Restrained Dendritic Growth of Adult-Born Granule Cells Innervated by Transplanted Fetal GABAergic Interneurons in Mice with Temporal Lobe Epilepsy

Visual Abstract


Introduction
Hippocampal neural circuit dysfunction is thought to contribute to epileptogenesis and the development of spontaneous seizures in temporal lobe epilepsy (TLE; Goldberg and Coulter, 2013;Alexander et al., 2016). Mounting evidence suggests that disruptions to the normal pattern of adult neurogenesis in the dentate gyrus (DG) promote the development of hyperexcitability (Parent et al., 1997;Jessberger et al., 2005Jessberger et al., , 2007. Prolonged seizures, defined as status epilepticus (SE), alter granule cell (GC) neurogenesis in the DG of the hippocampus (Parent et al., 1997;Hattiangady et al., 2004;Jessberger et al., 2005;Hattiangady and Shetty, 2010). Following SE, many adult-born GCs migrate to ectopic locations in the hilus or molecular layer of the DG (Scharfman et al., 2000(Scharfman et al., , 2003bScharfman, 2005;Parent et al., 2006) and form basal dendrites that establish abnormal and hyperexcitable neural circuits (Thind et al., 2008;Murphy et al., 2011). Studies have shown that the number of hilar ectopic adult-born GCs positively correlates with the seizure burden, and experimentally ablating GCs born after SE reduces the seizure burden in TLE (Hester and Danzer, 2013;Cho et al., 2015;Hosford et al., 2016Hosford et al., , 2017. Paradoxically, although ablating adult-born GCs in rodents with SE reduced the incidence of seizures, it increased the duration of residual seizures, possibly due to eliminating both normotopic and ectopic adult-born GCs (Hosford et al., 2016;Yu and Krook-Magnuson, 2017).
The patterns of excitatory and inhibitory connections to both the adult-born and developmentally-born GCs are also altered in TLE (Du et al., 2017). Several factors contribute to this phenomenon, including mossy fiber sprouting (Okazaki et al., 1995;Buckmaster et al., 2002;Scharfman et al., 2003a;Buckmaster, 2012), dendritic hypertrophy (Patel et al., 2004;Arisi and Garcia-Cairasco, 2007;Walter et al., 2007;Nishimura et al., 2008), and dendritic spine loss (Jiang et al., 1998). Prolonged SE can also trigger excitotoxic cell death of hippocampal GABAergic interneurons and lead to compensatory sprouting of surviving interneurons (Zhang et al., 2009;Long et al., 2011;Peng et al., 2013;Buckmaster et al., 2017). Exactly how each of these changes contributes to hyperexcitability is not fully understood (Beck et al., 1996;Bausch and McNamara, 2000;Cossart et al., 2001;Althaus et al., 2015). However, these and other experimental findings support the hypothesis that SE triggers structural and functional modifications to neural circuits in the DG that promote hyperexcitability and the development of recurrent spontaneous seizures (SRS) in TLE. Considerable research is now focusing on understanding the cellular and molecular mechanisms for these structural and functional changes in TLE (Pun et al., 2012;LaSarge et al., 2015LaSarge et al., , 2016Nishimura et al., 2017;Santos et al., 2017). Converging evidence supports the hypothesis that abnormal cellular growth mechanisms and altered gene expression lead to progressive changes that are sufficient to cause recurrent seizures and epilepsy (Parent et al., 1997(Parent et al., , 2006Walter et al., 2007).
Prior work established that transplanting fetal rodent GABAergic interneuron progenitors harvested from the medial-ganglionic eminence (MGE) or purified GABAergic interneurons derived from human embryonic stem cells are effective at suppressing seizures in developmental and chemoconvulsant models of epilepsy (Baraban et al., 2009;Zipancic et al., 2010;Hunt et al., 2013;Cunningham et al., 2014;Henderson et al., 2014;Lee et al., 2014;Shetty and Upadhya, 2016). Some evidence suggests that MGE transplants also reduce mossy fiber sprouting (Henderson et al., 2014; but see Hunt et al., 2013). After transplantation into the hippocampus of mice with TLE, fetal mouse MGE progenitors were linked to increased spontaneous IPSCs in hippocampal GCs (Henderson et al., 2014) and increased phasic and tonic inhibition in other populations of host brain neurons (Baraban et al., 2009;Hsieh and Baraban, 2017) via ␣4 subunit containing GABA A receptors (Jaiswal et al., 2015). The notion that transplanted neurons integrate into host brain circuits and provide synaptic inhibition is further supported by optogenetic experiments showing that stimulation of transplanted channelrhodopsin 2 (ChR2)-expressing GABAergic neurons induced strong IPSCs in hippocampal neurons (Henderson et al., 2014;Hsieh and Baraban, 2017). Other mechanisms may also be involved, as MGE progenitor grafts were shown to induce critical period plasticity and rewiring in the spinal cord and cerebral cortex (Park et al., 2002;Lee et al., 2007;Southwell et al., 2010;Chohan and Moore, 2016;Spatazza et al., 2017).
As a further step toward defining the cellular mechanisms responsible for fetal GABAergic interneuron-mediated effects in TLE, we examined whether transplantation of mouse MGE progenitors into the DG of naïve or TLE mice altered the morphologic development of adult-born GCs. We found that compared to adult-born GCs that were not innervated by the transplanted cells, the adult-born GCs with fetal transplant innervation exhibited significantly smaller dendritic arbors and shorter distal dendrites. These changes to the dendritic arbors were chiefly within the outer molecular layer of the DG, where adult-born GCs normally receive excitatory inputs from the lateral entorhinal cortex (Woods et al., 2018). The overall growth of dendritic arbors was also reduced in adult-born GCs in naïve, non-epileptic mice that received innervation from transplanted GABAergic interneurons, suggesting a general mechanism whereby inhibitory inputs from transplanted MGE-derived GABAergic interneurons diminish dendritic growth. Table 1 contains a summary of the experimental groups of mice that were used for electrophysiological recordings and morphology of adult-born GCs. All materials are listed in Table 2 except for common laboratory reagents that were purchased from Sigma.

Animals
Animal protocols were approved by the Institutional Animal Care and Use Committee (IACUC). As differences in estrous cycle in female mice were reported to affect seizure susceptibility, we performed all studies in male C57BL/6NHsd adult mice (Envigo) to reduce animal use (Müller et al., 2009). The mice were purchased at four to six weeks of age and maintained singly in self-ventilating  Zhao et al., 2011).

Pilocarpine and drug administration
The pilocarpine model was used to replicate TLE in sixto eight-week-old male mice (Envigo, C57BL/6N strain, 18 -22 g), as described previously (Henderson et al., 2014). Seizures were induced in male mice, since females in this mouse strain show significant differences in seizure latency, which could increase variability in patterns of recurrent seizure severity and incidence (Müller et al., 2009). On the day of seizure induction, the male mice were injected with methyl-scopolamine (0.5 mg/ml, i.p.; Sigma-Aldrich), followed 30 min later by an injection of pilocarpine hydrochloride (280 mg/kg, i.p.; Sigma-Aldrich). One or two supplemental doses of pilocarpine (30 -60 mg/kg, i.p.) were administered 30 min after the initial injection, if the mouse failed to develop seizures that progressed to status epilepticus (SE). Seizures were rated behaviorally, based on a modified Racine scale until the animal reached SE (Shibley and Smith, 2002;Henderson et al., 2014). One hour after continuous SE, seizures were attenuated by injecting midazolam (9.1-11.1 mg/kg, i.p.; Akorn). Ringer's solution (1 ml, i.p.; Henry Schein) was administered as needed, until recovery. A total of 124 mice were induced by pilocarpine. Of these, 77 developed SE (63%), 47 either died during SE or failed to develop SE and were euthanized (37%). Of the 77 SE mice, 44 underwent stereotaxic surgery to inject retrovirus and MGEderived GABAergic progenitors in the DG and 41 of these mice were then used for slice electrophysiology and neuronal reconstructions. A total of 17 C57BL/6N male mice (six-to eight-week-old, 18 -22 g) did not receive pilocarpine and were used as naïve, non-epileptic controls; 13 underwent stereotaxic surgery to inject retrovirus and MGE-derived GABAergic progenitors in the DG of the hippocampus. These mice were subsequently used for slice electrophysiology and neuronal reconstructions. Three additional SE mice received MGE-derived GABAergic progenitor transplants and were used for immunohistochemical analyses of the transplanted cells.

Viral production and injection
We labeled adult-born GCs in naïve mice and in mice with pilocarpine-induced TLE by stereotaxic injections of a Moloney murine leukemia virus (MMLV)-based retroviral vector (Luikart et al., 2011(Luikart et al., , 2012 into the subgranular zone of the DG. In mice with SE, the times of retrovirus (RV) injections were selected to coincide with early stages of seizure-induced upregulation of adult neurogenesis (one or two weeks after induction of SE) or the chronic phase of TLE (six or 12 weeks post-SE).
The MMLV-based retrovirus was produced by transient transfection of GP-293 retroviral packaging cell line with a retrovirus with internal ubiquitin promoter expressing mCherry (redRubi) and vesicular stomatitis virus glycoprotein (VSVg). The supernatant containing the viral particles was purified using PEG6000, as described previously (Luikart et al., 2011(Luikart et al., , 2012. The retrovirus titer produced in our laboratory was 1-10 ϫ 10 8 transfection units per milliliter (TU/ml).
Twenty-four hours before stereotaxic surgery, the mice were given overnight access to running wheels (ENV-044, ENV-044-02; Med Associates Inc.) to increase neurogenesis (Holmes et al., 2004). Thirty minutes before stereotaxic injections of virus, the mice were injected with Meloxicam (0.03 ml, s.c.; Boehringer Ingelheim). Anesthesia was induced by isoflurane inhalation (Isothesia, Henry Schein) and lidocaine hydrochloride was applied to the skin overlying the midline of the skull for topical anesthesia (2% topical; Hi·Tech Pharmacal). Each mouse received eight stereotaxic injections of retrovirus (0.5 l per site); the injections were made bilaterally into four sites in each hippocampus at a rate of 0.25 l/min with a glass syringe (10 l, removable needle, Hamilton) equipped with a 30-gauge needle (30 GA, 1", point style 2, 45°, Hamilton), at the following coordinates: AP -1.9 mm, ML Ϯ1.1, DV -2.5 and -2.3; AP -2.5, ML Ϯ2.1, DV -2.2 and -1.8. For two injection sites along one needle track, the first injection was made into the more ventral location and the needle remained in place for 2 min before a second injection was made into the more dorsal location. The needle was left in place for 5 min before being withdrawn and the skin incision was sealed (VetBond tissue adhesive, 3M Corp.). Mice were kept on a heating pad until fully awake and then housed in their home cages.

Transplantation of MGE-derived GABAergic progenitors
MGE-derived GABAergic progenitors were transplanted two weeks post-SE in mice receiving retrovirus at one, six, or 12 weeks post-SE, except for one group of mice that received retrovirus injections two weeks post-SE; in these mice the GABAergic progenitors were transplanted six weeks post-SE. Donor cells were harvested from the MGE of embryonic day (E)13.5 transgenic mouse embryos derived from adult female breeders (Envigo, C57BL/6NHsd) bred to VGAT-ChR2-EYFP line 8 (JAX Stock No. 014548, B6.Cg-Tg(Slc32a1-COP4‫ء‬H134R/ EYFP)8Gfng/J; Zhao et al., 2011). EYFP-positive embryos were identified with specialized goggles (FHS/F-01 headlamp equipped with FHS/EF-2G2 emission filters, Biolog-ical Laboratory Equipment Maintenance and Service Ltd.). The MGE on each side of the embryonic brain was isolated by free-hand dissection in cold HBSS (Sigma) without calcium or magnesium, as described previously (Xu et al., 2004;Henderson et al., 2014;Vogt et al., 2015). The tissue was incubated in 0.125% trypsin (Gibco) in HBSS at 37°C for 12 min, transferred with fire-polished glass pipettes into trypsin inhibitor (1ϫ, Gibco) for 10 min, then mechanically dissociated first with a large bore (10 times) then a small bore, fire-polished glass pipette (10 times). The dissociated cells were centrifuged and suspended in cell transplantation media. Cell counts were performed with a hemocytometer and the cells were then further diluted in cell transplantation media to reach a final concentration of 1 ϫ 10 5 cells/l. The transplantation media consisted of L-15 supplemented with fibroblast growth factor (Cell Signaling), epidermal growth factor (Cell Signaling), Caspase Inhibitor (Promega) and B27 (Gibco), as described previously (Henderson et al., 2014). The cell suspension was maintained on ice during determination of cell number and then 1 l (containing ϳ100,000 cells) was slowly injected over the course of 5 min into a single injection site into the hilus of the DG, by means of stereotaxic surgery. Bilateral injections were made into each DG (stereotaxic coordinates AP Ϫ2.5 mm, ML Ϯ2.1 mm, DV Ϫ2.2 and Ϫ1.8 mm). The needle was left at the injection site for 5 min before being withdrawn. The skin incision was closed, and the mice were allowed to recover on a heated pad, then returned to their cages.

Hippocampal slice electrophysiology
After allowing approximately eight weeks for maturation of adult-born GCs and interneuron transplants, we performed whole-cell patch-clamp recordings of GCs in hippocampal slices, as described previously (Henderson et al., 2014). The mice were anesthetized with a mixture of ketamine hydrochloride (120 mg/kg, i.p.; Ketaset, Zoetis) and xylazine (10 mg/kg, i.p.; Anased, Lloyd Laboratories), before rapidly removing the brain, and transferring it to oxygenated, ice-cold ACSF (high sucrose ACSF; 27.07 mM NaHCO 3 , 1.5 mM NaH 2 PO 4 , 1 mM CaCl 2 , 3 mM MgSO 4 , 2.5 mM KCl, and 222.14 mM sucrose). Brains were mounted onto the chilled stage of a vibratome (Leica VT1000S) in the horizontal plane, bisected in the sagittal plane, and thick slices (350 m) were cut from ventral to dorsal. The slices were maintained in oxygenated ACSF (37°C; 125 mM NaCl, 1 mM CaCl 2 , 3 mM MgSO 4 , 1.25 mM NaH 2 PO 4 , 25 mM NaHCO 3 , 2.5 mM KCl, 25 mM glucose, 3 mM myo-inositol, 2 mM Na-pyruvate, and 0.4 mM ascorbic acid) and for electrophysiology, slices were individually transferred into a recording chamber containing oxygenated ACSF (125 mM NaCl, 1.5 mM CaCl 2 , 1.0 mM MgSO 4 , 1.25 mM NaH 2 PO 4 , 25 mM NaHCO 3 , 3.5 mM KCl, 25 mM glucose, 3 mM myo-inositol, 2 mM Na pyruvate, and 0.4 mM ascorbic acid). Electrophysiological recordings were obtained from retrovirally (RV)-labeled GCs and unlabeled GCs in voltage-clamp mode at 34°C using a Cesium gluconate intracellular solution (135 mM gluconic acid, 135 mM CsOH, 1 mM EGTA, 8 mM MgCl, 0.1 mM CaCl 2 , 10 mM HEPES, 2 mM Mg-ATP, and 0.3 mM Na-GTP) containing biocytin (11 mM, Sigma), to allow us to recover and stain the electrophysiologicallyrecorded cells at the end of the experiment. To record IPSCs, the cells were voltage clamped at ϩ10 mV. To examine putative synaptic connections between the transplanted MGE-derived interneurons and GCs, we optogenetically activated transplanted ChR2-expressing interneurons using blue light pulses and recorded the responses in DG GCs, as described previously (Henderson et al., 2014). The blue light stimulus was triggered using a Master 8 stimulator (AMP instruments) and consisted of five pulses of blue light (5-ms duration each) with an interstimulus interval of 200 ms (Henderson et al., 2014).

Analysis of electrophysiological recordings
The electrophysiological recordings were analyzed offline using IGOR software. The amplitude of each lightinduced IPSC was measured. For each GC, the amplitude of induced IPSCs was determined by taking the average of all the induced IPSCs across a trial. We also determined the IPSC latency to the light stimuli by measuring the delay between onset of the light stimulus and IPSC onset in all adult-born GCs that were confirmed to be mCherry ϩ and thus RV labeled. To calculate the relationship between the magnitude of the light-induced currents in GCs and the number of MGE transplant-derived synaptic puncta, we correlated the peak amplitude of light-induced currents to the total number of synaptic appositions counted with the Spots function in IMARIS software, as described below.

Immunohistochemical staining of thick slices
Following electrophysiological recordings, the slices were fixed overnight in 4% paraformaldehyde (Electron Microscopy Sciences). The slices were then equilibrated in PBS containing 30% sucrose and frozen in tissue freezing medium (General Data).
To confirm RV labeling in electrophysiologicallyidentified GCs within the vicinity of transplants of MGEderived interneurons, we immunostained all the slices for mCherry, eYFP and biocytin. The slices were thawed and transferred to blocking buffer (5% normal goat serum containing 0.3% Triton-X) for 1 h at room temperature (RT). Slices were incubated for 48 h at RT in primary antibody solution in blocking buffer containing: chicken anti-GFP (1:1000, Aves), rabbit anti-mCherry (1:1000, Invitrogen), and streptavidin Alexa Fluor 647 (1:500, Invitrogen; omitted when staining slices that did not contain biocytin-filled cells). The slices were then washed for 1 h in PBS (KPBS, 0.02 M) and transferred into secondary antibody solution containing goat anti-rabbit Alexa Fluor 568 (1:1000, Life Technologies) and goat anti-chicken Alexa Fluor 488 (1:1000, Life Technologies) for 24 h, before a final wash for 1 h. Sections were mounted on Superfrost Plus slides in Prolong Gold with DAPI (Invitrogen) and stored at -20°C in the dark.

Phenotypic characterization of transplanted cells
Three additional SE mice with MGE-derived GABAergic progenitor transplants were perfused with 4% parafor-maldehyde in 0.1 M phosphate buffer (pH 7.4), and the brains were equilibrated in ascending sucrose solutions (10%, 20%, and 30% sucrose in 0.1 M phosphate buffer, pH 7.4), then frozen in tissue freezing media (General Data). Forty-micrometer-thick cryostat sections were collected in 0.1 M PBS and immunofluorescent staining was performed to detect transplanted GABAergic interneurons expressing ChR2-EYFP (chicken anti-GFP, GFP 1020, Aves), in combination with one of the following neurochemical markers: parvalbumin (PV; rabbit anti-PV, 1:100, PV-27, Swant), somatostatin (SOM; rat anti-SOM, 1:100, mab354 Millipore), calretinin (CR; rabbit anti-CR, 1:100, ab704, Abcam), or neuronal nitric oxide synthase (nNOS; rabbit anti-nNOS, 1:1000, AB5380, Millipore). Incubations were performed on free-floating sections at 4°C for 48 h with agitation. Sections were washed in PBS and secondary antibodies were applied. Nuclei were labeled with NeuroTrace Nissl 647 (1:300, Life Technologies), and sections were mounted onto glass slides in Prolong with DAPI (Life Technologies). We quantified the neurochemical phenotypes of all single and double-labeled cells contained in the hippocampus in a total of 20 -24 sections/ mouse. For each mouse, six sections spaced 400 m apart were immunostained for EYFP and one of the four neurochemical markers. Using optical slices obtained from confocal microscopic images of the hippocampal sections, we quantified the total number of individual, transplanted EYFP ϩ interneurons in each hippocampus and the number that co-expressed an additional neurochemical marker.

Confocal imaging
For imaging biocytin-filled GCs and the neurochemical phenotypes of transplanted cells, we collected optical slices through the entire thickness of 350-m-thick fixed and stained hippocampal slices that had been used for electrophysiological studies. Confocal microscopy (Zeiss LSM 510 or a Leica SP8) was performed at 0.5 m steps using a 25ϫ objective (Zeiss Plan-Neofluar 25ϫ, N.A. 0.8, water immersion objective) and a 63ϫ objective (Zeiss C-Apochromat 63ϫ objective, N.A. 1.2, water immersion). For Sholl analyses, RV-labeled GCs were imaged at lower magnifications to record their positions and capture the entire dendritic arbors (Zeiss Plan-Apochromat 10ϫ, N.A. 0.45 and Zeiss Plan-Neofluar 25ϫ, N.A. 0.8, water immersion objectives; z-stacks with a step size of 1 m and frame size, 1024 ϫ 1024 pixels). For detection and quantification of putative synaptic contacts formed by the transplanted GABAergic interneurons onto RV-labeled GCs, we imaged each RV-labeled neuron at high magnification (C-Apochromat 63ϫ objective, N.A. 1.2, water immersion). These high-resolution images were obtained as z-stacks with a step size of 0.3 m and resolution of 2048 ϫ 2048 pixels. All confocal images were stored on local hard drive (Backup Plus Desktop Drive, Seagate Technology) and archived in long-term cloud storage (RStore).

3D reconstructions and dendritic arbor morphometric analyses
To analyze dendritic morphology in RV-labeled adultborn GCs, confocal z-stacks of cells were imported into IMARIS neuron reconstruction software (8.4.0, Bitplane). The dendrites were traced using the filament tracer tool. Morphometric analyses of the reconstructed dendritic arbors were performed to quantify the following cell properties: total dendritic length, number of different order branches, length of different branches, and number of branch points. For dendritic branch lengths and total dendritic lengths, the distance from the center of the soma to the beginning of the primary dendrite was excluded. To determine dendritic complexity, 3D Sholl analyses were performed from the center of GC somas in IMARIS, identifying dendritic branch intersections at 10-m intervals.
Because electron microscopy has not been used to confirm that the axonal swellings formed by the transplanted neurons onto the GCs in the host brains are bona fide inhibitory GABAergic synapses, we refer to these morphologic swellings as "putative synapses" throughout the article. To quantify putative transplant-derived synapses onto RV-labeled GCs, the Spots function in IMARIS was used to identify green fluorescent boutons within 0.5 m of the cell that had a diameter of ϳ0.75 m or larger. Each putative synapse was checked manually to eliminate false signals. It seems probable that these putative synapses are sites of neurotransmitter release, as it was previously demonstrated by optogenetic studies that light-induced depolarization of the transplanted cells induced strong IPSCs in nearby GCs with short latencies (Henderson et al., 2014). Additionally, ϳ60% of these axonal swellings were previously found to be closely associated with postsynaptic clusters of gephyrin, an inhibitory synapse-specific postsynaptic density protein (Henderson et al., 2014).
We further classified the RV-labeled adult-born GCs based on patterns of innervation; the adult-born GCs that had fewer than 15 transplant-derived boutons on the entire cell were classified as "non-innervated" by the transplanted cells whereas GCs with Ͼ15 transplantderived putative synapses were classified as "innervated." The Sholl analyses were performed on adult-born GCs that were RV-labeled after SE as follows: RV1 week: 42 GCs (29 non-innervated and 13 innervated), RV2 weeks: 27 GCs (16 non-innervated and 11 innervated), RV6 weeks: 30 GCs (20 non-innervated and 10 innervated), and 12 weeks: 27 GCs (10 non-innervated and 17 innervated).

Statistical analyses
All statistical analyses were performed in SAS, version 9.4 (SAS Institute). Sholl intersection data were analyzed using a mixed-effects model with an auto-regressive variance structure and multi-level nesting along with the use of MULTTEST procedure to correct for false discovery rates, due to multiple t-tests (Wilson et al., 2017). Analyses of the length and numbers of different order branches were also performed using a mixed model. Comparisons of the means of the total dendritic lengths across different groups were done with ANOVA. Data are presented as the mean Ϯ SEM.

What is the spatiotemporal pattern of adultneurogenesis in the hippocampus at different times after SE?
Given that adult-born GCs display morphologic and functional abnormalities and are believed to contribute to hyperexcitability in the pilocarpine model, we first characterized adult-born GCs generated in naïve and epileptic mice at different times after SE by RV labeling and confocal microscopy. To this end, we made RV injections into the subgranular zone of the DG in naïve and epileptic mice at one, two, six, or 12 weeks after inducing SE and allowed one to two months for the adult-born neurons to reach maturity before analyzing patterns of neurogenesis (Fig. 1A). Retroviral expression of mCherry completely filled the soma, dendrites and frequently the axons of adult born GCs (Fig. 1B). In naïve mice, we observed strong labeling of GCs throughout the GC layer (GCL) of the DG (Fig. 1C). In epileptic mice, robust labeling of adult-born GCs was observed in the GCL and occasionally in the hilus or subgranular zones of the DG, following retroviral injections at one, two, or six weeks post-SE ( Fig.  1D-F). In agreement with prior studies, chronically epileptic mice injected with retrovirus 12 weeks post-SE exhibited sparser labeling with more frequent clumps of adultborn GCs (Fig. 1G).

Do adult-born GCs receive functional inhibition from developing MGE transplants?
Given that MGE-derived GABAergic interneuron progenitors take approximately four weeks to mature after transplantation (Southwell et al., 2010), we allowed approximately eight weeks for both the transplanted cells and the RV-labeled GCs to mature. The degree of inhibitory synapse formation from the transplants onto adultborn GCs was assessed first by combining optogenetic stimulation with patch-clamp recordings in acute hippocampal slices and second, by immunostaining the fixed slices and performing high-resolution confocal imaging.
We first measured the extent of light induced inhibitory currents in GCs, by recording from RV-labeled, adult-born GCs by whole-cell patch-clamp recordings, while optogenetically activating slices containing ChR2-expressing interneuron transplants (Fig. 1A, timeline). A representative slice showing transplanted GABAergic interneurons in the hilus and RV-labeled adult-born GCs in the GCL, is shown in Figure 2A. An example of a RV-labeled GC born one week post-SE that was filled with biocytin and further studied by electrophysiology is shown (Fig. 2B-D, insets, magnified views,). This neuron received putative transplantderived synapses onto the dendritic shafts and spines (Fig. 2E, arrows; Chiu et al., 2013). Optogenetic stimulation of transplant-derived interneurons induced strong IPSCs in this adult-born GC (Fig. 2G). A complete reconstruction of the dendritic arbor revealed 74 transplant-derived, putative synaptic contacts (Fig. 2F).
We conducted additional electrophysiological studies in 23 hippocampal slices from six mice with RV labeling at one week post-SE and transplants at two weeks post-SE. We successfully recorded from 23 mature GCs in these slices and 10 showed IPSCs in response to optogenetic stimulation. Six of the GCs in this subset were also RV labeled. Dendritic branching patterns of all six GCs were reconstructed in IMARIS and we found an average of ϳ180 putative synapses from the transplanted interneurons onto these GCs and a mean lightinduced IPSC amplitude of 63.79 Ϯ 28.42 pA. These results, summarized in Table 1, establish that GCs born one week post-SE are heavily innervated by MGE transplants made a week later and that activating the putative synaptic inputs onto these cells induces powerful postsynaptic inhibition.

What functional subtypes of interneurons are present in the transplants?
To further characterize the transplanted GABAergic interneurons that could be responsible for driving the observed IPSCs in adult-born GCs, we evaluated the neurochemical subtypes of interneurons within the transplants. We selected molecular markers with little to no overlap in different classes of MGE-derived GABAergic interneurons and show typical immune-staining patterns in Extended Data Figure 2-1. Approximately 37% of the grafted neurons co-expressed VGAT and PV (207 PV ϩ / 554 EYFP ϩ cells) and 34% co-expressed VGAT and SOM (232 SOM ϩ /685 EYFP ϩ cells). A smaller percentage of ϳ12% co-expressed VGAT and nNos (73 nNos ϩ /588 EYFP ϩ cells), and ϳ8% co-expressed CR (52 CR ϩ /651 EYFP ϩ cells). While the immunohistochemical data show that ϳ71% of the transplanted GABAergic interneurons were either PV-or SOM-positive, any or all of these interneuron subtypes were associated in close proximity with the adult-born GCs.

Do MGE transplants innervate GCs born at asynchronous stages of development?
To determine whether transplanted GABAergic interneurons can innervate adult-born GCs at different stages of maturity, we made retroviral injections at two weeks post-SE and transplants at six weeks post-SE, and then recorded from GCs in these mice four to eight weeks later. In 8/15 cells with strong IPSCs in response to optogenetic stimulation (Fig. 3A-A''), post hoc staining confirmed RV labeling in 2/8 (mean IPSC amplitudes: 83.8 and 127.4 pA, respectively). These results suggest that optogenetic stimulation of the transplanted cells induced strong inhibition in many of the adult-born GCs ( Fig.  3A-C).

Do later-born GCs receive functional inhibition after MGE transplants have matured?
An additional 13 GCs were recorded from mice with retroviral injections at six weeks post-SE and transplants at two weeks post-SE (Fig. 3B-B"), eight were confirmed to have RV labeling, and 5/8 showed light-induced IPSCs (mean IPSC amplitude 175.8 Ϯ 51.4 pA). To test whether GCs born long after transplants were made would show strong transplant-derived synaptic inhibition, we characterized adult-born GCs in mice that received transplants at two weeks post-SE and retroviral injections at 12  . We were further able to confirm RV labeling post hoc in one of these cells. As a whole, these findings show that transplanted GABAergic interneurons are capable of providing functional inhibition onto GCs born several months after development of TLE. In the adult-born GCs that did not show light induced currents (Fig. 3D,E), confocal microscopy showed that the cells generally had Ͻ15 transplant-derived synaptic boutons onto their somas or dendrites. In total, we recorded from 55 GCs in mice with retroviral birth-dating at one, two, six, or 12 weeks post-SE. Of the 55 GCs with strong light-induced inhibitory currents, we confirmed retroviral mCherry expression in 14/55 by post hoc immunostaining. Twelve of the 14 RV-labeled GCs that were optogenetically responsive showed a delay of 3 ms or less (mean 1.6 ms, range 1-3 ms) between the beginning of blue light stimulus and IPSC onset, indicating a monosynaptic origin of the optogenetically-induced inhibitory currents onto them. Together, these findings indicate that MGE-derived interneurons continue to innervate and form functional inhibitory connections with adult-born GCs over a relatively long period, in excess of 10 weeks after transplantation.

What is the relationship between the number of synaptic inputs from the transplants and lightinduced IPSCs?
Having established that transplanted GABAergic interneurons induce hyperpolarizing currents in GCs born at different times after the development of TLE and transplantation, we next asked whether the number of putative synaptic inputs correlated with the strength of the postsynaptic inhibition, which would further strengthen the evidence that the MGE cells formed functional inhibitory synapses onto adult-born GCs. To address this question, we completed high-resolution reconstructions of 24 GCs from 13 mice that were optogenetically stimulated during electrophysiological recordings. As shown in Figure 3, the number of morphologically-identified putative synaptic boutons onto individual GCs ranged from 14 to 1103 putative synapses/cell. Moreover, the number of these inputs was highly correlated with the peak amplitudes of optogenetically-induced IPSCs onto each cell (R 2 ϭ 0.838; Fig. 3F). Remarkably, the largest number of transplantderived synaptic boutons (1103) was onto a GC born 10 weeks after the GABAergic interneurons were transplanted in a TLE mouse (12 weeks post-SE; Fig. 3C). These findings in chronically epileptic mice indicate that the strength (amplitude) of the light-induced inhibitory synaptic currents in adult-born GCs is strongly related to the density of innervation from transplanted interneurons and that long after transplantation, MGE-derived GABAergic interneurons retain a robust capacity for forming putative inhibitory synapses with new adult-born GCs.

Does functional innervation of adult-born GCs require epilepsy-induced circuit reorganization in host brain?
Many studies have shown that pilocarpine-induced SE results in selective loss of SOM ϩ interneurons in the hippocampus and sprouting of subsets of interneurons, as well as other host brain changes (Peng et al., 2013). To determine whether changes in the host brain in TLE are required for the transplanted cells to form synaptic inputs onto adult-born neurons, we studied the innervation of adult-born GCs in naïve, non-epileptic mice by RV labeling adult-born GCs in six-to eight-week-old naïve mice and then a week later, we transplanted fetal GABAergic interneuron progenitors into the DG. After allowing eight weeks for maturation of the RV-labeled GCs and transplanted GABAergic interneurons, we made patch-clamp recordings from the labeled GCs, while activating the slices with blue light. Of 39 GCs recorded in slices from 11 naive mice with transplants, the majority showed robust light-induced responses (24/39 cells, mean IPSC amplitude: 76.15 Ϯ 20.51 pA). One representative example of an RV-labeled, adult-born GC from a naïve mouse is shown in Figure 4A-C. This GC showed strong lightinduced IPSCs (Fig. 4D) and had extensive transplantderived putative synapses onto its soma and dendrites (Fig. 4F). Based on our Sholl analyses, this GC had a typical branching profile (Fig. 4E). In just under half (42%) of the 31/39 adult-born GCs with confirmed RV labeling in naïve mice, we were able to demonstrate postsynaptic inhibition by optogenetically stimulating the transplants (13/31 RV-labeled GCs). The remaining 18 RV-labeled GCs were non-innervated and served as an internal control group (also see Fig. 6). Taken together, these results  demonstrate that MGE-derived GABAergic progenitors transplanted into the adult DG robustly innervate adultborn GCs in both epileptic and naïve, non-epileptic adult mice.

Does innervation by MGE transplants alter the dendritic morphology of adult-born GCs?
As changes in dendritic morphology are welldocumented in hippocampal GCs in TLE, our central question was whether the increased inhibition from the transplanted GABAergic interneurons altered the dendritic growth of adult-born GCs. To address this question, we compared dendritic morphologies in innervated versus non-innervated adult-born GCs, using high-resolution 3D reconstructions and Sholl analyses of 157 RV-labeled adult-born GCs; including 126 GCs from SE mice and 31 GCs from naïve mice. An example of a GC born at six weeks post-SE and innervated by a transplant made at two weeks post-SE is shown in Figure 5. The transplant contained a large VGAT-EYFP ϩ interneuron in the vicinity of this GC (Fig. 5B, white arrow), shown at higher magnification in Figure 5C. Excitation of the ChR2-expressing axons in this slice evoked IPSCs in host brain GCs (Fig.  5D). The Sholl analysis of the dendritic arbor of this GC indicated a relatively normal pattern of dendritic branching (Fig. 5E) and the corresponding 3D morphologic reconstruction revealed a high density of transplant-derived putative synapses onto this cell's dendritic arbor (Fig. 5F).
To determine whether putative synaptic innervation from the transplanted interneurons induced structural changes to the dendritic arbors of adult-born GCs, we examined the total dendritic lengths (Fig. 6, Table 3). Strikingly, the sizes of the dendritic arbors in GCs with heavy input from the transplanted GABAergic interneurons were significantly smaller, compared to noninnervated GCs in the same animals (total dendritic lengths were on average, 1700.0 Ϯ 92.0 m in the population of innervated GCs vs an average of 2055.6 Ϯ 116.4 m in the sample of non-innervated GCs, p ϭ 0.0235; Fig. 6B). Moreover, a significant effect of transplant-derived innervation was found for GCs born one, two, or six weeks post-SE. When these populations of adult-born neurons were innervated by the transplants, they developed significantly shorter dendrites compared to non-innervated adult-born GCs generated at the same time. Although the RV-labeled GCs born 12 weeks post-SE were also heavily innervated by the transplants and showed a similar trend of having shorter dendrites compared to non-innervated GCs, overall the differences in dendritic arbor size were not statistically significant (1917.0 Ϯ 147 vs 2143.3 Ϯ 167.3 m, p ϭ 0.3211). The results of the statistical tests of significance for all Sholl data are shown in Table 4.
Overall, the total dendritic lengths for the innervated versus non-innervated GCs in different groups and the number of GCs that were analyzed for each group were as follows: in naive mice the total dendritic branches were 1700.0 Ϯ 92.0 m (n ϭ 13 transplant-innervated GCs) versus 2055.6 Ϯ 116.4 m (n ϭ 18 non-innervated GCs; p ϭ 0.0235). In epileptic mice with RV labeling at one week post-SE, the total dendritic lengths of the transplant innervated versus non-innervated adult-born GCs were 1690.1 Ϯ 101.5 m (n ϭ 13 transplant innervated GCs) versus 2039.9 Ϯ 84.0 m (n ϭ 29 non-innervated GCs; p ϭ 0.0129). In epileptic mice with RV labeling at two weeks post-SE, the total dendritic lengths of the transplant innervated versus non-innervated adult-born GCs were 1577.7 Ϯ 140.5 m (n ϭ 11 transplant-innervated GCs) versus 2161.6 Ϯ 93.0 m (n ϭ 16 non-innervated GCs; p ϭ 0.0027). In epileptic mice with RV labeling at six weeks post-SE, the total dendritic lengths of the transplant innervated versus non-innervated adult-born GCs were 1825.3 Ϯ 120.2 m (n ϭ 10 transplant innervated GCs) versus 2390.0 Ϯ 70.8 m (n ϭ 20 non-innervated GCs; p ϭ 0.0010). In epileptic mice with RV labeling at 12 weeks post-SE, the total dendritic lengths of the transplant innervated versus non-innervated adult-born GCs were 1917.0 Ϯ 147 m (n ϭ 17 transplant innervated GCs) versus 2143.3 Ϯ 167.3 m (n ϭ 10 non-innervated GCs; p ϭ 0.3211).
Given that the total dendritic lengths of adult-born GCs were significantly smaller when they were innervated by Movie 1. 3D animation of an adult-born GC (red) that was labeled by an RV injection into the dentate gyrus subgranular zone. This GC was born 12 weeks after SE and 10 weeks after transplantation of MGE-derived GABAergic progenitors. The animation shows sites of putative transplant-derived synaptic boutons terminating across this cell's soma and dendrites (yellow). Optogenetic stimulation of the transplanted interneurons near this GC resulted in robust, short-latency IPSCs in the cell (Fig. 3 C, C', C"), confirming input from the transplanted interneurons. Only the soma and proximal dendrites are shown in this animation and the complete reconstruction is shown in the extended data in Fig. 6  the MGE transplants, we next compared differences in proximal versus distal dendrite lengths (Fig. 6C-H). The GCs from naïve mice that received input from the MGE transplants showed reductions in the lengths of both proximal and distal dendrites, compared with noninnervated GCs (radii 50 -70 m and radii 180 -240 m, p Ͻ 0.05). However, in epileptic mice, dendritic growth was significantly more stunted distally in GCs that were innervated by the transplants (radii 170 -270 m, p Ͻ 0.01; Fig. 6D), whereas proximally, dendritic growth was unaffected. The most significant differences were found in 3rd through 7th order branches in GCs innervated by the transplants. For example, the average length of 3rd order branches was 64.4 Ϯ 6.8 m in innervated GCs versus 78.7 Ϯ 4.4 m in non-innervated GCs (p ϭ 0.0390); the average length of 5th order branches in transplantinnervated GCs was 72.2 Ϯ 8.0 versus 108.9 Ϯ 5.1 m in non-innervated GCs (p ϭ 0.0390); the average length of 6th order branches was 34.8 Ϯ 8.6 m in innervated GCs, versus 71.3 Ϯ 5.5 m in non-innervated GCs (p Ͻ 0.0001), and the average length of 7th order branches was 7.5 Ϯ 6.1 m in innervated GCs versus 22.0 Ϯ 3.9 m in non-innervated GCs (p ϭ 0.0197). We also found that adult-born GCs generated shortly after induction of SE showed the greatest effects; for example, GCs born one week post-SE had significantly shorter 3rd, 4th, and 6th order branches (3rd order branches: innervated 47.6 Ϯ 13.6 m versus non-innervated 78.9 Ϯ 7.6 m; p ϭ 0.0271); 4th order branches: innervated 73.1 Ϯ 10.7 m versus non-innervated 95.6 Ϯ 6.0 m; p ϭ 0.0437); 6th order branches: innervated 34.6 Ϯ 15.4 m versus noninnervated 66.7 Ϯ 8.6 m; p ϭ 0.0438). In contrast, GCs generated at two weeks post-SE only showed significantly shorter 5th order branches (82.5 Ϯ 16.0 vs 117.6 Ϯ 10.2 m; p ϭ 0.0385), and similarly, GCs generated six weeks post-SE only showed significantly shorter 5th order (50.3 Ϯ 17.3 vs 108.1 Ϯ 10.0 m, p ϭ 0.0024) and 6th order branches (24.1 Ϯ 20.5 vs 80.6 Ϯ 11.8 m; p ϭ 0.0102). We further compared branching patterns based on Sholl analyses for the entire populations of innervated versus non-innervated GCs born one, two, six, and 12 weeks post-SE (Fig. 6E-H; Table 3). In all of these groups, except for the GCs born 12 weeks post-SE, we found that input from the MGE transplants was linked to a significant reduction in the size of the dendritic arbors, with more restricted growth of the distal dendritic branches. While the GCs generated 12 weeks post-SE did not show statistically significant differences in branching, they did show a similar trend toward shorter distal dendritic branches ( Fig. 6H; Table 3, dendritic length data).
A larger sample of representative RV-labeled adult-born GCs are shown in Extended Data Figures 6-1, 6-2, 6-3, 6-4, 6-5. While none of the GCs born in naïve mice were dysmorphic, GCs in epileptic mice exhibited highly abnormal arbors with hilar basal dendrites. Taken together, these results indicate that in epileptic and naïve mice, adult-born GCs with putative synapses formed by the MGE transplants had significantly shorter dendritic arbors.

Discussion
We have used a retroviral approach to label adult-born GCs at successively different times in naïve or epileptic mice in combination with transplantation of MGE progenitors from E13.5 ChR2-EYFP ϩ mouse embryos either before or after RV labeling of adult-born GCs. We allowed for maturation of the transplants and adult-born GCs, then tested for functional integration of the transplants by optogenetically stimulating the transplanted cells, while patch-clamping nearby RV-labeled GCs in hippocampal slices. The GCs were concurrently filled with biocytin during recordings, and after fixing the slices, 3D reconstructions and morphometric analyses were conducted.
The major new findings of this study are that fetal ChR2-expressing mouse GABAergic progenitor transplants in the hippocampus of naïve or epileptic adult mice innervate dentate GCs born weeks or even months after the time of transplantation. Optogenetic stimulation of the transplants induced strong IPSCs in adult-born GCs, and the magnitude of these currents correlated with the number of transplant-derived (putative) GABAergic synapses  on these adult-born GCs. Additionally, in either naïve, non-epileptic mice, or epileptic mice, adult-born GCs receiving dense innervation from the transplanted GABAergic interneurons had significantly more compact dendritic arbors, due to reduced growth of distal dendrites.
Remarkably, GCs born during the chronic phase of epilepsy as late as 12 weeks post-SE were heavily innervated by the interneuron transplants and the magnitudes of optogenetically-induced IPSCs were as large as those observed in populations of adult-born GCs generated shortly after transplantation. However, later born GCs showed a more modest trend toward reduced dendritic growth which was not significantly different from noninnervated GCs, suggesting that transplant-induced effects on adult-born GCs and their dendritic growth may wane over time. It was previously shown that astrogliosis increases in the DG during the chronic phase of TLE, leading to compromised GABAergic inhibition, which could reduce efficacy of the both endogenous and transplanted GABAergic interneurons (Eid et al., 2004;Dengler et al., 2017). Indeed, a previous study examining MGE transplants in mice with TLE, showed that seizure suppression did not endure during the later phases of TLE (Henderson et al., 2014). However, further work is needed to determine whether shorter distal branches in adult-born GCs lead to functional changes that reduce either the connectivity or excitability of dentate GCs.

Relationship to prior studies of GC structural changes in epilepsy
The present study examined structural changes in the dendrites of normotopic adult-born GCs in naïve and SE mice following transplantation and integration of GABAergic progenitors harvested from the fetal mouse MGE. We confined our analyses to normotopic GCs, to increase our sample sizes. It would be important in future studies, however, to investigate whether inputs from transplanted fetal GABAergic cells also restrict the growth of dendrites formed by ectopic GCs born in the epileptic brain, as the ectopic adult-born GCs in particular, are thought to contribute to epileptogenesis and hyperexcitability.
Several studies in rodent models of TLE have compared dendritic arbor complexity in normotopic versus hilar ectopic GCs and reported greater distal GC dendrite branching (Sholl intersections) in epileptic compared to naïve rats (Cameron et al., 2011). Similarly, in tissue from human patients with TLE or extra-hippocampal lesions, GCs were found to have increased dendritic lengths in the inner molecular layer of the DG (von Campe et al., 1997). Our findings suggest that transplant-derived input onto GCs reduces both normal dendritic growth and epilepsyinduced dendritic overgrowth.

Role of GABA and downstream signaling pathways in structural changes in adult-born GCs
The observed changes in dendrite growth could be mediated by the spontaneous or synaptic release of GABA from the transplanted interneurons. Adult-born GCs receive their first inputs from GABAergic interneurons (Ge et al., 2006). GABA A receptor antagonists have revealed tonic GABA currents in developing GCs as early as 3 d after birth and GABA-mediated synaptic currents as early as 7 d after cell birth (Ge et al., 2006). In contrast, the first glutamatergic currents occur ϳ14 d after GC birth (Ge et al., 2006). Newborn GCs express high levels of Na ϩ -K ϩ -2Cl Ϫ transporter NKCC1 and thus, have a high intracellular chloride concentration, which causes GABA continued dendritic branching both proximally and distally, compared with non-innervated GCs. D, Grouped Sholl data showing significantly reduced dendritic arbors in innervated, adult-born GCs compared to non-innervated adult-born GCs in SE mice with transplants. While the innervated GCs formed similar patterns of proximal branching, they had significantly fewer distal branches, compared to non-innervated adult-born GCs. Significant differences found for radii at 150 m from the soma (p ϭ 0.03) and all radii between 170 and 290 m from the soma (p Ͻ 0.01). E-H, Sholl data broken down by the birthdate post-SE of adult-born GCs. Graphs indicate that innervated adult-born GCs had significantly fewer distal dendrites, compared to non-innervated adult-born GCs. E, Innervated GCs born one week post-SE had significantly fewer intersections at radii between 200 and 270 m from the soma, compared to non-innervated GCs (p Ͻ 0.05). F, Innervated GCs born two weeks post-SE had significantly reduced dendritic branching at radii 200 -250 m compared to the non-innervated GCs (p Ͻ 0.05). G, Innervated GCs born six weeks post-SE had significantly fewer dendritic branches at radii 150 -270 m compared to non-innervated GCs (p Ͻ 0.05). H, Innervated GCs born 12 weeks post-SE showed trend toward reduced branching compared to non-innervated GCs, but this trend did not reach significance. Asterisks indicate statistically significant differences between groups. to be depolarizing, before a developmental switch to a hyperpolarizing effect of GABA (Ben-Ari et al., 2012). The depolarizing effects of GABA are seen for at least two weeks after GCs are born in the rodent (Ge et al., 2006). The initial GABA-mediated depolarization appears to be important for regulating dendritic growth, as knocking down NKCC1 reduced dendritic length and branching up to 14 d after cell birth (Ge et al., 2006). Growth-modulating effects of synaptic or extra synaptic GABA could be mediated by a number of different intracellular signaling pathways. For example, a major mechanism for GABA-mediated modulation of intracellular pathways is through activation of voltage-gated calcium channels that are activated by GABA-mediated depolarization, resulting in calcium influx (Schmidt-Hieber et al., 2004). Consistent with this, application of exogenous GABA, to immature cerebellar granule neurons, increased calcium influxes and increased dendritic complexity whereas antagonists of voltage-gated calcium channels, or inhibiting calcium/calmodulin-dependent protein kinase IV, reduced dendritic growth (Redmond et al., 2002;Borodinsky et al., 2003). The effects of GABA and calcium influxes on dendrite growth are likely to be mediated through molecules such as CaMKII and CaMKIV, or downstream signaling proteins, such as mitogen-activated protein kinases (MAPKs) and protein kinase A (PKA; Ghosh and Greenberg, 1995), as inhibition of CaMKIV has been linked to reduced dendritic growth, while increased CaM-KIV activation has been shown to increase dendritic growth (Redmond et al., 2002). Considering the importance of voltage-gated calcium influxes in GC maturation, it will be important to determine whether innervation of adult-born GCs by fetal GABAergic interneurons alters key intracellular calcium-dependent signaling pathways. One downstream effector of CaMKIV is cAMP response element-binding protein (CREB; Redmond et al., 2002). In normal mice, a high percentage of adult-born GCs express phospho-CREB, the active phosphorylated form of CREB (Nakagawa et al., 2002). Retroviral expression of dominant-negative isoforms of phospho-CREB in newly generated GCs reduced dendritic growth and altered the orientation of dendrites (Jagasia et al., 2009). Conversely, upregulating phospho-CREB in adult-born GCs by a pharmacological approach increased dendritic length and branching (Fujioka et al., 2004). Similarly, incubation of primary hippocampal cultures with cAMP agonists increased the expression of phospho-CREB as well as dendritic length and branching (Fujioka et al., 2004). The expression of phospho-CREB in newly-generated GCs begins around 5 d after the cells become postmitotic and lasts up to ϳ21 d. The temporal pattern of phospho-CREB expression coincides with the time during which GABA exerts a depolarizing effect on these cells, indicating that GABA-mediated depolarization may play a key role in activation of CREB (Jagasia et al., 2009). One study found that knock-down of NKCC1 in newborn GCs reduced levels of phospho-CREB, indicating a critical role for GABA-mediated depolarization in phosphorylation of CREB at early stages of GC maturation (Jagasia et al., 2009).
Brain-derived neurotrophic factor (BDNF) is a key downstream target of CREB. The expression of phospho-CREB in immature adult-born GCs is correlated with maximal expression of BDNF (Bender et al., 2001). Increased glutamatergic neurotransmission increases BDNF levels, while increased GABAergic inhibition decreases BDNF levels (Zafra et al., 1991). Hippocampal slice cultures treated with BDNF showed increased MAP-2 expression and increased dendritic growth (Marty et al., 1996). Moreover, transplant-derived GABAergic synapses onto adultborn GCs might counteract Hebbian plasticity at excitatory synapses from the entorhinal cortex, as these synapses are also modulated by BDNF (Asztely et al., 2000). Given these prior observations, decreased BDNF release may be responsible for reduced dendritic growth that we observed in adult-born GCs receiving dense inputs from the MGE transplants.

Summary and conclusions
Our results add to a growing body of literature showing that transplantation of MGE-derived GABAergic interneuron progenitors into the mature brain alters host brain neural circuitry. Here, we provide novel findings showing that putative synaptic input from transplants of GABAergic interneuron progenitors was associated with reduced growth of adult-born GC dendrites. The transplanted interneurons formed putative inhibitory synapses with both proximal and distal dendritic branches, and an overall shortening of dendritic branches was found in adult-born GCs innervated by the transplants in naïve mice. In epileptic mice however, the effect was limited to distal dendrites.