex was analyzed separately. Due to resource limitations, and pilot analyses showing sex had no impact on MRI parameters (Jantzie, Getsy et al. 2015, Robinson, Berglass et al. 2016, Robinson, Winer et al. 2016), separate analyses by sex were not performed for the DTI studies. Correlative relationships between DTI parameters and gait metrics were performed using a bivariate Pearson’s correlation for continuous variables. Since the dependent measures were not collected in a within subjects design, adult Sham, TSHI and TSHI+EPO rats that underwent gait analysis were sex matched, and randomly compared to an independent sample of adult rats undergoing DTI analysis. A Pearson’s product-moment correlation was then run to assess the relationship between individual DTI (i.e. FA, RD) and gait metrics (i.e. ataxia coefficient, stride length variation, paw area).

RESULTS

Gait disturbance from prenatal TSHI resolves with post-injury EPO treatment

Detailed digital gait pattern analyses revealed TSHI-veh rats (n=30) had a markedly abnormal gait compared to shams (n=26) that improved with postnatal EPO (TSHI-EPO, n=22, Fig. 1). Specifically, TSHI-veh rats spent less time in hindlimb swing or propel phases than shams or TSHI-EPO rats (all p<0.001, Fig. 1A,B), consistent with a shorter stride length and increased stepping frequency (Jantzie, Corbett et al. 2015). TSHI-veh rats showed more variation in hindlimb stride length, swing duration and ataxia (all p<0.001, Fig. 1C–E). Interestingly, TSHI-veh rats contacted less hindlimb paw area at peak stance, and used less maximal force (all p<0.001, Fig. 1F,G), consistent with toe-walking and similar to the spastic gait in children with cerebral palsy. Notably, forelimb gait metrics were less affected by TSHI compared to hindlimbs, similar to children born preterm with cerebral palsy and complex motor impairment (Table 1). Nevertheless, compared to shams or TSHI-EPO rats, forelimb stride time and length were both shorter in the TSHI-veh rats (all p<0.001, Fig. 1H,I).

Figure 1.

Prenatal TSHI induces gait deficits that are mitigated by post-injury EPO treatment. Hindlimb swing phase (A) and propel phase (B) are significantly shorter following TSHI in vehicle-treated rats (dark gray), compared to sham (light gray) and TSHI-EPO treated (hashed) rats. Hindlimb stride length variation (C) and ataxia (D) are much greater in TSHI-veh rats, compared to shams and TSHI-EPO rats. The hindlimb paw area at peak stance (E), and maximal pressure exerted by the hindlimb at peak stance (F) are both less in TSHI-veh rats, compared to shams and TSHI-EPO rats. Forelimb stride time (G) and stride length (H) are also shorter in TSHI-veh rats compared to shams and TSHI-EPO rats (two-way ANOVA with Bonferroni’s post-hoc correction, ***p ≤ 0.001)

Table 1.

Two-way ANOVA with Bonferroni post-hoc correction,

Male	Hindlimb			Forelimb
Gait Metric	Sham	TSHI-veh	TSHI-EPO	Sham	TSHI-veh	TSHI-EPO
Swing Time (s)	0.097 ± 0.004	0.073 ± 0.004c	0.096 ± 0.004c	0.010 ± 0.005	0.086 ± 0.004	0.102 ± 0.17a
Propel Time (s)	0.167 ± 0.007	0.128 ± 0.006c	0.165 ± 0.008b	0.100 ± 0.007	0.083 ± 0.005	0.091 ± 0.006
Brake Time (s)	0.033 ± 0.003	0.027 ± 0.002	0.033 ± 0.003	0.083 ± 0.007	0.060 ± 0.004a	0.074 ± 0.006
Stance Time (s)	0.201 ± 0.007	0.155 ± 0.006c	0.197 ± 0.007c	0.183 ± 0.008	0.143 ± 0.007c	0.165 ± 0.007
Stride Time (s)	0.30 ± 0.01	0.23 ± 0.01c	0.29 ± 0.11c	0.28 ± 0.01	0.23 ± 0.01b	0.27 ± 0.01a
Stride Length (cm)	8.9 ± 0.3	6.9 ± 0.3c	8.8 ± 0.3c	8.5 ± 0.3	6.9 ± 0.3b	8.0 ± 0.3a
Step Frequency	3.5 ± 0.2	4.8 ± 0.2c	3.6 ± 0.2c	3.7 ± 0.2	4.8 ± 0.3b	4.0 ± 0.2
Stride Length CV	19 ± 3	41 ± 3c	20 ± 3c	35 ± 7	41 ± 3	37 ± 3
Swing Duration CV	27 ± 4	52 ± 3c	27 ± 4c	36 ± 4	50 ± 3a	39 ± 4
Ataxia Coefficient	0.67 ± 0.11	1.33 ± 0.09c	0.64 ± 0.11c	1.10 ± 0.12	1.38 ± 0.11	1.29 ± 0.12
Paw Area at Peak Stance (cm2)	1.29 ± 0.08a	1.21 ± 0.07	1.64 ± 0.09c	0.60 ± 0.03	0.61 ± 0.03	0.66 ± 0.04
Maximal Force (cm2/s)	98.4 ± 7.5b	96.3 ± 6.7	133.7 ± 7.9b	41.0 ± 1.6	43.6 ± 1.8	46.0 ± 1.9

Female	Hindlimb			Forelimb
Gait Metric	Sham	TSHI-veh	TSHI-EPO	Sham	TSHI-veh	TSHI-EPO
Swing Time (s)	0.098 ± 0.004	0.079 ± 0.004b	0.101 ± 0.005b	0.102 ± 0.003	0.086 ± 0.003b	0.105 ± 0.004b
Propel Time (s)	0.155 ± 0.005b	0.138 ± 0.005a	0.178 ± 0.006c	0.094 ± 0.004	0.068 ± 0.004c	0.091 ± 0.005b
Brake Time (s)	0.033 ± 0.022b	0.027 ± 0.002a	0.025 ± 0.002	0.076 ± 0.004	0.061 ± 0.004a	0.078 ± 0.005a
Stance Time (s)	0.188 ± 0.005	0.165 ± 0.005b	0.203 ± 0.005c	0.169 ± 0.006	0.129 ± 0.006c	0.169 ± 0.007c
Stride Time (s)	0.287 ± 0.007	0.244 ± 0.007c	0.304 ± 0.008c	0.271 ± 0.009	0.215 ± 0.009c	0.275 ± 0.010c
Stride Length (cm)	8.6 ± 0.2	7.3 ± 0.2c	9.1 ± 0.3c	8.1 ± 0.3	6.5 ± 0.3c	8.2 ± 0.3c
Step Frequency	3.7 ± 0.1	4.4 ± 0.1b	3.4 ± 0.2c	3.924 ± 0.173	5.041 ± 0.178c	3.796 ± 0.206c
Stride Length CV	21 ± 2	39 ± 2c	16 ± 3c	29 ± 3	48 ± 3c	31 ± 3c
Swing Duration CV	26 ± 3	46 ± 3c	19 ± 3c	32 ± 3	57 ± 3c	33 ± 4c
Ataxia Coefficient	0.68 ± 0.07	1.30 ± 0.08c	0.55 ± 0.09c	1.01 ± 0.09	1.64 ± 0.09c	1.09 ± 0.11c
Paw Area at Peak Stance (cm2)	1.53 ± 0.06	1.23 ± 0.06b	1.72 ± 0.07c	0.69 ± 0.03	0.53 ± 0.03c	0.70 ± 0.03c
Maximal Force (cm2/s)	124.1 ± 5.2	100.5 ± 5.4b	140.2 ± 6.2c	45.3 ± 1.5	36.4 ± 1.2c	47.7 ± 1.8c

^ap<0.05;

^bp<0.01;

^cp<0.001.

Letter in TSHI-veh column indicates statistical comparison between sham and TSHI-veh. Letter in TSHI-EPO column indicates statistical comparison between TSHI-veh and TSHI-EPO. Letter in sham column indicates statistical comparison with TSHI-EPO. Female TSHI-veh rats (n=16), female shams (n=17) and female TSHI-EPO rats (n=12). Male TSHI-veh rats (n=14), male shams (n=11) and male TSHI-EPO rats (n=10, Table 1).

Overall, sex differences in gait in TSHI-veh rats were subtle. Female TSHI-veh rats (n=16), shams (n=17) and TSHI-EPO rats (n=12) demonstrated a pattern of deficits similar to all animals independent of sex (Table 1). Female TSHI-veh rats used less paw area and maximal force in both hindlimbs and forelimbs compared to male animals, and compared to female sham and female TSHI-EPO rats. Male TSHI-veh rats (n=14) also showed hindlimb deficits similar to all animals independent of sex (male shams (n=11) or male TSHI-EPO rats (n=10, Table 1)). In sum, detailed gait analyses show TSHI-veh rats exhibit a sustained motor deficit phenotypically similar to children with a spastic gait from preterm birth, including a shorter, more variable stride, and less paw contact area consistent with toe-walking and ataxia. Importantly, these findings indicate that postnatal EPO treatment mitigates motor deficits in both sexes.

Impaired social interaction resolves with post-injury EPO treatment

To quantify social interaction, pairs of rats with the same sex, injury and treatment group, but from different litters, were observed and interactions were scored. Sham (10 pairs, n=20) and TSHI-EPO (8 pairs, n=16) rats spent significantly more time interacting, including sniffing, playing and grooming, than TSHI-veh rats (10 pairs, n=20, both p=0.001, Fig. 2A). Mean duration of each interaction period was also shorter for TSHI-veh rats (both p<0.001, Fig. 2B). When analyzed separately by sex, both male and female rats showed a similar pattern, with EPO treatment ameliorating deficits in rat social behavior in both sexes (Fig. 2C–F). Olfactory testing confirmed primary sensory deficits were not related to the impaired social interaction observed in the TSHI animals (Suppl. 1), and that all rats were able to distinguish social and food odors. These findings show in utero TSHI causes impaired social interaction in young adult rats of both sexes, and that the social interaction deficits can be restored by postnatal EPO treatment.

Figure 2.

Postnatal EPO treatment mitigates deficits in social interaction following prenatal TSHI. The total time of interaction (A) and mean duration of each period of interaction (D) was shorter in pairs of TSHI-veh rats (n=20), compared to pairs of sham (n=20) or TSHI-EPO treated rats (n=8). When analyzed by sex, the total time of interaction for male (B) and female (C) rats, and mean duration of each period of interaction for male (E) and female (F) rats showed the same pattern. (two-way ANOVA with Bonferroni’s post-hoc correction, *p < 0.05, **p ≤ 0.01, ***p ≤ 0.001)

Hyperactivity persists after post-injury EPO treatment

Assessment of behavior in a circular open field revealed TSHI-veh rats (n=31) traveled faster and a greater distance compared to shams (n=29, both p=0.001, Fig. 3A,B), consistent with hyperactivity and increased mobility. Notably, TSHI-veh rats maintained the excess activity throughout the 15 minute observation period, whereas sham and TSHI-EPO rats showed diminished activity as they spent more time in the arena (p=0.006, Fig. 3C). Rats naturally avoid an exposed, open area. TSHI-veh rats, however, showed a lack of environmental awareness by crossing into, spending more time in, and traveling more in the arena center compared to shams (all p=0.001, Fig. 3D–F). TSHI-veh rats also crossed into the perimeter region more than sham rats (p=0.001), consistent with their hyperactivity. Compared to shams, TSHI-veh rats exhibited a higher proportion of center/total entries (p=0.02, Fig. 3G). Similarly, the proportion of center/total distance traveled, and center/total mobile time was longer for TSHI-veh rats, compared to shams (both p=0.001, Fig. 3H,I). Unlike the recovery with EPO treatment that was observed with gait and social interaction, TSHI-veh and TSHI-EPO treated rats performed the same on all OF parameters and as visualized on composite occupancy heat maps (Fig. 3 and Suppl Fig 2). Similarly, both male and female TSHI rats demonstrated the same patterns of hyperactivity compared to shams, and lack of improvement with EPO (Suppl Table 1).

Figure 3.

Prenatal TSHI induces hyperactivity and disinhibition. In an open field arena, TSHI-veh and TSHI-EPO rats have increased mean speed (A), increased total distance traveled (B), and increased total distance travelled in the final 5 minutes (C) of open field arena exploration, compared to sham animals. Similarly, TSHI-veh and TSHI-EPO rats had more center entries (D), spent more time in the center of the field (E) and traveled more distance in the center (F) in the open field, compared to shams. TSHI-veh and TSHI-EPO rats also had a greater percentage of center entries (G), center distance (H) and total time in the center (I). (two-way ANOVA with Bonferroni’s post-hoc correction,*p < 0.05, **p ≤ 0.01, ***p ≤ 0.001)

MRI microstructural abnormalities improve with post-injury EPO treatment

Diffusion MRI analyses were used to establish differences in structure following in utero injury, including changes in fractional anisotropy (FA), axial diffusivity (AD) and radial diffusivity (RD) (Fig. 4). ROI studied are shown in Supplemental Figure 3. Specifically, FA in the capsular white matter/corona radiata white matter was reduced in TSHI-veh brains (n=8) compared to shams (n=10, p<0.001). Notably, this injury-induced change in fractional anisotropy was resolved following EPO treatment, consistent with improved microstructure and tract organization (n=9, p=0.001, Fig. 5A). Mean diffusivity (MD) for each region analyzed is shown in Supplemental Figure 3. In callosal and capsular white matter/coronal radiatia, MD is increased in TSHI-veh pups compared to sham. Notably, EPO-treatment mitigated increased MD in the corpus callosum. To achieve precise, specific analyses of directional diffusion, diffusion along the primary eigenvector parallel (AD), and perpendicular minor eigenvectors (RD) to axons was investigated. The analyses revealed increased AD and RD in TSHI-veh brains (p=0.005 and p=0.003, respectively, Fig. 5B,C) compared to shams, suggestive of impaired axonal integrity and sustained deficits in myelination. EPO treatment ameliorated white matter RD abnormalities (p=0.035). Together, these results highlight that prenatal TSHI leads to sustained impairments in white matter microstructure, and that EPO treatment during the period of postnatal myelination improves RD after injury.

Figure 4.

DTI directionally encoded color maps in the coronal plane show differences in white matter microstructure and deep gray matter in TSHI-veh rats compared to Sham and TSHI-EPO rats. Red color indicates horizontal tracts, blue color indicates anterior to posterior tracts orthogonal to the plane, and green indicates vertical directionality. Specifically, injured gray and white matter regions (white arrows) show a loss of directionality, where a loss of color is equal to a loss of microstructural organization.

Figure 5.

DTI reveals significant differences in FA and directional diffusion following prenatal TSHI, with postnatal EPO attenuation of regional microstructural abnormalities. In white matter, TSHI-veh animals have significantly decreased FA (A) with increased AD (B) and RD (C). EPO attenuates abnormalities in FA and RD. By contrast, no differences are observed in sensory cortex FA, AD or RD (D, E, F). Like the white matter, hippocampal FA is reduced in TSHI-veh brains (G), with associated elevations in AD (H) and RD (I). Similarly, striatal FA is decreased in TSHI-veh brains compared to shams (J), with increases in AD (K) and RD (L). In the hippocampus and striatum, postnatal EPO treatment normalizes abnormalities in RD. In the thalamus, FA is reduced in TSHI-veh compared to Sham (M). Whereas no changes were observed in AD in the thalamus (N), RD was increased in TSHI-veh animals and this elevation was attenuated in EPO treated animals (O), similar to the effect of EPO on RD in the white matter, hippocampus and striatum. (two-way ANOVA with Bonferroni’s post-hoc correction, *p <0.05, **p ≤ 0.01, ***p ≤ 0.001)

Changes in diffusion and microstructure were also examined in essential gray matter regions including in the sensory barrel cortex, hippocampus, striatum and thalamus. No differences in FA, MD, AD or RD were present in sensory cortex between sham and TSHI-veh animals (Fig. 5D–F, Suppl. Fig 3). In deep gray matter, diffusion abnormalities were present in the hippocampus, striatum and thalamus of TSHI-veh brains compared to shams (Fig. 5G–O), suggesting disrupted microstructural organization in deep gray matter after prenatal TSHI. Specifically, hippocampus, striatum and thalamus each exhibited a pattern of microstructural damage similar to that observed in white matter, including MD (Suppl. Fig 3). Compared to shams, hippocampal FA in TSHI-veh rats was reduced (p=0.021), and AD and RD increased (p=0.001 and p=0.002, respectively, Fig. 5G–I). Likewise, striatal FA in TSHI-veh rats was reduced (p=0.011), and striatum AD and RD were increased in TSHI-veh rats (p=0.002 and p=0.001, respectively, Fig. 5J–L). Similarly, thalamus FA was lower in TSHI-veh brains compared to shams (p=0.006), and thalamus RD was elevated in TSHI-veh brains compared to shams (p=0.009, Fig. 5M,O). No difference in AD was observed between groups in the thalamus (Fig. 5N). Notably, postnatal EPO treatment reversed RD abnormalities in hippocampus (p=0.037), striatum (p=0.047) and thalamus (p=0.034) (Fig. 5I,L,O), suggesting gray matter microstructure is also responsive to EPO treatment. In summary, deep gray matter structure exhibited a similar pattern of microstructural abnormalities to white matter after injury, with reduced FA, and increased MD, AD and RD.

To establish a relationship between the DTI changes and gait abnormalities, we performed a correlative analysis of precise DTI and individual gait metrics. A Pearson’s product-moment correlation was run to assess the relationship between FA and RD in the white matter, hippocampus, striatum and thalamus and hindlimb stride phase, propel phase, stride length variation, ataxia coefficient, paw area at peak stance, and maximal paw pressure, and forelimb stride phase and length (Table 2). As expected, there was strong positive correlation between white matter FA and swing (r(27) =0.492, p=0.008) and a negative correlation with swing variation (r(27)=−.544, p=0.003) and ataxia coefficient (r(27)=−.567, p=0.002). Similarly, there were significant relationships between radial diffusion in major gray matter structures and gait, including paw area at peak stance and maximal paw pressure (Table 2). Collectively, these data reflect EPO’s influence on myelination and align these structural measures with improved functional outcome.

Table 2.

Pearson Correlation between gait parameters (Figure 1) and DTI parameters (Figure 5) with significant changes between vehicle-treated TSHI rats versus EPO-treated TSHI rats and shams. (n=9–11)

Hindlimb	White Matter FA		White Matter RD		Hippocampus RD		Striatum RD		Thalamus RD
	Pearson r	p value	Pearson r	p value	Pearson r	p value	Pearson r	p value	Pearson r	p value
Swing Phase	0.492	0.008	−0.377	0.044	−0.466	0.012	−0.508	0.006	−0.494	0.006
Propel Phase	0.537	0.003	−0.216	0.260	−0.439	0.019	−0.403	0.033	−0.365	0.051
Stride Length Variation	−0.544	0.003	0.459	0.012	0.496	0.007	0.495	0.007	0.293	0.124
Swing Duration Variation	−0.544	0.003	0.459	0.012	0.496	0.007	0.495	0.007	0.293	0.124
Ataxia Coefficient	−0.567	0.002	0.392	0.036	0.452	0.016	0.517	0.005	0.332	0.079
Paw Area at Peak Stance	0.307	0.112	−0.210	0.275	−0.422	0.025	−0.354	0.065	−0.225	0.240
Maximal Pressure	0.247	0.205	−0.159	0.411	−0.375	0.049	−0.275	0.156	−0.144	0.457
Forelimb
Stride Phase	0.570	0.002	−0.373	0.046	−0.433	0.018	−0.450	0.016	−0.326	0.084
Stride Length	0.613	0.001	−0.418	0.024	−0.500	0.007	−0.490	0.008	−0.389	0.037

EPO treatment mitigates acute CHOP expression

To demonstrate an acute mechanism associated with EPO’s putative beneficial effect, we performed immunohistochemistry and a quantitative electrochemiluminescent immunoassay for the transcription factor CHOP at P5. CHOP is a downstream effector of ER stress, and immediately upstream of initiation of apoptosis (Zhao et al. 2015, Carloni et al. 2016). Thus, we chose to evaluate CHOP at the a point in development approximately equivalent to 35 weeks gestation in humans, and immediately following EPO administration in treated pups. A marked increase in CHOP immunoreactivity is present in vehicle-treated TSHI animals compared to Shams. Notably, EPO treatment markedly attenuated CHOP expression in white matter (Fig. 6A). Concomitant with these changes in CHOP immunolabeling, MECI analyses confirms vehicle-treated TSHI cortex has approximately 18% more CHOP protein compared to SHAM (Fig. 6B). EPO treatment reduces CHOP levels in TSHI pups compared to vehicle-treated TSHI pups, consistent with acute protection from cell death (n = 3–6, p=0.007). Thus, neonatal EPO treatment mitigates signs of ER stress in the developing CNS, consistent with our prior findings of acute CNS recovery after prenatal injury with neonatal EPO treatment ((Mazur, Miller et al. 2010, Jantzie, Getsy et al. 2014, Jantzie, Corbett et al. 2015, Jantzie, Winer et al. 2016).

Figure 6.

Prenatal injury elevates expression of CHOP, a marker endoplasmic reticulum stress. (A) At P5 expression of CHOP is elevated in TSHI-veh brain, and reduced after neonatal EPO treatment, particularly in white matter as shown in the anterior commissure (*). (B) Elevation in CHOP protein expression was confirmed using a multiplex electrochemiluminescent immunoassay (two-way ANOVA with Bonferroni correction, **p<0.01).

EPO treatment ameliorates long-term changes in myelin and axons

Given the acute effect on CHOP observed with EPO administration from P1-P5, we next evaluated chronic mechanisms of EPO repair by evaluating myelination and axon health in adult animals after prenatal injury. Specifically, we stained myelin with black gold and performed immunohistochemistry for neurofilament. Neonatal EPO treatment attenuates myelin loss and preserves axons in white matter in the corpus callosum and internal capsule in adult TSHI rats (Fig. 7). Myelin staining demonstrates that EPO treatment improves axonal bundling in the anterior portion of the internal capsule and anterior commissure in adult TSHI rats after in utero injury, compared to TSHI-veh rats (Fig. 7C,F). As expected given the integrated health of axons and myelination, reduced neurofilament expression is present in white matter following prenatal injury (Fig. 7H). Together, the improvement in myelination and neurofilaments in adult TSHI rats following neonatal EPO treatment is consistent with the enhanced functional performance and microstructural integrity on DTI, and corroborates our prior findings of chronic recovery (Mazur, Miller et al. 2010, Jantzie, Getsy et al. 2014, Jantzie, Corbett et al. 2015, Jantzie, Winer et al. 2016).

Figure 7.

In adult rats, myelination (black gold staining) and neurofilament expression is reduced in TSHI-veh rats, and improves after neonatal EPO treatment. In white matter of the corpus callosum and internal capsule (A–C) and anterior commissure (D–F), compared to shams (A, D), loss of myelination is apparent in TSHI-veh rats (B,E). The abundance of myelin is increased in adult EPO-treated TSHI rats (C, F). Neurofilament expression is also reduced in adult TSHI-veh white matter, including the lateral internal capsule (H). After EPO treatment, neurofilament expression in TSHI-EPO white matter (I) is similar to that found in shams (G).

DISCUSSION

Using a prenatal rat model of in utero transient systemic hypoxia-ischemia, we demonstrate significant deficits in gait, activity and social interaction. Importantly, we provide evidence in a clinically relevant platform that deficits in social interaction and complex motor behavior are reversible with neonatal EPO treatment, together with improved brain microstructure on DTI, and recovery of myelination and neurofilament expression. Using sophisticated, multimodal, digital gait analyses, we demonstrate that prenatal TSHI causes gait abnormalities in rats similar to those observed in children born very preterm with a spastic gait, including shorter stride, reduced paw contact and ataxia (Kurz et al. 2012, Arpin et al. 2013), and similar to other models of prenatal hypoxia-ischemia (Delcour et al. 2011, Coq et al. 2016). Interestingly, EPO treatment mitigates specific gait abnormalities, including impairment in stride, stance and paw placement consistent with prior, less sophisticated gait analyses (Mazur, Miller et al. 2010). These findings are also in alignment with the results from a small early clinical trial that showed a reduction in cerebral palsy at two years in preterm survivors who received erythropoiesis-stimulating agents as neonates (Ohls, Kamath-Rayne et al. 2014). These results emphasize that prenatal TSHI induces chronic motor deficits that mimic cerebral palsy that is reversible with EPO treatment, and demonstrate that this model supports preclinical testing of emerging therapeutics for neonates with prenatal injury.

Given that children born very preterm are also prone to impaired social interactions (Arpi and Ferrari 2013, Healy et al. 2013, Anderson 2014, Johnson and Marlow 2014, Pyhala et al. 2014, Montagna and Nosarti 2016, Wolford et al. 2017), we subsequently evaluated rat social interaction including standard introduction, play and grooming behaviors. Quantification of interaction time revealed TSHI-veh rats had significantly limited social interaction compared to shams; this is the first time to our knowledge that impaired social interaction has been demonstrated in a preclinical model of encephalopathy of prematurity. Remarkably, we found that EPO restored deficits in social interaction and increased interaction time, consistent with a small early clinical trial that showed improved behavioral outcomes in preschoolers who were preterm and received erythropoiesis-stimulating agents as neonates (Lowe et al. 2017). Without doubt, deficits in social interaction are multifactorial, in which limited awareness of and limited attention to social cues are additive, together with reduced sensory stimuli integration (Wickremasinghe et al. 2013, Williamson and Jakobson 2014, Montagna and Nosarti 2016). Subplate abnormalities contribute to abnormal cerebral circuit development both in disorders of impaired socialization (Hutsler and Casanova 2016) and encephalopathy of prematurity (Volpe 1996, Robinson 2005, Volpe 2009, Kinney et al. 2012, Pogledic et al. 2014), and subplate neurons are essential to integration of multiple thalamocortical connections (Kanold and Luhmann 2010). Previously, we showed subplate loss and impaired cortical maturation also occur following prenatal TSHI, and resolve with EPO treatment (Jantzie, Corbett et al. 2015). Reduced GABAergic neurotransmission has also been implicated in decreased sociability (Paine et al. 2017). Loss of expression of GABAergic markers was observed in postmortem brain from preterm infants with diffuse white matter gliosis (Robinson, Li et al. 2006). Loss of the potassium chloride co-transporter KCC2 and GABA_AR subunits occurs after prenatal TSHI, and is reversible with postnatal EPO treatment (Jantzie, Getsy et al. 2014, Jantzie, Corbett et al. 2015). Taken together, these data demonstrate that deficits in social behavior are potentially reversible with a clinically available therapy. Moreover, these findings solidify the utility of this preclinical platform to mechanistically evaluate critical molecular and cellular substrates of social behavior.

Microstructural DTI abnormalities may be informative as a biomarker of global white matter injury and recovery as emerging interventions are tested in both preclinical and clinical trials. Indeed, defective myelination correlates with DTI microstructural abnormalities following preterm birth (Ment et al. 2009), suggestive of a therapeutic window while myelination progresses (Thompson et al. 2015). When compared to placebo-treated controls, very preterm infants treated with neuroprotective EPO regimens have fewer gray and white matter MRI abnormalities at term (Leuchter, Gui et al. 2014), and show cognitive improvement at two years (Ohls, Kamath-Rayne et al. 2014) and at preschool age (Ohls et al. 2016). Importantly, EPO has been demonstrated to be safe (McAdams et al. 2013, Ohls, Kamath-Rayne et al. 2014, Fauchere, Koller et al. 2015), with initial concerns for exacerbation of retinopathy of prematurity now disproven (Ohlsson and Aher 2014, Fauchere, Koller et al. 2015, Song et al. 2016). In the present study, we found a pattern of DTI abnormalities in young adult rats following prenatal injury that mimic those found in children born very preterm, and that EPO repair of structural connectivity correlates with improved gait and functional motor outcome (Song et al. 2002, Huppi and Dubois 2006, Constable et al. 2008, Ment, Hirtz et al. 2009, Thompson, Lee et al. 2015). The chronic reductions in FA, and increases in white matter in MD, AD and RD we found are comparable to other rodent models of perinatal brain injury (Morken et al. 2013), and similar to other preclinical models of white matter injury (Ruest et al. 2011, Aung et al. 2013, Tuor et al. 2014, van de Looij et al. 2014, Egger et al. 2016). These microstructural abnormalities suggest both impaired axonal integrity and reduced myelination (Song, Sun et al. 2002, Huppi and Dubois 2006, Ment, Hirtz et al. 2009), consistent with clinical data (Groeschel et al. 2014, Thompson et al. 2014, Sripada et al. 2015, Thompson, Lee et al. 2015), and our histological analyses of myelination and neurofilament protein expression in adult animals, confirming prior reports and quantitative assessment (Robinson, Petelenz et al. 2005). Importantly, our data indicate that postnatal EPO treatment can reverse abnormalities in brain microstructure, which is important in the context of recent recognition of white matter plasticity and relationship to axons and myelin together (Almeida and Lyons 2017). White matter from TSHI-EPO treated rats showed restoration of microstructure FA and RD that was indistinguishable from shams. These findings are consistent with recovery of myelination and neurofilament expression with EPO treatment as shown here with black gold impregnation and immunohistochemistry, and previously observed with histological and biochemical assays (Mazur, Miller et al. 2010, Jantzie, Miller et al. 2013, van de Looij, Chatagner et al. 2014, Hassouna et al. 2016, Jantzie, Winer et al. 2016), and prior clinical reports that show EPO-treated preterm infants have fewer gross MRI abnormalities present in white matter than placebo-treated preterm infants (Leuchter, Gui et al. 2014), and increased white matter FA (O’Gorman, Bucher et al. 2015).

With our multimodal DTI investigation, we also studied deep gray matter changes associated with prenatal injury as it is known that white and gray matter changes culminate in complex functional deficits (Volpe 2009). Similar to prior pathological and radiological studies of preterm infants (Pierson et al. 2007, Kinney 2009, Nagasunder et al. 2011, Salvan et al. 2014, Cai et al. 2017), we found deep gray matter abnormalities are more prominent than cortical changes. Specifically, we found changes in multiple diffusion metrics, including FA, RD and AD in the hippocampus, striatum and thalamus but not sensory cortex. Indeed, the thalamus appears particularly vulnerable to microstructural changes in preterm neonates and childhood survivors (Nagasunder, Kinney et al. 2011, Paquette et al. 2015, Menegaux et al. 2017). Distinct overlap exists between the evolution of thalamo-cortical and intrahemispheric cortico-cortico connectivity, suggestive of developmental synergy between thalamic morphology and the temporal emergence of cortical networks (Rose et al. 2014, Ceschin, Wisnowski et al. 2015, Cai, Wu et al. 2017, Hwang et al. 2017). Importantly, altered functional connectivity is identifiable in preterm infants prior to birth (Thomason et al. 2017), emphasizing the contribution of intrauterine insults to early CNS injury.

Most interestingly, RD showed recovery with EPO treatment in all three regions of deep gray matter. Because EPO treatment restored white and gray matter RD levels following prenatal injury, improved functional outcomes observed in TSHI-EPO rats may be related to restoration of myelination (Jantzie, Miller et al. 2013, Egger, Janz et al. 2016). Interestingly, development of AD is exceedingly complex in young children and clinical data emphasize directional diffusion metrics mature non-linearly, individually, regionally and along gradients (Krogsrud et al. 2016). Consistent with these findings, our results support the concept that neural activity, axonal membrane density and coherence, and myelin sheaths drive diffusion, with exclusive contributions to directionality, and thus unique, specific responses to injury and/or treatment (Beaulieu 2002, Concha 2014, Krogsrud, Fjell et al. 2016). Likewise, our data support the concept that a prominent component of functional recovery observed in both rodents and humans with CNS injury from prematurity reflects the influence of EPO on the recovery and trajectory of myelination.

Among preterm children, follow-up studies in early childhood report higher rates of neurologic and early cognitive impairments in boys compared to girls (Johnson et al. 2009, Moore et al. 2012, Peacock et al. 2012, Skiold et al. 2014, Kuban et al. 2016). Recent multicenter trials confirm boys have a higher prevalence of cognitive and behavioral impairment than girls in nearly all measures, and more often require assistive devices to ambulate (Kuban, Joseph et al. 2016). We found gait deficits tended to be worse in males, but both sexes showed recovery with EPO treatment. We also found hyperactivity and social interaction after prenatal injury, and in response to EPO treatment were similar in male and female rats. Notably, children born preterm often exhibit hyperactivity and limited inhibition (Hutchinson et al. 2013, Wilson-Ching et al. 2013, Anderson 2014, Murray et al. 2014). Widespread disparities in white matter microstructure and organization have been found in adolescents with ADHD (King et al. 2015). Little is known about the role of sex (King, Yurgelun-Todd et al. 2015), or recently discovered novel molecular mediators of attention, such as stress-inducible phosphoprotein 1 (STIP1) (Beraldo et al. 2015, King, Yurgelun-Todd et al. 2015) and serotonin deficiency (Banerjee and Nandagopal 2015, Whitney et al. 2016), in the elaboration of attention deficits and concomitant DTI abnormalities over time. Here resources did not allow sufficient DTI analyses to investigate the impact of sex on microstructural abnormalities after injury, and in response to EPO treatment. Further study is required to establish sex differences in phenotypic expression of impaired attention, and network trajectories. Importantly, the present study does reveal treatment response to EPO is independent of sex. Indeed, EPO treatment showed benefit and reversed deficits in multiple outcome measures in both male and female rats, including gait, social behavior and multiple DTI metrics. By contrast, EPO treatment was ineffective in mitigating the hyperactivity and hypermobility observed in our studies of open field behavior, emphasizing that the molecular and neural substrates of attention and activity are likely independent of those fundamental to the other functional outcomes studied here.

Directed translation of interventions, including adequate testing of target populations, requires a preclinical model that recapitulates the spectrum of deficits arising from preterm birth, especially behavior and cognition. Here, we provide first evidence of a valuable preclinical platform of prenatal injury that includes deficits in multiple behavioral realms relevant to preterm children as they age. Further, we demonstrate that EPO treatment following prenatal TSHI restores multiple functional deficits and acutely reduces CHOP expression, a marker of ER stress, consistent with our prior reports that EPO attenuates cleaved caspase-3 protein expression in a similar developmental window (Mazur, Miller et al. 2010, Jantzie, Getsy et al. 2014, Jantzie, Corbett et al. 2015, Jantzie, Winer et al. 2016). These results emphasize potential reversibility of deficits commonly found in the preterm patient population. EPO’s multiple mechanism of action and repair make it an attractive therapeutic candidate with benefit owing to reduced apoptotic death initiated from mitochondrial and endoplasmic reticulum stress, improved oligodendrogenesis, neurogenesis, inhibitory circuit formation, and reduced inflammation (Mazur, Miller et al. 2010, Jantzie, Miller et al. 2013, Jantzie, Corbett et al. 2015, Hassouna, Ott et al. 2016).

Supplementary Material

1. Suppl. Figure 1.

Odor detection is not impacted by injury or treatment. There was no difference in the time spent sniffing water, social and food odors between the sham (n=23), TSHI-veh (n=28) and TSHI-EPO (n=14) groups.

2. Suppl. Figure 2.

EPO treatment had no impact on hypermobility and lack of inhibition in TSHI rats. In open field heat maps, shams spend the majority of the time near the periphery. There was no perceivable difference between the TSHI-veh and TSHI-EPO groups.

3. Suppl. Figure 3.

EPO treatment normalizes changes in mean diffusivity (MD). Regions of interest analyzed in DTI studies (A, representative anterior slice = top, representative posterior slice = bottom). In corpus callosum (B) and white matter (C), striatum (E), hippocampus (F) and thalamus (G), TSHI-veh animals have significantly increased MD compared to sham rats. Notably, EPO attenuates abnormalities in MD in the corpus callosum, hippocampus and thalamus. By contrast, no differences in MD are observed in sensory cortex (D). (blue = corpus callosum; yellow = lateral capsular white matter and corona radiata; green = sensory barrel cortex; orange = striatum; purple = hippocampus; red = thalamus. Two-way ANOVA with Bonferroni’s post-hoc correction, *p < 0.05, **p ≤ 0.01, ***p ≤ 0.001)

Highlights.

• Prenatal hypoxia-ischemia results in impaired social interaction in juvenile rats

• Diffusion tensor imaging microstructural abnormalities parallel functional outcome

• Neonatal erythropoietin restores gait and social deficits after prenatal injury

• Erythropoietin restores radial diffusivity in white matter and deep gray matter