Impairment of motor coordination and interneuron migration in perinatal exposure to glufosinate-ammonium

Glufosinate-ammonium (GLA) is a broad-spectrum herbicide for agricultural weed control and crop desiccation. Due to many GLA-resistant crops being developed to effectively control weeds and increase harvest yields, herbicide usage and the residual GLA in food has increased significantly. Though perinatal exposure by the residual GLA in food might affect brain development, the developmental neurotoxicity of GLA is still unclear. Therefore, this study aimed to investigate the effects of perinatal exposure to GLA on cortical development. The analysis revealed that perinatal GLA exposure altered behavioral changes in offspring, especially motor functional behavior. Moreover, perinatal GLA exposure affected cortical development, particularly by disrupting interneuron migration. These results provide new evidence that early life exposure to GLA alters cortical development.

The open field test. Open field tests were performed at PND 17 using a method previously described 14,15 .
The open field arena consisted of 25 × 25 × 30 cm square acryl boxes. Rats were acclimated in the testing room for 30 min prior to starting the test. Rats were placed in the center of the open field arena and left to freely explore for 15 min for the test session with a video recording system. Test chambers were cleaned with 70% isopropanol and distilled water before each session. The average speed, total distance, mobility rate, and time spent on the edge were automatically analyzed using ToxTrac program 16 . Rotarod test. Accelerating rotarod tests were performed as described previously 17,18 . At postnatal week 7, rats were tested on a rotarod (Panlab Harvard Apparatus, Barcelona, Spain) accelerating from 4 to 40 rpm in 300 s. Rats were evaluated for 9 trials per session. At least 180 s of resting time was allowed between each trial. The end of a trial was determined when rats fell off the rod or when they reached 300 s. The latency to fall, speed, and time were recorded for each trial.
Immunofluorescence staining. Immunofluorescence staining was performed as described previously 19 .
Primary cortical neuron culture. Primary cortical neuron cultures were developed as described previously 20 . GD 18 rat embryo cortices were isolated and dissected with trypsin. The cortical neurons were then plated on coverslips coated with 100 mg/mL poly d-Lycine (Sigma, St. Louis, MO) in neurobasal media supplemented with B27 (Invitrogen, Carlsbad, CA) and cultured for 1-7 days. For the neuronal viability test, Cell Counting Kit-8 (Dojindo Molecular Technologies, Rockville, MD) was used according to the manufacturer's protocol; absorbance was measured at 450 nm/640 nm using a SynergyMx microplate reader (BioTek, Winooski, VT). Data are represented as the means of triplicate values; trials were repeated independently at least 3 times.
Quantification and statistical analyses. Statistical analyses were performed using SigmaStat 3.5 software (Systat Software, San Jose, CA). Tests performed before e-weaning stage including righting reflex, grip strength, open field test were analyzed both male and female together using a Kruskal-Wallis one-way analysis of variance on ranks with the Dunn's method. The rotarod test was conducted the male and female were analyzed separately using a Kruskal-Wallis one-way analysis of variance on ranks with the Dunn's method. To count the Sox2 + and calbindin D + cells, at least 3 embryos were analyzed for each group. The number of Sox2-expressing cells were automatically measured in a 500 × 500 pixel area of the ventricular zone using an ImageJ program with an Image-based Tool for Counting Nuclei plugin (National Institutes of Health, Bethesda, MD). The number of calbindin D-expressing cells were counted in a cortical plate 300 pixels wide. To measure neurite length, primary www.nature.com/scientificreports/ cortical neurons were harvested at 1 day in vitro (DIV) and Tuj1 + neurites were automatically analyzed using ZEN software (Zeiss, Oberkochen, Germany). Cortical neuron cultures were repeated independently at least 3 times. Statistical significance was considered when p < 0.05 for * or p < 0.001 for ***. Values are expressed as the mean ± the standard error of the mean.
Perinatal exposure of glufosinate-ammonium induced an abnormal righting reflex response and motor coordination at postnatal day 6. Since decreases in weight can facilitate a developmental delay in motor functions, motor coordination was tested at PND 6 using a righting reflex test to investigate the ability of pups to flip onto their feet from the supine position ( Fig. 2A,B). A righting reflex test can examine trunk control abilities and postural imbalances 13,21,22 . The righting reflex index value and the average latency to reflex were not different between controls and GLA-treated pups (righting reflex index: GLA 0 mg/ kg, 2.77 ± 0.27 s, GLA 100 mg/kg, 2.44 ± 0.25 s; GLA 250 mg/kg, 4.59 ± 0.92 s; average latency to reflex: GLA 0 mg/kg, 2.77 ± 0.27 s; GLA 100 mg/kg, 2.44 ± 0.25 s; GLA 250 mg/kg, 4.59 ± 0.92 s) (Fig. 2C,D). Though their average latency was not altered, several pups exposed to 250 mg/kg GLA exhibited a long latency to flip from the supine position. These pups tried to return to the prone position constantly but failed to achieve the prone position rapidly.
To identify motor coordination defects, pups were divided into three groups according to flip latency (fast, moderate, and late); further analysis focused on the late latency group. Especially, late latency pups were not present in the GLA 0 mg/kg or GLA 100 mg/kg groups, but late latency group population increased and the latency to reflex also dramatically increased among GLA 250 mg/kg pups (late latency: 8.33 ± 2.41%, p < 0.05; latency of www.nature.com/scientificreports/ reflex: 23.47 ± 3.84 s, p < 0.05). (Fig. 2E,F). These data indicate that GLA exposure during the perinatal periods may lead to weakness in the limbs and trunk muscles as well as a motor coordination imbalance.
Perinatal exposure to glufosinate-ammonium decreases the hanging impulse at postnatal day 15. Due to the abnormal righting reflex response at PND 6, muscle weakness in all 4 limbs was evaluated by a grip strength test at PND 15 (Fig. 3A); this test assessed the ability of pups to resist falling from a wire mesh. The grip strength index and the latency to fall were similar between controls and GLA-exposed pups (grip strength index: GLA 0 mg/kg, 4.32 ± 0. 16 (Fig. 3B,C). Also, the hanging impulse was calculated to reflect the force needed to resist gravity. As the hang time involves maintaining a minimum force required to oppose the gravitational force, a hanging impulse is an advantageous analytical tool for measuring phasic tension 22 (Fig. 4B,C). Also, the mobility rate was only reduced in 100 mg/kg GLA-exposed pups (GLA 0 mg/kg, 93.74 ± 1.67%; GLA 100 mg/kg, 87.21 ± 2.19%, p < 0.05; GLA 250 mg/kg, 91.06 ± 1.29%) (Fig. 4D). An analysis of the time spent on the edges during the open field area exploration revealed that all GLAexposed pups spent significantly more time on the edges of the maze compared to control pups (GLA 0 mg/kg, 92.75 ± 1.43%; GLA 100 mg/kg, 97.87 ± 1.40%; GLA 250 mg/kg, 97.43 ± 0.63%; p < 0.05) (Fig. 4E).
Perinatal exposure to glufosinate-ammonium induced abnormal motor coordination and balance in male rats. Accelerating rotarod tests were conducted at postnatal week 7 to examine whether perinatal GLA exposure causes motor coordination and balance impairments; the latency to fall speed and time were measured. Male pups exposed to 250 mg/kg GLA showed significantly abnormal motor coordination and balance (latency to fall speed: GLA 0 mg/kg, 13  www.nature.com/scientificreports/ kg, 59.51 ± 2.82 s; p < 0.05) (Fig. 5A,B). These data indicate early motor coordination defects persist until the juvenile stage in male rats after perinatal GLA exposure.
Abnormal cortical interneuron development after perinatal exposure to glufosinate-ammonium. To address the underlying mechanism of motor coordination defects, rat premotor cortical areas were analyzed at GD 18. First, cortical progenitor cells in the cortex were investigated regarding whether neural progenitor cell survival and proliferation were altered by perinatal GLA exposure. Immunohistochemistry was performed with a Sox2 antibody for neural progenitor cells and a Tuj1 antibody for cortical neurons (Fig. 6A-C). In the ventricular zone, the number of Sox2 + neural progenitor cells did not change after GLA exposure (GLA 0 mg/kg, 685.75 ± 57.65 cells; GLA 100 mg/kg, 693.50 ± 57.02 cells; GLA 250 mg/kg, 755.25 ± 26.87 cells) (Fig. 6D). These data indicate that GLA exposure did not affect neural progenitor cell survival or proliferation. Abnormal interneuron development, including migration and circuit formation, clearly affect motor function 23,24 . To identify whether cortical interneurons are intact after GLA exposure, interneuron development in the motor cortical area was examined. Immunohistochemistry was performed with a calbindin D antibody for cortical interneurons (Fig. 6E-G). Most calbindin D + interneurons were differentiated from the ganglionic eminence and migrated to the marginal zone and subplate at embryonic day 16-18 in the developing mouse   (Fig. 6H). Especially, calbindin D + interneurons were reduced in the marginal zone and subplate; the total number of interneurons did not change. These numbers indicate that interneuron migration from the ganglionic eminence was disrupted by GLA exposure. In addition, the axon lengths of calbindin D + interneurons were slightly reduced after GLA exposure. (Fig. 6E'-G' arrow heads) Thus, perinatal GLA exposure disrupted cortical interneuron migration and axon outgrowth.

Discussion
In this study, perinatal GLA exposure induced abnormal motor coordination by disrupting cortical interneuron migration. Cortical circuits consist of glutamatergic excitatory neurons and γ-aminobutyric acid (GABA)ergic inhibitory interneurons. The balance of excitatory and inhibitory neurons is important for proper function; changes in this neural balance could cause multiple neurodevelopmental disorders in humans 3-6 . Meanwhile, evidence suggests that GABAergic inhibitory interneurons in the primary motor cortex directly control voluntary movement [25][26][27][28][29] . In this study, perinatal GLA exposure induced weakness of the limbs and trunk muscles as well as a motor coordination imbalance in rats. Decreased body weight in pups raises the possibility of delayed development, but the adolescent motility impairment demonstrated by the rotarod test supported the possibility that GLA directly affected the balance of excitatory and inhibitory during brain development. Also, perinatal GLA exposure decreased motility and increased the time spent on the edges in the open field test. Since altered GABAergic inhibitory interneuron development could affect developmental disorders including autism and intellectual disabilities, a relationship with these diseases should be investigated.
Moreover, perinatal GLA exposure disrupted the migration of interneurons expressing calbindin D. Interestingly, abnormal migration of calbindin D + interneurons showed a dose-response relationship with GLA  www.nature.com/scientificreports/ be divided by their markers, including parvalbumin, somatostatin, and serotonin receptor 3A 30 . As immature calbindin D-expressing interneurons did not express their subtype marker at GD 18, interneuron subtypes were not distinguished. Further investigation should be performed to determine the type of inhibitory neurons that experience migration changes and the functional disability outcomes due to interneuron alterations after perinatal GLA exposure. Mammalian glutamine synthetase regulates toxic ammonia and glutamate levels by converting glutamine 31,32 . Given that GLA acts a glutamine synthetase inhibitor, exposure to GLA could increase blood ammonia levels; also, hyperammonemia during pregnancy could lead to encephalopathy. However, perinatal GLA exposure did not induce a blood ammonia concentration in the dams in this study (Supplementary Fig. S1). This indicates GLA would not cause a change in the ammonia concentration in the fetal brain; thus, abnormal motor coordination and disrupted interneuron migration would not be due to GLA-related hyperammonemia. Meanwhile, it is well known that glutamine synthetase is expressed in the brain, especially astrocytes, but its expression in the developing cortex is still unclear 33 . We found that glutamine synthetase was expressed in developing cortical neurons ( Supplementary Fig. S2). Notably, glutamate is a neurotransmitter in the brain that regulates the growth rate and branches of axons and dendrites during brain development 34,35 . This axon and dendrite outgrowth is an important process to establish functional neural circuits and neural migration 36 . Furthermore, glutamine synthetase inhibition could increase brain glutamate levels and lead to abnormal migration and neurite outgrowth in inhibitory neurons. It is not clear whether prenatal GLA exposure increases brain glutamate levels; hence, brain glutamate levels and the role of glutamine synthetase should be further investigated. In conclusion, the present results provide new evidence that early life exposure to the GLA might affect the cortical development and facilitate behavioral changes.