Depletion of embryonic microglia using the CSF1R inhibitor PLX5622 has ad- verse sex-specific effects on mice, including accelerated weight gain, hyperac- tivity and anxiolytic-like behaviour

Jessica M. Rosin, Siddharth R. Vora, Deborah M. Kurrasch PII: S0889-1591(18)30375-1
DOI: https://doi.org/10.1016/j.bbi.2018.07.023
Reference: YBRBI 3458

To appear in: Brain, Behavior, and Immunity

Received Date: 12 April 2018
Revised Date: 5 July 2018
Accepted Date: 25 July 2018

Please cite this article as: Rosin, J.M., Vora, S.R., Kurrasch, D.M., Depletion of embryonic microglia using the CSF1R inhibitor PLX5622 has adverse sex-specific effects on mice, including accelerated weight gain, hyperactivity and anxiolytic-like behaviour, Brain, Behavior, and Immunity (2018), doi: https://doi.org/10.1016/j.bbi.2018.07.023

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

1Depletion of embryonic microglia using the CSF1R inhibitor PLX5622 has
2adverse sex-specific effects on mice, including accelerated weight gain,
3hyperactivity and anxiolytic-like behaviour
4
1,2,3#
Jessica M. Rosin1,2,3, Siddharth R. Vora4, and Deborah M. Kurrasch 5

61Department of Medical Genetics, Cumming School of Medicine, University of Calgary,
7Calgary, Alberta, CANADA
82Alberta Children’s Hospital Research Institute, University of Calgary, Calgary, Alberta
9CANADA
103Hotchkiss Brain Institute, University of Calgary, Calgary, Alberta CANADA
114Oral Health Sciences, Faculty of Dentistry, University of British Columbia, Vancouver, British

12
13
14
15
Columbia, CANADA

16#Corresponding author:
17Deborah M Kurrasch
18Cummings School of Medicine
19University of Calgary
203330 Hospital Drive NW, Room HSC 2215

21
22
Calgary, AB T2N 4N1

23Office: 403/210-6713
24Fax: 403/270-0737

25
26
27
28
29
[email protected]

30 Keywords

31

32

33
Microglia, hypothalamus, embryonic development, POMC neurons, weight gain, hyperactivity

34Abstract

35Microglia are the resident immune cells in the central nervous system (CNS). Originally thought

36to be primarily responsible for disposing of cellular debris and responding to neural insults,

37emerging research now shows that microglia are highly dynamic cells involved in a variety of

38neurodevelopmental processes. The hypothalamus is a brain region critical for maintaining

39homeostatic processes such as energy balance, thirst, food intake, reproduction, and circadian

40rhythms. Given that microglia colonize the embryonic brain alongside key steps of hypothalamic

41development, here we tested whether microglia are required for the proper establishment of this

42brain region. The Colony-stimulating factor-1 receptor (Csf1r) is expressed by microglia,

43macrophages and osteoclasts, and is required for their proliferation, differentiation, and survival.

44Therefore, to eliminate microglia from the fetal brain, we treated pregnant dams with the CSF1R

45inhibitor PLX5622. We showed that approximately 99% of microglia were eliminated by

46embryonic day 15.5 (E15.5) after pregnant dams were placed on a PLX5622 diet starting at E3.5.

47Following microglia depletion, we observed elevated numbers of apoptotic cells accumulating

48throughout the developing hypothalamus. Once the PLX5622 diet was removed, microglia

49repopulated the postnatal brain within 7 days and did not appear to repopulate from Nestin+

50precursors. Embryonic microglia depletion also resulted in a decreased litter size, as well as an

51increase in the number of pups that died within the first two postnatal days of life. In pups that

52survived, the elimination of microglia in the fetal brain resulted in a decrease in the number of

53Pro-opiomelanocortin (POMC) neurons and a concomitant accelerated weight gain starting at

54postnatal day 5 (P5), suggesting that microglia could be important for the development of cell

55types key to hypothalamic satiety centers. Moreover, surviving PLX5622 exposed animals

56displayed craniofacial and dental abnormalities, perhaps due to non-CNS effects of PLX5622 on

57macrophages and/or osteoclasts. Finally, depletion of microglia during embryogenesis had long-

58term sex-specific effects on behaviour, including the development of hyperactivity and

59anxiolytic-like behaviour in juvenile and adult female mice, respectively. Together, these data

60demonstrate an important role for microglia during the development of the embryonic

61

62
hypothalamus, and perhaps the CNS more broadly.

63 1. Introduction

64 Microglia are the resident mononuclear phagocytic immune cells of the central nervous

65system (CNS). During embryogenesis, macrophages generated from erythro-myeloid progenitors

66(EMPs) in the yolk sac travel to the neuroectoderm and invade the developing CNS to generate

67microglia (Ginhoux et al., 2010; Gomez Perdiguero et al., 2015; Kierdorf et al., 2013). Microglia

68continue to develop and alter their morphology and expression signatures throughout the CNS

69during embryonic, postnatal, and adult stages (Butovsky et al., 2014; Kierdorf et al., 2013;

70Matcovitch-Natan et al., 2016; Thion et al., 2018), suggesting differing roles for microglia across

71defined developmental stages. Moreover, following the closure of the blood brain barrier,

72peripheral myeloid cells are rarely found to infiltrate the brain parenchyma during normal

73healthy CNS conditions (Ginhoux et al., 2010; Mildner et al., 2007), demonstrating that

74microglia truly represent a unique tissue-specific macrophage important for the CNS.

75 Originally thought to be primarily responsible for disposing of cellular debris and

76responding to neural insults, emerging research now shows that microglia are highly dynamic

77cells involved in a variety of developmental processes. Study of embryonic microglia shows that

78they can play a role in neural progenitor maintenance (Antony et al., 2011), in addition to neural

79progenitor engulfment and elimination at the termination of neurogenesis (Cunningham et al.,

802013). In the embryo, fetal microglia can direct cortical neuron migration and differentiation, as

81well as modulate the wiring and lamination of cortical neurons (Aarum et al., 2003; Squarzoni et

82al., 2014). Postnatally, microglia influence neuronal connectivity via their roles in dendritic spine

83formation (Miyamoto et al., 2016) and synaptic pruning (Paolicelli et al., 2011; Schafer et al.,

842012; Tremblay et al., 2010). This continuously growing body of literature supporting a role for

85microglia both as the tissue-specific immune cells of the brain, and as functional players

86embryonically and postnatally, highlight the importance of microglia for normal healthy CNS

87development. The functional role of microglia during embryonic and postnatal development is

88further supported by C-X3-C motif chemokine receptor 1 (Cx3cr1) knockout mice, whereby

89disrupting microglia signaling causes hyperactivity, anxiolytic-like behaviours, and phenotypes

90resembling depression in mutant female mice (Bolos et al., 2018).

91 The Colony-stimulating factor-1 receptor (Csf1r) is expressed by microglia, macrophages

92and osteoclasts, and is required for their proliferation, differentiation, and survival (Li et al.,

932006; Patel and Player, 2009). Csf1r knockout causes complete microglia ablation that leads to

94defects in the postnatal olfactory bulb, as well as an increase in cortical neurons and astrocytes

95with a concomitant decrease in oligodendrocyte cell number (Erblich et al., 2011). Moreover,

96these Csf1r knockout mice often do not survive to adulthood (Erblich et al., 2011; Ginhoux et al.,

972010). In contrast, microglia ablation in the adult using the CSF1R inhibitor PLX3397 results in

98~99% microglia depletion in the brain; however, no cognitive or behavioural impairments were

99identified (Elmore et al., 2014). Following removal of this CSF1R inhibitor, microglia repopulate

100the adult brain within one week, and were thought to repopulate from the proliferation of Nestin-

101expressing neural cells that differentiate into microglia (Elmore et al., 2014). However, a more

102recent study shows that microglia in the adult brain might actually repopulate from the self-

103renewal of the few microglia that remain following microglia depletion using the CSF1R

104inhibitor PLX5622 (Huang et al., 2018).

105 Microglia are also important for hypothalamic satiety signaling and circuitry (Gao et al.,

1062017a; Gao et al., 2017b; Urabe et al., 2013; Valdearcos et al., 2017). For example, mice

107deficient in Brain-derived neurotrophic factor (BDNF) specifically in haematopoietic cells

108develop hyperphagia, obesity and insulin resistance (Urabe et al., 2013). Moreover, microglia-

109specific lipoprotein lipase knockout mice, when challenged with a high-carbohydrate high-fat

110(HCHF) diet, display Pro-opiomelanocortin (POMC) neuronal loss and weight gain (Gao et al.,

1112017a). Furthermore, microglia express leptin receptor (Chang et al., 2017) and mice with

112myeloid-specific leptin receptor elimination show accelerated weight gain and hyperphagia (Gao

113et al., 2017b), demonstrating that microglia signaling plays a role in proper maintenance of

114energy balance.

115 Here, we studied the timeline of PLX5622 mediated elimination of microglia in the

116embryonic hypothalamus, and examined the consequences of loss of fetal microglia on

117behavioural outcomes later in life. We showed that approximately 99% of microglia are

118eliminated by embryonic day 15.5 (E15.5) after treatment of pregnant dams with PLX5622 diet

119starting at E3.5. Following microglia depletion, we observed elevated numbers of dead cells

120accumulating throughout the developing hypothalamus. Moreover, elimination of microglia in

121the fetal brain decreased the number of POMC+ neurons and led to a concomitant accelerated

122weight gain starting at postnatal day 5 (P5). In addition, PLX5622 exposed animals displayed

123craniofacial and dental abnormalities. Finally, depletion of microglia during embryogenesis had

124long-term sex-specific effects on behaviour, including the development of hyperactivity and

125anxiolytic-like behaviour in juvenile and adult female mice, respectively. Together, these data

126demonstrate that microglia appear to play an important role during the development of the

127

128
embryonic hypothalamus, and perhaps across the entire CNS.

1292. Materials & Methods

1302.1. Animals

131 CD1 mice (Charles River) were used for all experiments not involving transgenic lines.

132
Cx3cr1-CreERT2 mice (B6.129P2(Cg)-Cx3cr1
tm2.1(cre/ERT2)Litt
/WganJ Stock No. 021160, The

133

Jackson Laboratory) were crossed to Rosa26tdTomato
tm14(CAG-
mice (B6.Cg-Gt(ROSA)26Sor

134tdTomato)Hze/J Stock no. 007914, The Jackson Laboratory) for embryonic flow cytometry

135experiments. Nestin-CreERT mice (B6.Cg-Tg(Nes-cre)1Kln/J Stock no. 003771, The Jackson

136
flox/flox
Laboratory) were crossed with Csf1r
tm1.2Jwp
mice (B6.Cg-Csf1r
/J Stock No. 021212, The

137Jackson Laboratory) and used with the Rosa26tdTomato reporter line for Nestin repopulation

138experiments. Microglia knockdown was achieved by administering the Plexxikon CSF1R

139inhibitor PLX5622 (1200 PPM added to chow AIN-76A, Research Diets) to pregnant dams

140starting at embryonic day 3.5 (E3.5). Control dams received control diet (AIN-76A, Research

141Diets). Both mice generated from Control diet and PLX5622 diet fed dams were put on a wet

142diet after weaning, in order to help mice experiencing feeding problems due to craniofacial

143and/or dental abnormalities. Cre recombinase was induced using 1mg 4-Hydroxytamoxifen

144(H7904, Sigma) dissolved in corn oil and administered intraperitoneally over two days as two 0.5

145mg doses. Animal protocols were approved by the University of Calgary Animal Care

146

147
Committee and followed the Guidelines for the Canadian Council of Animal Care.

148 2.2. Immunohistochemistry

149 Embryonic day 11.5 (E11.5), E13.5, E15.5, E17.5, and postnatal day 0 (P0), P2, P4

150brains were collected in ice-cold phosphate-buffered saline (PBS) and fixed in 4%

151paraformaldehyde (PFA) overnight at 4°C. The brains were then washed in PBS and equilibrated

152in 20% sucrose/PBS overnight at 4°C. Brains were embedded in Clear Frozen Section

153Compound (VWR, 95057-838) and cryosectioned (10–20 µm sections). For

154immunohistochemistry (IHC), cryosections were rehydrated in PBS, washed with PBT (PBS

155with 0.1% Triton-X), blocked using 5% normal donkey serum (NDS, Sigma) for 1 h at room

156temperature (RT), and exposed to rabbit anti-Fezf1 (1:100, Fitzgerald 70R-7693), rabbit anti-

157Iba1 (1:500, Wako 019-19741), goat anti-Iba1 (1:500, Abcam ab107159), mouse anti-Ki67

158(1:200, BD Pharmingen 550609), mouse anti-Nestin (1:4, DSHB AB2235915), rabbit anti-

159cleaved active Caspase 3 (1:500, BD Pharmingen 559565), rabbit anti-POMC (1:200, Phoenix

160Pharmaceuticals H-029-30), and/or goat anti-LepR (1:200, R&D Systems AF497) at 4°C

161overnight. Slides were then washed with PBT and exposed to secondary antibody (1:200, Alexa

162488 or 555 donkey anti-rabbit IgG, donkey anti-goat IgG, and/or donkey anti-mouse IgG, Life

163Technologies) for 2 h at RT. Sections were mounted using Aqua Poly/Mount (Polysciences Inc.).

164Fluorescent IHC images were captured on a Zeiss Axioplan 2 fluorescent microscope. Brightness

165and/or contrast of the entire image was adjusted using Adobe Photoshop CS5.1 if deemed

166

167
appropriate.

168 2.3. MicroCT imaging

169 Crania of euthanized P21 and P28 animals were scanned using the Scanco Medical

170µCT100 scanner at 55kVp, 200uA with an isometric resolution of 34.4µ. Following

171reconstruction, 3D volumes were rendered using Drishti software (V2.6.4). All scans were

172

173
visualized using similar thresholding and rendering parameters.

174 2.4. Flow cytometry

175 Whole E15.5 brains were extracted from Cx3cr1-CreERT2; Rosa26tdTomato embryos in

176HBSS (14025-092, Life Technologies). Whole brain tissue was dissociated into smaller pieces in

177EH media (DMEM (11965-092, Life Technologies), 10 mM HEPES (15630-080, Life

178Technologies), 2% FBS) using a scalpel. EH media (15 mL) was added to the dissociated tissues,

179and filtered through a 70 µm strainer (542070, Greiner Bio-One). The filtered cell suspension

180was then mixed with 5.4 mL Percoll (17-0891-02, GE Healthcare) stock solution (SIP: 9 parts

181(v/v) Percoll to 1 part 10 x PBS). 5 mL of Percoll d = 1,08 g/l was then placed under the cell

182suspension/SIP mixture. The sample was then centrifuged at 1200 g for 30 min at 20°C. Once

183the myelin debris was removed, the interface between the two liquid layers (cell suspension) was

184pipetted and placed into a clean tube, where EH media was added until the volume reached 45

185mL. The cell suspension was then centrifuged at 1350 rpm for 10 min at 4°C. The supernatant

186was removed and the cell pellet was reconstituted in 2 mL of Flow-PBS (1 x PBS with 3%

187BSA). The resulting cell suspension was analyzed by the University of Calgary’s Flow Core

188

189
Facility using a BD LSR II machine and analyzed using FlowJo software.

190 2.5. Mouse weight and size measurements

191 Control diet (n = 74-76) and PLX5622 diet (n = 45-52) pups (depending on postnatal

192deaths) generated from 6 dams for each experimental diet were weighed (g) on a scale every day

193from P0 to P21. 20 male mice (12 Control diet and 8 PLX5622 diet animals) and 36 female mice

194(28 Control and 8 PLX5622 animals) generated from 3 dams for each experimental diet were

195also weighed (g) at 6 weeks of age. Length (cm) and width (cm) measurements were taken from

196P0 (36 Control diet, 33 PLX5622 diet pups), P5 (34 Control diet, 30 PLX5622 diet pups) and

197

198
P21 mice (6 Control diet, 6 PLX5622 diet mice).

199 2.6. Developmental behavioural testing

200 Assessment of developmental behaviours began on P5 and continued through until P15.

201Behavioural analysis was performed on 40 Control diet and 16 PLX5622 diet pups generated

202from 3 dams for each experimental diet. Specific behaviours were assessed each day over the

203

204
indicated age range for each mouse pup.

205 2.6.1. Surface righting reflex (P5–P11)

206 Each pup was placed on its back on a flat surface. The time for the pup to return to its

207

208
four limbs was recorded. The cutoff time for this test was 30 sec.

209 2.6.2. Negative geotaxis (P5–P11)

210 Each pup was placed face down on a ramp with a 25-degree incline, covered with a paper

211 mesh to enable traction. The time for the pup to turn 180 degrees with its head oriented to the top

212

213
of the incline was recorded. The cutoff time for this test was 30 sec.

214 2.6.3. Locomotion (P5–P15)

215 Each pup was placed on a flat surface in the center of a circle (13 cm diameter). The time

216 for all four legs of the pup to exit the circle completely was recorded. The cutoff time for this test

217

218
was 30 sec.

219 2.7. Adolescent and adult behavioural testing

220 Behaviour was tested at 42–46 (adolescent) and 84–90 (adult) days of age using the same

22120 male mice (12 Control diet and 8 PLX5622 diet animals) and 36 female mice (28 Control and

2228 PLX5622 animals) generated from 3 dams for each experimental diet. Testing was conducted

223over 5 days (adolescent) or 7 days (adult), with at least 24 hours between each test. The tests

224were conducted in the order presented below. All adolescent and adult behavioural tests were

225performed in the University of Calgary’s HBI Advance Light and Optogenetics core facility

226

227
(HALO).

228 2.7.1. Open field (P42 adolescent, P84 adult)

229 Locomotion was assessed by observing the activity of mice in a square open field (60 cm

230x 60 cm) divided into a 5 x 5 grid (each of the squares comprising the grid were 12 cm x 12 cm).

231Open field lighting was set to 100 Lux. Mice were placed individually into the center of the field

232and allowed to explore for 10 min. The activity of the mice in the field was recorded using a

233movement tracking system (SMART 3 software) connected to a camera mounted above the field.

234The total distance travelled (cm), time spent resting (s), and percentage of time spent in each

235zone (open field was divided into 3 zones; zone 1 contained the outer squares, zone 2 contained

236the middle squares, and zone 3 contained the center square) was quantified. The mean scores +/-

237

238
S.E.M. were plotted.

239 2.7.2. Elevated plus maze (P43 adolescent, P85 adult)

240 Locomotion was assessed by observing the activity of mice in the elevated plus maze,

241which has two open and two closed arms (each arm is 21.6 cm in length and 7 cm wide).

242Elevated plus maze lighting was set to 60-70 Lux in the open arms, and 20-30 Lux in the closed

243arms. Mice were placed individually into the junction of the open and closed arms and allowed

244to explore for 10 min. The activity of the mice in the elevated plus maze was recorded using a

245movement tracking system (SMART 3 software) connected to a camera mounted above the

246elevated plus maze. The total distance travelled (cm), number of zone transitions, and percentage

247of time spent in each zone (open arms vs closed arms) was quantified. The mean scores +/-

248

249
S.E.M. were plotted.

250 2.7.3. Rotarod (P44-P46 adolescent, P86-P88 adult)

251 The rotarod is a test of motor ability in which the animal must continuously walk and

252maintain its balance on a rotating cylinder. The apparatus consists of a 10 cm wide lane with a

253cylinder, 3 cm in diameter, elevated to a height of 28 cm. Training was organized into 3 blocks

254of 3 trials conducted over 2 days. For each training day, a different rotational speed was used: 4

255rotations per minute (rpm) on day 1 and 10 rpm on day 2. During the day of testing, trials were

256organized into 3 blocks of 1 trial and the rotational speed was set to accelerate from 4 rpm to 40

257rpm over 5 min. For each trial, the mouse was placed on the cylinder facing against the direction

258of rotation and the latency to fall was recorded. The maximum length of training trials was

259180 sec, while the maximum length of the testing trials was 300 sec. For each animal, the score

260(latency before falling) for the best of three trials was plotted, which are depicted as mean scores

261

262
+/- S.E.M.

263 2.7.4. Three-chamber sociability test (P89-P90 adult)

264 Sociability between mice was evaluated using the three-chamber sociability test, whereby

265mice were placed in the center of a three-chamber field (see diagram in Supplementary Fig. 6E)

266and their activity in each of the three-chambers was evaluated. Three-chamber lighting was set to

26780 Lux. On day 1, mice were placed in the three-chamber for 10 min, where both chambers

268adjacent to the center chamber housed empty wire cages (used to determine if there is a side

269preference). During testing on day 2, mice were placed in the three-chamber for 5 min, where

270one chamber adjacent to the center chamber contained a cage with a stranger same-sex mouse

271and the other chamber adjacent to the center housed an empty wire cage. The activity of the mice

272in the three-chambers was recorded using a movement tracking system (SMART 3 software)

273connected to a camera mounted above the three-chamber field. The percentage of time spent in

274

275
each of the three-chambers was quantified. The mean scores +/- S.E.M. were plotted.

276 2.8. Quantification and statistical analysis

277 Every 10 µm Fezf1+ section (usually between 5-14 sections, depending on age) was used

278for microglia counts, cell death counts, and repopulation counts (n = 3 from 2-3 dams), while

279only the first three 10 µm Fezf1+ sections were used for POMC cell counts (n = 3, from 3 dams).

280Quantitative results for all cell counts, weight/size measurements, and behaviour are represented

281by mean scores +/- S.E.M. and were analyzed by two-tailed unpaired t-tests using Prism 6

282

283
(GraphPad Software).

2843. Results

2853.1. Maternal exposure to the CSF1R inhibitor PLX5622 dramatically reduced the number of

286microglia present in the embryonic brain

287 Considering that microglia colonize the embryonic brain alongside key steps of

288hypothalamic development (Marsters et al., 2016; Nesan and Kurrasch, 2016), we tested whether

289microglia are required for the proper establishment of the hypothalamus. To eliminate microglia

290from the fetal brain, we treated pregnant CD1 dams with the Colony-stimulating factor-1

291receptor (CSF1R) inhibitor PLX5622, since CSF1R activity is required for microglia survival (Li

292et al., 2006; Patel and Player, 2009). Pregnant dams were placed on either a control or PLX5622

293diet starting at embryonic day 3.5 (E3.5) in order to avoid disrupting the implantation process,

294given the importance of endometrial macrophages (Care et al., 2013; Wang et al., 2016).

295Samples were collected for immunohistochemistry (IHC) analysis and cell counts at E11.5,

296E13.5 and E15.5 following exposure to control or PLX5622 diet from E3.5 onwards (Fig. 1A).

297Iba1 antibody staining revealed that the number of microglia in the hypothalamus had

298dramatically decreased by E11.5 following exposure to the PLX5622 diet at E3.5 (Fig. 1B-C,

299arrowheads), and only approximately 12.5% of microglia remained (Fig. 1H; p = 0.0047). E13.5

300PLX5622 diet exposed embryos also showed a dramatic decrease in hypothalamic microglia

301numbers as compared to controls (Fig. 1D-E, arrowheads), with only approximately 7.2% of

302microglia remaining (Fig. 1H; p = 0.0001). By E15.5, the majority of microglia were found to be

303depleted from the embryonic hypothalamus of PLX5622 diet exposed animals (Fig. 1F-G,

304arrowheads), with approximately 99.9% of microglia eliminated from the E15.5 hypothalamus

305(Fig. 1H; p < 0.0001).

306 Considering that the few remaining microglia in the E15.5 hypothalamus following

307exposure to the PLX5622 diet appeared to be undergoing cell death (Supplementary Fig. 1L-M,

308arrowheads), we considered E15.5 to be our “depleted” time point for the hypothalamus moving

309forward. However, in order to ensure that this level of microglia depletion was not specific to the

310hypothalamus, and that it was indeed occurring throughout the E15.5 brain, we performed flow

311cytometry on E15.5 control and PLX5622 diet treated Cx3cr1-CreERT2; Rosa26tdTomato brains

312(Supplementary Fig. 1N). Whole brain analysis of all the cells in control diet fed animals

313demonstrated that the E15.5 embryonic brain is comprised of approximately 3.8% microglia

314(Supplementary Fig. 1O). Following exposure to the PLX5622 diet from E3.5 to E15.5, only

315approximately 1.7% of total microglia remain, demonstrating that approximately 98.3% are

316depleted throughout the entire E15.5 brain (Supplementary Fig. 1P). Together, these data

317indicate that exposure of pregnant dams to the CSF1R inhibitor PLX5622 from E3.5 to E15.5

318results in approximately 98.3-99.9% microglia depletion throughout the brain, and that at least

319

320
some of the ~1.7% remaining microglia appeared to be unhealthy and dying.

3213.2. Embryonic microglia depletion resulted in the accumulation of dead cells throughout the

322embryonic brain across development

323 Given that microglia are phagocytic immune cells that play an important role in cleaning

324up dead cells and cellular debris (Giulian et al., 1989), we asked if apoptotic cells were

325accumulating in the developing brain following the depletion of microglia. Similar to what was

326outlined above, E11.5, E13.5 and E15.5 brains were harvested from pregnant dams that were

327placed either on a control or PLX5622 diet starting at E3.5. Staining with an active cleaved

328Caspase 3 (CC3) antibody at both E11.5 and E13.5 revealed a significant increase in the number

329of CC3+ cells in the developing hypothalamus of PLX5622 exposed embryos as compared to

330controls (Fig. 1I-L, arrowheads), with an approximate 11-times increase in CC3+ cells in

331PLX5622 brains (Fig. 1O; E11.5 p < 0.0001, E13.5 p < 0.0001). Similarly, at E15.5 we observed

332a significant increase in cell death throughout the developing hypothalamus (Fig. 1M-N,

333arrowheads; Fig. 1O; p = 0.001). Together, Iba1 and CC3 double staining showed that as

334microglia numbers decreased there was a noticeable increase in cell death (Supplementary Fig.

3351B-G, arrowheads). However, when we examined the CC3+ cells more closely we only observed

336a few Iba1+/CC3+ dying microglia (Supplementary Fig. 1I, L-M, arrowhead) and NeuN+/CC3+

337dying neurons (Supplementary Fig. 1H, J-K, arrowhead), since the vast majority of CC3+ cells

338appeared to be apoptotic bodies. This indicates that during microglia depletion, dead and dying

339cells accumulate throughout the embryonic hypothalamus, and likely throughout the entire

340

341
embryonic brain.

342 3.3. Embryonic microglia do not appear to repopulate from Nestin+ neuronal cells

343 To determine whether embryonic microglia repopulated from Nestin+ neuronal cells, as

344has been proposed in the adult (Elmore et al., 2014), we removed pregnant dams from the

345PLX5622 diet at E15.5 and studied microglial repopulation. By E17.5, two days following the

346removal of the diet, we observed Ki67+ cycling microglia within the hypothalamus and

347throughout the brain (Fig. 2A-B, arrowhead); however, we were unable to find any obvious

348Iba1/Nestin double-positive repopulating microglia (Fig. 2C-E, arrowheads). To test whether our

349timing was off and in fact we were missing a transient Nestin-expressing Iba1+ microglia during

350the repopulation process, we generated transgenic E17.5 Nestin-CreERT; Rosa26tdTomato embryos

351that were exposed to the PLX5622 diet from E3.5 to E15.5 (Fig. 2F), thus, any repopulating

352microglia that was derived from a Nestin+ precursor would be detected by a tdTomato signal.

353We did not observe a strong overlap of Nestin-CreERT; Rosa26tdTomato signal in Iba1+

354repopulating microglia throughout the embryonic brain (Fig. 2G-H, arrowhead), and specifically

355when we analyzed the E17.5 embryonic hypothalamus we detected approximately 2.7% (2

356microglia/75 total microglia counted) that were double-positive (Fig. 2M); however, the

357tdTomato signal was also very weak (Fig. 2L-L’’’, arrowhead).

358 To test if Nestin+ neuronal precursors were required for microglia repopulation, we

359
flox/flox
generated Nestin-CreERT;Csf1r
;Rosa26tdTomato embryos. Since CSF1R is required for

360microglia proliferation, differentiation and survival (Li et al., 2006; Patel and Player, 2009), the

361elimination of this receptor in Nestin+ cells should shunt microglia repopulation if they are

362indeed arising from a Nestin+ precursor. Similar to what was previously outlined, pregnant dams

363were exposed to the PLX5622 diet from E3.5 to E15.5 and allowed to repopulate from E15.5 to

364
flox/flox
E17.5 (Fig. 2I). Staining for Iba1 in E17.5 Csf1r
flox/flox
as compared to Nestin-CreERT; Csf1r

365
flox/flox
animals showed that microglia still repopulate in Nestin-CreERT; Csf1r

mutant embryos

366(Fig. 2J-K, arrowheads), and that there was no significant difference in the numbers of

367repopulating microglia (Fig. 2N; p = 0.1835). Together, these data suggest that although Nestin

368may be expressed in a limited number of repopulating microglia, Nestin is not required for their

369

370
full repopulation.

371 3.4. Microglia repopulate the brain seven days after removal of the CSF1R inhibitor PLX5622

372 Given that microglia appeared to be 98.3-99.9% depleted by E15.5 (depending on the

373brain region examined), we next removed the pregnant dams from the PLX5622 diet at E15.5

374and studied microglia repopulation (Fig. 3A). Samples were collected for IHC analysis and

375counts at E17.5, postnatal day 0 (P0), P2 and P4 (Fig. 3A). Staining with Iba1 at E17.5 revealed

376~2.3% of total microglia numbers when compared to controls within 48 hours after removal of

377PLX5622 (Fig. 3B-C, arrowheads, Fig. 3J; p < 0.0001). By birth (E19/P0; approximately three

378days after removal of PLX5622), we observed a noticeable increase in the number of

379repopulating microglia in the hypothalamus of PLX5622 exposed animals as compared to

380controls (Fig. 3D-E), with microglia repopulation now approximately 43.4% of controls (Fig. 3J;

381p = 0.0402). At the same time, we noticed that embryonic microglia depletion caused a decreased

382litter size and an increase in the number of pups that died within the first two postnatal days of

383life (Table 1), which did not appear to preferentially effect either sex (data not shown). By two

384days after birth (P2; approximately five days after removal from PLX5622), we observed a

385striking increase in the number of repopulating microglia in the hypothalamus of PLX5622

386exposed animals as compared to controls (Fig. 3F-G, K-L), with microglia repopulation levels at

387approximately 80.1% of control (Fig. 3J; p = 0.0288). And finally, the number of microglia in

388the P4 (approximately seven days after PLX5622 removal) hypothalamus of PLX5622 diet

389exposed animals and controls appeared comparable (Fig. 3H-I, M-N), with some PLX5622

390animals showing up to 99.8% repopulation by this time-point (Fig. 3J; p = 0.2159). Combined,

391these data demonstrate that approximately seven days are required for microglia to repopulate

392

393
the brain following exposure to the CSF1R inhibitor PLX5622 during embryogenesis.

3943.5. Embryonic microglia depletion resulted in accelerated weight gain and a decrease in

395POMC neurons postnatally

396 Although embryonic microglia depletion resulted in a decreased litter size (Table 1), the

397majority of pups born to dams exposed to the PLX5622 diet do indeed survive. Therefore, we

398next assayed these surviving PLX5622 exposed pups for any obvious growth or developmental

399defects (Fig. 4A). At birth, P0 PLX5622 exposed pups were comparable in weight and length to

400controls (Fig. 4B, G; length p = 0.9825, weight p = 0.4121); however, P0 PLX5622 pups were

401slightly wider then controls (Fig. 4C; p = 0.0117). Moreover, the elimination of microglia in the

402fetal brain did not appear to influence their weight from P1 to P4 (Fig. 4G). By P5 PLX5622

403exposed pups were significantly shorter (Fig. 4D; p = 0.0005), wider (Fig. 4E; p < 0.0001), and

404heavier (Fig. 4G; p < 0.0001) than controls. Moreover, PLX5622 exposed pups continued to gain

405weight at an accelerated rate as compared to controls from P5 to P15 (Fig. 4G).

406 Given that microglia disruption in adult mice results in weight gain and loss of

407POMC neurons (Gao et al., 2017a; Gao et al., 2017b), we examined whether POMC neurons

408were affected in postnatal brains exposed to PLX5622 during embryonic development. IHC

409staining in P4 control and PLX5622 exposed animals showed a significant decrease in POMC+

410neurons (Fig. 4H-I, Supplementary Fig. 2A-F, arrowheads), with an approximately 45%

411reduction in the number of POMC neurons found in the tuberal hypothalamus of PLX5622

412brains as compared to controls (Fig. 4J; p = 0.0016). We also examined postnatal microglia for

413evidence of leptin signaling, similar to what is observed in the adult (Chang et al., 2017; Gao et

414al., 2017b), and did not detect leptin receptor expression at E15.5 (Fig. 4K, Supplementary Fig.

4152G-H, arrowheads) or P2 in hypothalamic microglia (Fig. 4L, Supplementary Fig. 2I-J,

416arrowheads). However, similar to previous findings in the adult (Chang et al., 2017), we detected

417leptin receptors on adult hypothalamic microglia (Fig. 4M, Supplementary Fig. 2K-L,

418arrowheads). Together, these data suggest that microglia might be important for the development

419

420
of specific neurons within hypothalamic satiety centers in a leptin receptor-independent manner.

4213.6. Embryonic exposure to the CSF1R inhibitor PLX5622 resulted in craniofacial and dental

422defects

423 Although embryonic microglia depletion resulted in accelerated weight gain from P5 to

424P15, we noticed that PLX5622 exposed pups began to lose weight starting at P16 (Fig. 4G; p =

4250.0488). PLX5622 exposed pups continued to lose weight up until weaning (P21), and were also

426significantly shorter compared to controls (Fig. 4F; p < 0.0001). Given that CSF1R is expressed

427on microglia, macrophages, and osteoclasts (Li et al., 2006; Patel and Player, 2009), and that P21

428PLX5622 exposed mice did not look as healthy as controls and appeared to be eating less of their

429food, we examined their craniofacial development. Gross morphological examination of the teeth

430of P21 control and PLX5622 exposed animals demonstrated that PLX5622 animals had altered

431morphology of upper incisors (Supplementary Fig. 3A-B, arrow) and noticeably smaller lower

432incisors (Supplementary Fig. 3A-B, arrowhead). We also observed eye defects in some PLX5622

433animals (50% of males in 3/3 litters; 50% of females in 1/3 litters), where they were unable to

434open one eye until later in adulthood (Supplementary Fig. 3C-D, dashed-circle).

435 Further examination of P21 and P28 control and PLX5622 exposed mice (Fig. 5A) using

436microCT (µCT) scanning revealed distinct doming at the posterior aspect of the cranial vault in

437PLX5622 exposed animals (Fig. 5B-E, #) and the absence of the expected flattening of the mid-

438cranial vault seen in control animals (Fig. 5B-E, green arrows). The upper and lower incisors in

439PLX5622 exposed animals lacked the typical tooth curvature seen in control animals and

440displayed distinct notches in enamel close to the incisal edges (Fig. 5B-G, asterisk) as well as

441ectopic enamel ridges further apically (Fig. 5B-G, arrowheads). Few animals also displayed

442malocclusions, with overlap of the incisor tips (data not shown). The continuous eruption of

443incisors did not appear to be inhibited, since most animals displayed contact of the upper and

444lower incisors (Fig. 5E). The first molars also displayed altered cuspal morphology, with a

445mesial enamel protuberance resembling an extra cusp visible in the PLX5622 diet group (Fig. F-

446G). The molar crowns were also mesio-distally elongated and narrow bucco-lingually compared

447to controls (Fig. 5F-G, dashed-arrow and unfilled arrowheads respectively). In addition, the

448molar roots appeared to be slightly taurodontic in the PLX5622 diet group compared to controls

449(data not shown). Notably, this phenotype was observed in all male and female animals exposed

450to the PLX5622 diet.

451 Closer examination of the dentition of PLX5622 animals prior to P21 demonstrated that

452these dental phenotypes were also apparent during this time (data not shown), which is consistent

453with the time in which we observed the PLX5622 exposed mice losing weight and doing poorly

454(Fig. 4G). While these dental defects continued to P28 (Fig. 5D-G), supplying PLX5622 exposed

455animals with a wet diet enabled both female and male mice to continue to grow to a healthy

456weight that was comparable to controls by 6 weeks of age (Fig. 4N-O; female p = 0.1342, male p

457= 0.8945). Together, these data demonstrate that exposure to the CSF1R inhibitor PLX5622 in

458utero resulted in craniofacial and dental defects postnatally, likely due to the effects of PLX5622

459

460
on osteoclasts.

4613.7. Depletion of embryonic microglia resulted in hyperactivity and anxiolytic-like behaviour in

462female mice

463 Considering that adult female Cx3cr1 knockout mice exhibit behavioural alterations

464(Bolos et al., 2018), we examined if microglia depletion embryonically causes neural defects that

465affect normal behaviours in juvenile, adolescent and/or adult PLX5622 exposed mice. First, we

466assayed whether embryonic microglia depletion affected juvenile behaviours between P5 and

467P15 (Fig. 6A). Although we did not observe significant changes in locomotion or negative

468geotaxis responses in PLX5622 exposed pups as compared to controls (Supplementary Fig. 4A,

469C), we detected a delay in PLX5622 pup surface righting as compared to controls

470(Supplementary Fig. 4B; P5 p < 0.0001), which recovered as the pups aged (Supplementary Fig.

4714B).

472 Next, we examined whether embryonic microglia depletion affected adolescent behaviour

473at six weeks of age, specifically between P42 and P46 (Fig. 6A). Open field analysis on P42

474female control and PLX5622 exposed mice demonstrated that PLX5622 adolescent female mice

475were hyperactive (Fig. 6B-C), traveling a significantly greater distance (Fig. 6F; p = 0.0084),

476with a significantly lower resting time (Fig. 6G; p = 0.0028). In contrast, adolescent control and

477in utero PLX5622 exposed female mice spent a comparable amount of time in the inner and

478outer regions of the open field (Fig. 6H; outer p = 0.8788, inner p = 0.8784). Similarly, analysis

479of elevated plus maze results from P43 control and in utero PLX5622 exposed mice

480demonstrated that PLX5622 adolescent female mice were hyperactive (Fig. 6D-E), traveling a

481significantly greater distance (Fig. 6I; p = 0.0012) and showing a significant increase in the

482number of zone transitions (Fig. 6J; p = 0.0005). However, adolescent control and in utero

483PLX5622 exposed female mice spent a comparable amount of time in the open and closed arms

484of the elevated plus maze (Fig. 6K; closed arms p = 0.0594, open arms p = 0.3269). Adolescent

485P46 in utero PLX5622 exposed female mice also spent a significantly longer period of time on a

486rotarod accelerating from 4 to 40 rpm as compared to controls (Supplementary Fig. 7A; p =

4870.0273).

488 Perhaps unexpectedly, adolescent in utero PLX5622 exposed male mice were not

489hyperactive (Supplementary Fig. 5A-D), since they did not show a change in distance traveled in

490the open field (Supplementary Fig. 5E; p = 0.4979) or the elevated plus maze (Supplementary

491Fig. 5H; p = 0.1390). Moreover, adolescent in utero PLX5622 exposed male mice showed no

492change in open field resting time (Supplementary Fig. 5F; p = 0.8344) or elevated plus maze

493zone transitions (Supplementary Fig. 5I; p = 0.2250). Adolescent male PLX5622 in utero

494exposed mice also showed no change in open field zone localization (Supplementary Fig. 5G;

495outer p = 0.3345, inner p = 0.3429) or elevated plus maze arm localization as compared to

496controls (Supplementary Fig. 5J; closed arms = 0.7930, open arms p = 0.4076). Adolescent P46

497in utero PLX5622 exposed male mice also showed no difference in latency to fall when placed

498on a rotarod (Supplementary Fig. 7B; p = 0.9832).

499 Similar to the adolescent male results, twelve-week old in utero PLX5622 exposed adult

500male mice likewise were not hyperactive (Supplementary Fig. 6A-D), given that they did not

501display a change in distance traveled in the open field (Supplementary Fig. 6E; p = 0.2985) or

502the elevated plus maze (Supplementary Fig. 6H; p = 0.2965). Moreover, adult male PLX5622 in

503utero exposed mice showed no change in open field resting time (Supplementary Fig. 6F; p =

5040.5017) or elevated plus maze zone transitions (Supplementary Fig. 6I; p = 0.8632). Adult male

505PLX5622 in utero exposed mice also showed no change in open field zone localization as

506compared to controls (Supplementary Fig. 6G; outer p = 0.6547, inner p = 0.6534) or elevated

507plus maze arm localization (Supplementary Fig. 6J; closed arms = 0.7250, open arms p =

5080.9810). Adult P88 in utero PLX5622 exposed male mice also showed no difference in latency

509to fall when placed on a rotarod (Supplementary Fig. 7D; p = 0.0700). Furthermore, three-

510chamber analysis (Supplementary Fig. 7E) of adult P90 control and in utero PLX5622 exposed

511male mice showed a comparable percent time spent in each of the three chambers

512(Supplementary Fig. 7F; empty cage p = 0.8512, center p = 0.8151, mouse p = 0.8106).

513 Finally, we examined whether embryonic microglia depletion affected adult female

514behaviour at twelve weeks of age, specifically between P84 and P90 (Fig. 6A). Open field

515analysis on P84 female control and in utero PLX5622 exposed mice showed no difference in

516distance traveled (Fig. 7A-B, E; p = 0.3289) and no difference in resting time (Fig. 7F; p =

5170.1111). Adult control and in utero PLX5622 exposed female mice also showed a comparable

518amount of time in the inner and outer regions of the open field (Fig. 7G; outer p = 0.6488, inner

519p = 0.6378). Similarly, analysis of elevated plus maze results from P85 control and in utero

520PLX5622 exposed female mice showed no difference in zone transitions (Fig. 7I; p = 0.2786),

521with PLX5622 exposed adult females actually traveling less than controls (Fig. 7C-D, H; p =

5220.0321). However, analysis of adult control and PLX5622 female mice showed that PLX5622 in

523utero exposed female mice spent significantly less time in the closed arms of the elevated plus

524maze (Fig. 7J; closed arms p = 0.0039, open arms p = 0.1440). Adult P88 PLX5622 in utero

525exposed female mice were found to show no difference in latency to fall when placed on a

526rotarod (Supplementary Fig. 7C; p = 0.2836). Furthermore, three-chamber testing

527(Supplementary Fig. 7E) of adult P90 control and in utero PLX5622 exposed female mice

528showed a comparable amount of time spent in each of the three chambers (Supplementary Fig.

5297F; empty cage p = 0.4685, center p = 0.2099, mouse p = 0.5361). Taken together, these data

530demonstrate that depletion of microglia during embryogenesis has long-term sex-specific effects

531on behaviour, including the development of hyperactivity and anxiolytic-like behaviour in

532

533
juvenile and adult female mice, respectively.

534 4. Discussion

535 To our knowledge, we are the first to show that approximately 99% of microglia are

536eliminated by E15.5 when pregnant dams are placed on the CSF1R inhibitor PLX5622 starting at

537E3.5. Following microglia depletion, we observed elevated numbers of apoptotic cells

538accumulating throughout the developing hypothalamus. Moreover, elimination of microglia in

539the fetal brain decreased the number of POMC neurons and led to a concomitant accelerated

540weight gain in postnatal animals. In addition, we observed craniofacial and tooth defects in the

541PLX5622 exposed animals. Finally, depletion of microglia during embryogenesis had long-term

542sex-specific effects on behaviour, including the development of hyperactivity and anxiolytic-like

543behaviour in juvenile and adult female mice, respectively. Together, these data demonstrate that

544microglia appear to play an important role during the development of the hypothalamus, and

545

546
likely other regions of the embryonic brain.

5474.1. In utero exposure to the CSF1R inhibitor PLX5622 resulted in ~99% microglia depletion

548and an accumulation of apoptotic cells in the embryonic brain

549 Other groups have used a diet to administer CSF1R inhibitors to adult mice and have

550demonstrated a range of success (~80% to 99% depletion) with varying timelines to total

551depletion (7 to 21 days) (Acharya et al., 2016; Chalmers et al., 2017; Elmore et al., 2014; Jin et

552al., 2017; Li et al., 2017; Rice et al., 2017; Spangenberg et al., 2016; Szalay et al., 2016).

553Specifically, the use of the PLX5622 diet on adults has been documented to result in ~90%

554depletion over two to six weeks in a model of radiation-induced cognitive deficits (Acharya et

555al., 2016), ~90% depletion over seven days in a neuronal lesion injury model (Rice et al., 2017),

556and ~80% depletion over 28 days in an Alzheimer model system (Spangenberg et al., 2016).

557Embryonically, there are no reports on the use of CSF1R inhibitors administered maternally

558through diet. At present, embryonic microglia have been disrupted using Cx3cr1-/-, CR3-/-, and

559DAP12-/- knockout animals (Bolos et al., 2018; Squarzoni et al., 2014). Moreover, microglia

560have been depleted using approaches such as Cx3cr1CreER to drive diphtheria toxin receptor

561expression in microglia (Parkhurst et al., 2013), liposomal clodronate injections (Cunningham et

562al., 2013), CSF1R-specific blocking antibodies (Hoeffel et al., 2015; Squarzoni et al., 2014), or

563Csf1r complete knockout animals (Erblich et al., 2011). In our system, in utero exposure of

564embryos to the CSF1R inhibitor PLX5622 from E3.5 to E15.5 via maternal diet resulted in

56598.3% to 99.9% microglia depletion, depending on the brain region analyzed. This allowed us to

566circumvent the early lethality of the Csf1r complete knockout models, as complete nulls often do

567not survive to adulthood (Erblich et al., 2011; Ginhoux et al., 2010), while still achieving a high

568level of microglia depletion.

569 Similar to other studies that have disrupted microglia signaling (Fourgeaud et al., 2016),

570we observed apoptotic cells accumulating throughout the hypothalamus as development

571proceeded in brains lacking microglia. Whether these apoptotic cells were caused by the absence

572of microglia, as if they required specific microglial signals for survival, or whether the presence

573of these apoptotic cells is a reflection of defects in normal phagocytosis, remains unknown.

574Moreover, while it is possible that the accumulation of these dead cells themselves could disrupt

575key brain developmental processes, such as neurogenesis/gliogenesis, neuronal/glial activity,

576and/or synaptic function, we propose it is the loss of microglia influence on specific neural

577developmental steps that is responsible for the phenotypic results we observe in PLX5622

578animals. Consistently, our results match well with groups that have used genetic knockout

579models to disrupt microglia signaling, such as Cx3cr1 knockout female mice that develop

580hyperactivity, anxiolytic-like behaviour and phenotypes resembling depression (Bolos et al.,

5812018). However, future analyses of the cellular environment and cytokine/chemokine changes

582following microglia depletion should be examined in order to better understand the signaling

583

584
deficits in the brain following microglia depletion.

585 4.2. Microglia do not appear to repopulate from Nestin+ neuronal cells during embryogenesis

586 Given that the use of CSF1R inhibitors does not affect the blood-brain barrier in adult

587animals (Elmore et al., 2014; Huang et al., 2018), we expected that microglia repopulation

588following depletion in the developing brain likewise occurs under a normal physiological state.

589Although others have shown that repopulating microglia express Nestin in the adult (Elmore et

590al., 2014), we observed very few repopulating Iba1/Nestin double-positive cells in the E17.5

591brain. Moreover, when we conditionally eliminated Csf1r specifically in Nestin+ cells, we did

592not observe a decrease in the number of repopulating microglia at E17.5, suggesting that

593embryonic microglia repopulate from a non-Nestin+ cell. However, given that we did detect a

594small proportion (~2.7%) of Nestin+/Iba1+ microglia, it is possible that Nestin becomes

595transiently expressed in repopulating microglia without being required. A recent study of

596microglia repopulation in the adult shows that newly born microglia arise from the few microglia

597that remain following depletion (Huang et al., 2018), and we propose this could also be the case

598

599
for embryonic microglia.

6004.3. Exposure to the CSF1R inhibitor PLX5622 during embryogenesis caused POMC neuronal

601loss, accelerated weight gain and craniofacial defects postnatally

602 Here, we uncovered accelerated weight gain in postnatal PLX5622 in utero exposed

603animals. Although the confounding dental phenotype prevented us from tracking this phenotype

604into adulthood, to the best of our knowledge this early postnatal weight gain is highly unusual,

605even among animal models with hypothalamic defects. Indeed, across various genetic

606backgrounds, usually the weight gain occurs later in adulthood and often time requires high-fat

607diet challenges (Gao et al., 2017a; Gao et al., 2017b; Urabe et al., 2013; Valdearcos et al., 2017;

608Yi et al., 2017; Zhan et al., 2013). Even Lepob/Lepob mice with a disruption in the leptin gene do

609not display increased body weight until >P15 (Bouyer and Simerly, 2013), thereby making this

610observed postnatal phenotype even more striking. Interestingly, the elimination of leptin receptor

611specifically in myeloid cells in adult animals causes accelerated weight gain (Gao et al., 2017b);

612however, we did not detect the expression of leptin receptors on embryonic or early postnatal

613hypothalamic microglia, thereby imply a different mechanism of action underlying the weight

614gain observed in these postnatal microglia depleted pups. Yet, at least in the adult, microglia are

615important for influencing hypothalamic energy balance signaling by other, non-leptin pathways

616(Gao et al., 2017a; Gao et al., 2017b; Urabe et al., 2013; Valdearcos et al., 2017). Specifically,

617BDNF deficiency (Urabe et al., 2013) or the disruption of lipoprotein lipase activity (Gao et al.,

6182017a) within microglia also causes a decrease in POMC neurons and/or weight gain. Therefore,

619additional studies are required to determine whether embryonic and/or early postnatal microglia

620target these same pathways or if they differentially influence the development of hypothalamic

621satiety centers. Furthermore, given that PLX5622 also causes depletion of macrophages and

622osteoclasts, it is possible that non-microglial effects mediate some or all of this weight gain

623phenotype.

624 Surprisingly, a marked cranial and dental phenotype was observed in the animals exposed

625to the PLX5622 diet during gestation. Considering that CSF1R is expressed in microglia,

626macrophages, and osteoclasts (Li et al., 2006; Patel and Player, 2009), one or more of these cell

627types must also contribute to craniofacial morphogenesis. As teeth expand during growth,

628removal of bone at the tooth-bone interface (TBI) is carried out by Tartrate-resistant acid

629phosphatase (TRAP)-positive osteoclasts (Alfaqeeh et al., 2013). This removal of bone is

630presumably required for normal tooth morphogenesis. Moreover, it is known that CSF-1 and

631Receptor activator of nuclear factor kappa-B ligand (RANKL) are requisite for osteoclast-

632precursor cell proliferation, osteoclast differentiation, survival and migration (Arai et al., 1999;

633Biskobing et al., 1995; Fuller et al., 1993; Sakai et al., 2006). Hence, the dental phenotype found

634in in utero PLX5622 diet treated animals may be a consequence of disturbance in

635osteoclastogenesis at the TBI, and a resulting increase in constriction of the developing enamel

636organs.

637 However, it is also possible that CSF1R expressed either in cells of the osteoclast,

638macrophage, microglial lineages and/or cells directly involved in tooth formation, play a role in

639various aspects of odontogenesis and tooth shape determination, independent of surrounding

640bone. The finding of ectopic enamel ridges towards the apical aspects of the incisors, and slightly

641taurodontic roots, supports such a direct role for CSF1R during tooth morphogenesis.

642Osteoclastic activity is certainly required for tooth eruption, with CSF-1 expression increased in

643the occlusal aspects of emerging murine teeth (Heinrich et al., 2005). However, in these pups

644exposed to PLX5622 during gestation, the eruption of teeth is not impeded. A possible

645explanation is that since the diet was discontinued at birth, osteoclast recruitment, expansion, and

646differentiation transpired normally at the time when teeth begin to erupt. The increased doming

647of the cranial vault may also reflect an altered morphology of the underlying brain during growth

648or a direct effect on bone formation and/or remodeling. Further investigations into the cranial

649and dental phenotypes resulting from PLX5622 exposure during gestation are being performed to

650

651
address these hypotheses.

6524.4. Microglia depletion in utero resulted in the development of hyperactivity and anxiolytic-like

653behaviours in female mice

654 Although previous studies using the CSF1R inhibitor PLX3397 showed that microglia

655depletion in the adult causes no behavioural deficits (Elmore et al., 2015; Elmore et al., 2014),

656here we demonstrated that the elimination of microglia embryonically resulted in behavioural

657phenotypes in both adolescent and adult animals. Specifically, PLX5622 gestationally exposed

658adolescent female mice were hyperactive using open field and elevated plus maze tests.

659Moreover, PLX5622 gestationally exposed adolescent female mice were found to perform better

660than controls on the rotarod. Interestingly, these phenotypes were not observed in PLX5622 in

661utero exposed adolescent male mice, demonstrating a sex-specific effect of microglia on the

662developing brain. Similarly, only in utero PLX5622 exposed adult female mice were shown to be

663less anxious than controls when analyzed using the elevated plus maze. However, this decrease

664in anxiety was only observed in the elevated plus maze and was not apparent in the open field,

665suggesting the phenotype is a mild decrease. Notably, our findings are consistent with those

666reported by Cx3cr1 complete knockout mice (Bolos et al., 2018), suggesting that disruption of

667microglia signaling through CX3CR1 to surrounding cells in the brain could be the cause of the

668behavioural phenotypes observed in our microglia depleted females. However, there are subtle

669differences in the timing of these behavioural alterations, and we did not test for depression-like

670behaviours in our animals. Therefore, additional behavioural tests should be performed on in

671utero exposed PLX5622 mice to directly compare the full spectrum of behavioural changes that

672result from microglial depletion. Combined, these studies suggest that microglia are important

673for development of key brain centers, which then leads to lasting effects on behaviour, since

674microglia have long repopulated in our model by the time the behavioural testing is conducted.

675 At the same time, we also acknowledge that our behavioural findings could be a result of

676a delay in microglia maturation and/or functional changes to the repopulating microglia during

677the postnatal period, which then influences behavioural measures later in life. Given that we did

678not observe complete microglia repopulation until P4 (seven days after removal of PLX5622),

679and that the repopulating microglia appeared to be more amoeboid and immature as compared to

680controls, these phenotypes could be indicative of changes to microglia signaling and/or function.

681Consistent with this notion, temporary depletion of microglia using liposomal clodronate

682specifically during the early postnatal period causes hyperactivity and anxiolytic-like behaviours

683in rats (VanRyzin et al., 2016), suggesting that microglia in the postnatal brain might also

684influence behavioural outcomes.

685 While our results demonstrate that fetal microglia are important for the development of

686normal mouse behaviours in females, we were surprised to find that microglia depletion

687embryonically did not result in any overt behavioural consequences in adolescent or adult male

688mice. These findings could reflect microglial sex differences, as microglia in the male brain

689mature on a different developmental trajectory as compared to microglia in the female brain

690(Bolton et al., 2017; Hanamsagar et al., 2018; Nelson et al., 2017; Schwarz et al., 2012; Thion et

691al., 2018). Especially considering that in P2 PLX5622 exposed brains the repopulating microglia

692were often more amoeboid, with thicker microglia projections, and it was not until around P4,

693the time at which repopulation was complete, that the repopulated microglia in PLX5622

694exposed animals were nearly indistinguishable from those found in control brains. These subtle

695differences in microglia morphology during the repopulation process postnatally could

696potentially contribute to the sex differences we observe, given the differences that exist in the

697normal developmental trajectory present in female and male mouse brains. Therefore, further

698behavioural testing of male mice exposed to PLX5622 gestationally will be required to more

699confidently conclude whether microglial loss during embryogenesis simply does not disrupt

700behaviour in males, or whether it differentially disrupts behaviours that have not been uncovered

701yet. Furthermore, it will be important to investigate early requirements for microglia in a sex-

702specific manner by analyzing changes in cell numbers (e.g., neuronal and glial), localization and

703connectivity, including changes in secreted factors and signaling molecules, separately in males

704

705
and females to try to determine what underlies these sex-specific changes in behaviour.

706 5. Conclusions

707 Here we explored the effect of embryonic microglia depletion on hypothalamic-mediated

708energy balance, as well as juvenile, adolescent and adult behaviours. Specifically, we

709demonstrated that the elimination of microglia in utero caused a decrease in POMC neurons and

710accelerated weight gain postnatally, suggesting that microglia could be important for the

711development of key neurons within hypothalamic satiety centers. Importantly, embryonic

712microglia depletion also showed long-term sex-specific effects on behaviour. Specifically,

713embryonic microglia depletion resulted in hyperactivity in juvenile female mice and the

714development of anxiolytic-like behaviour in adult female mice. Finally, we illustrate the use of

715CSF1R inhibitors through diet to pregnant dams as a means to efficiently eliminate microglia in

716the developing embryonic brain. Collectively, these findings demonstrate that microglia appear

717to play an important role during the development of the embryonic hypothalamus, and perhaps

718

719
across the entire CNS.

720Funding Sources

721J.M.R. is supported by a Canadian Institute of Health Research (CIHR) Postdoctoral Fellowship

722

723
(MEF-140891). D.M.K. is supported by CIHR (MOP-275053).

724Author contributions

725JMR and DMK conceived and designed the experiments. SRV performed and analyzed all

726microCT results. JMR performed all other experiments. JMR prepared the manuscript. JMR,

727

728
SRV and DMK edited the manuscript. All authors read and approved the final manuscript.

729Acknowledgements

730The authors thank Candi Grivel and Gaurav Kaushik for animal care, Taylor Chomiak for

731behavioural test training, Natasha Klenin for genotyping, and the entire Kurrasch lab for critical

732discussion. The authors also thank the University of Calgary’s Flow Core Facility and HBI

733Advance Light and Optogenetics core facility (HALO). We would also like to acknowledge

734

735
Plexxikon for providing the CSF1R inhibitor PLX5622 under the MTA.

736 Conflicts of interest

737

738
The authors declare no conflict of interest

739 Appendix A. Supplementary data

740

741

742

743

744

745

746

747
Supplementary data associated with this article can be found in the online version.

748Figure legends

749Figure 1. Embryonic microglia are depleted by E15.5 and show an accumulation of dead

750cells within the tuberal hypothalamus when using the CSF1R inhibitor PLX5622. (A)

751Schematic of CD1 crossing, dam diet exposure period during pregnancy, and sample collection

752time-points. Expression of Iba1 in E11.5 (B, C), E13.5 (D, E), and E15.5 (F, G) control diet (B,

753D, F) and PLX5622 diet (C, E, G) exposed embryonic brains. White arrowheads mark Iba1+

754microglia. (H) Quantification of the number of Iba1+ microglia present in the tuberal

755hypothalamus of control or PLX5622 diet exposed embryos at E11.5 (p = 0.0047), E13.5 (p =

7560.0001), and E15.5 (p < 0.0001). Expression of active cleaved Caspase 3 (CC3) in E11.5 (I, J),

757E13.5 (K, L), and E15.5 (M, N) control diet (I, K, M) and PLX5622 diet (J, L, N) exposed

758embryonic brains. White arrowheads mark CC3+ cells. (O) Quantification of the number of

759CC3+ cells present in the tuberal hypothalamus of control or PLX5622 diet exposed embryos at

760E11.5 (p < 0.0001), E13.5 (p < 0.0001), and E15.5 (p = 0.001). Dashed-lines outline the ventricle

761

762
and pial surface. 3V, third ventricle.

763Figure 2. Microglia do not appear to repopulate from Nestin+ neural precursor cells

764following removal of the CSF1R inhibitor PLX5622. (A, B) Iba1 and Ki67 expression in the

765brains of E17.5 embryos allowed to recover from PLX5622 diet exposure (E3.5 to E15.5) for 2

766days. White arrowheads mark Iba1/Ki67 double-positive microglia. (C-E) Iba1 and Nestin

767expression in the brains of E17.5 embryos allowed to recover from PLX5622 diet exposure (E3.5

768to E15.5) for 2 days. White arrowheads mark Iba1+/Nestin- microglia. (F) Schematic of Nestin-

769
CreERT crossing to Rosa26
tdTomato
, dam diet exposure period during pregnancy (E3.5 to E15.5),

7704OHT injections (E15.5 and E16.5), and sample collection at E17.5. (G, H) Iba1 and Nestin-Cre

771expression in the brains of E17.5 embryos allowed to recover from PLX5622 diet exposure (E3.5

772to E15.5) for 2 days. White arrowheads mark Iba1+/Nestin- microglia. (I) Schematic of Nestin-

773
flox/flox
CreERT; Csf1r
crossing to Rosa26tdTomato
flox/flox
; Csf1r
, dam diet exposure period during

774 pregnancy (E3.5 to E15.5), 4OHT injections (E15.5 and E16.5), and sample collection at E17.5.

775
flox/flox
(J, K) Iba1 and Nestin-Cre expression in the brains of E17.5 Nestin-CreERT; Csf1r
knockout

776embryos allowed to recover from PLX5622 diet exposure (E3.5 to E15.5) for 2 days. White

777arrowheads mark Iba1+/Nestin- microglia. (L-L’’’) Iba1 and Nestin-Cre expression in the brains

778
flox/flox
of E17.5 Nestin-CreERT; Csf1r
knockout embryos allowed to recover from PLX5622 diet

779exposure (E3.5 to E15.5) for 2 days. White arrowheads mark Iba1+/Nestin+ double-positive

780microglia. (M) Quantification of the proportion of Iba1+ repopulating microglia that are

781
tdTomato
tdTomato+ in the tuberal hypothalamus of E17.5 Nestin-CreERT; Rosa26
embryos allowed

782 to recover from PLX5622 diet exposure (E3.5 to E15.5) for 2 days. (N) Quantification of the

783
flox/flox
number of Iba1+ microglia in the tuberal hypothalamus of E17.5 Csf1r
control and Nestin-

784
flox/flox
CreERT; Csf1r

knockout embryos allowed to recover from PLX5622 diet exposure (E3.5 to

785

786

E15.5) for 2 days (p = 0.1835).

787Figure 3. Microglia repopulate 7 days after removal of the CSF1R inhibitor PLX5622. (A)

788Schematic of CD1 crossing, dam diet exposure period during pregnancy (E3.5 to E15.5), and

789sample collection time-points. Expression of Iba1 in E17.5 (B, C), newborn (E19/P0; D, E), P2

790(F, G), and P4 (H, I) control diet (B, D, F, H) and PLX5622 diet (C, E, G, I) exposed brains.

791White arrowheads mark Iba1+ microglia. (J) Quantification of the number of Iba1+ microglia

792present in the tuberal hypothalamus of control or PLX5622 diet exposed embryos at E17.5 (p <

7930.0001), birth (E19/P0; p = 0.0402), P2 (p = 0.0288), and P4 (p = 0.2159). (K-N) Higher

794magnification images of Ibat+ microglia in P2 (K, L) and P4 (M, N) control diet (K, M) and

795PLX5622 diet (L, M) exposed brains. White arrowheads mark microglia projections. Dashed-

796

797
lines outline the ventricle and pial surface. 3V, third ventricle.

798Figure 4. Embryonic microglia depletion results in accelerated weight gain and a decrease

799in POMC neurons in postnatal animals. (A) Schematic of CD1 crossing, dam diet exposure

800period during pregnancy (E3.5 to P0), and sample collection or weight/measurement time-points.

801(B) Quantification of mouse pup length (mm) of P0 control and PLX5622 exposed animals (p =

8020.9825). (C) Quantification of mouse pup width (mm) of P0 control and PLX5622 exposed

803animals (p = 0.0117). (D) Quantification of mouse pup length (mm) of P5 control and PLX5622

804exposed animals (p = 0.0005). (E) Quantification of mouse pup width (mm) of P5 control and

805PLX5622 exposed animals (p < 0.0001). (F) Quantification of mouse length (cm) of P21 control

806and PLX5622 exposed animals (p < 0.0001). (G) Quantification of mouse pup weight (g) of

807control and PLX5622 exposed animals from P0 to P21. (H, I) Expression of

808proopiomelanocortin (POMC) in P4 control diet (H) and PLX5622 diet (I) exposed brains. White

809arrowheads mark POMC neurons. Dashed-lines outline the pial surface. (J) Quantification of the

810number of POMC neurons in the tuberal hypothalamus of P4 control and PLX5622 exposed

811animals (p = 0.0016). (K-M) Expression of Iba1 and leptin receptor (LepR) in E15.5 (K), P2 (L),

812and adult (M) wild-type brains. White arrowheads mark Iba1+ microglia that are either LepR- or

813LepR+. (N) Quantification of mouse weight (g) of 6-week old female control and PLX5622

814exposed animals (p 0.1342). (M) Quantification of mouse weight (g) of 6-week old male control

815

816
and PLX5622 exposed animals (p 0.8945).

817Figure 5. Embryonic exposure to the CSF1R inhibitor PLX5622 results in craniofacial

818defects. (A) Schematic of CD1 crossing, dam diet exposure period during pregnancy (E3.5 to

819P0), and sample collection for microCT (µCT) scanning. (B-G) µCT scans of crania of P21 (B,

820C) and P28 (D-G) animals exposed to control diet (B, D, F) or PLX5622 diet (C, E, G) during

821gestation. (B-E) Lateral views show the doming of the posterior aspect of the cranial vault in

822PLX5622 diet animals (C, E; #) and the absence of typical vault flattening observed in control

823animals (B, D; green arrows). Notching of the incisal edges of the enamel is observed in

824PLX5622 diet animals (C, E; *) with a lack of the expected curved morphology of incisors in

825controls (B, D; curved green arrow). Ectopic enamel ridges are observed on the facial surfaces of

826upper and lower incisors of PLX5622 animals (C, E, G; filled pink arrowheads). Altered first

827molar shape can also be observed with an elongation in the mesio-distal direction (G; dashed

828pink arrow) and constriction in the bucco-linugal direction (G; open pink arrowheads) in

829PLX5622 animals. Images are from representative scans of n = 4 animals at each time point (2M,

830

831
2F) per group. Scale bar = 10mm.

832Figure 6. Depletion of embryonic microglia results in hyperactivity in adolescent female

833mice. (A) Schematic of CD1 crossing, dam diet exposure period during pregnancy (E3.5 to P0),

834and juvenile, adolescent and adult testing periods. (B, C) Representation of open field results for

835a 6-week old female control diet (B) and PLX5622 diet (C) exposed animal. (D, E)

836Representation of elevated plus maze results for a 6-week old female control diet (D) and

837PLX5622 diet (E) exposed animal. (F) Quantification of the total distance (cm) traveled by 6-

838week old female control and PLX5622 mice in an open field over a 10-minute period (p 0.0084).

839(G) Quantification of the time (s) 6-week old female control and PLX5622 mice spent resting in

840an open field over a 10-minute period (p = 0.0028). (H) Quantification of the percent time 6-

841week old female control and PLX5622 mice spent in each zone of an open field over a 10-minute

842period (outer p = 0.8788, inner p = 0.8784). (I) Quantification of the total distance (cm) traveled

843by 6-week old female control and PLX5622 mice in an elevated plus maze over a 5-minute

844period (p 0.0012). (J) Quantification of the number of zone transitions made by 6-week old

845female control and PLX5622 mice in an elevated plus maze over a 5-minute period (p = 0.0005).

846(K) Quantification of the percent time 6-week old female control and PLX5622 mice spent in

847each region of an elevated plus maze over a 5-minute period (closed arms p = 0.0594, open arms

848

849
p = 0.3269).

850Figure 7. Depletion of embryonic microglia results in anxiolytic-like behaviour in adult

851female mice. (A, B) Representation of open field results for a 12-week old female control diet

852(A) and PLX5622 diet (B) exposed animal. (C, D) Representation of elevated plus maze results

853for a 12-week old female control diet (C) and PLX5622 diet (D) exposed animal. (E)

854Quantification of the total distance (cm) traveled by 12-week old female control and PLX5622

855mice in an open field over a 10-minute period (p 0.3289). (F) Quantification of the time (s) 12-

856week old female control and PLX5622 mice spent resting in an open field over a 10-minute

857period (p = 0.1111). (G) Quantification of the percent time 12-week old female control and

858PLX5622 mice spent in each zone of an open field over a 10-minute period (outer p = 0.6488,

859inner p = 0.6378). (H) Quantification of the total distance (cm) traveled by 12-week old female

860control and PLX5622 mice in an elevated plus maze over a 5-minute period (p 0.0321). (I)

861Quantification of the number of zone transitions made by 12-week old female control and

862PLX5622 mice in an elevated plus maze over a 5-minute period (p = 0.2786). (J) Quantification

863of the percent time 12-week old female control and PLX5622 mice spent in each region of an

864

865

866

867

868

869

870

871

872

873

874

875

876

877

878

879

880

881

882

883

884

885
elevated plus maze over a 5-minute period (closed arms p = 0.0039, open arms p = 0.1440).

886References

887Aarum, J., Sandberg, K., Haeberlein, S.L., Persson, M.A., 2003. Migration and differentiation of
888neural precursor cells can be directed by microglia. Proc Natl Acad Sci U S A 100, 15983-
88915988.
890Acharya, M.M., Green, K.N., Allen, B.D., Najafi, A.R., Syage, A., Minasyan, H., Le, M.T.,
891Kawashita, T., Giedzinski, E., Parihar, V.K., West, B.L., Baulch, J.E., Limoli, C.L., 2016.
892Elimination of microglia improves cognitive function following cranial irradiation. Sci Rep 6,
89331545.
894Alfaqeeh, S.A., Gaete, M., Tucker, A.S., 2013. Interactions of the tooth and bone during
895development. J Dent Res 92, 1129-1135.
896Antony, J.M., Paquin, A., Nutt, S.L., Kaplan, D.R., Miller, F.D., 2011. Endogenous microglia
897regulate development of embryonic cortical precursor cells. J Neurosci Res 89, 286-298.
898Arai, F., Miyamoto, T., Ohneda, O., Inada, T., Sudo, T., Brasel, K., Miyata, T., Anderson, D.M.,
899Suda, T., 1999. Commitment and differentiation of osteoclast precursor cells by the sequential
900expression of c-Fms and receptor activator of nuclear factor kappaB (RANK) receptors. J Exp
901Med 190, 1741-1754.
902Biskobing, D.M., Fan, X., Rubin, J., 1995. Characterization of MCSF-induced proliferation and
903subsequent osteoclast formation in murine marrow culture. J Bone Miner Res 10, 1025-1032.
904Bolos, M., Perea, J.R., Terreros-Roncal, J., Pallas-Bazarra, N., Jurado-Arjona, J., Avila, J.,
905Llorens-Martin, M., 2018. Absence of microglial CX3CR1 impairs the synaptic integration of
906adult-born hippocampal granule neurons. Brain Behav Immun 68, 76-89.
907Bolton, J.L., Marinero, S., Hassanzadeh, T., Natesan, D., Le, D., Belliveau, C., Mason, S.N.,
908Auten, R.L., Bilbo, S.D., 2017. Gestational Exposure to Air Pollution Alters Cortical Volume,
909Microglial Morphology, and Microglia-Neuron Interactions in a Sex-Specific Manner. Front
910Synaptic Neurosci 9, 10.
911Bouyer, K., Simerly, R.B., 2013. Neonatal leptin exposure specifies innervation of
912presympathetic hypothalamic neurons and improves the metabolic status of leptin-deficient mice.
913J Neurosci 33, 840-851.
914Butovsky, O., Jedrychowski, M.P., Moore, C.S., Cialic, R., Lanser, A.J., Gabriely, G.,
915Koeglsperger, T., Dake, B., Wu, P.M., Doykan, C.E., Fanek, Z., Liu, L., Chen, Z., Rothstein,
916J.D., Ransohoff, R.M., Gygi, S.P., Antel, J.P., Weiner, H.L., 2014. Identification of a unique
917TGF-beta-dependent molecular and functional signature in microglia. Nat Neurosci 17, 131-143.
918Care, A.S., Diener, K.R., Jasper, M.J., Brown, H.M., Ingman, W.V., Robertson, S.A., 2013.
919Macrophages regulate corpus luteum development during embryo implantation in mice. J Clin
920Invest 123, 3472-3487.
921Chalmers, S.A., Wen, J., Shum, J., Doerner, J., Herlitz, L., Putterman, C., 2017. CSF-1R
922inhibition attenuates renal and neuropsychiatric disease in murine lupus. Clin Immunol 185, 100-
923108.
924Chang, K.T., Lin, Y.L., Lin, C.T., Tsai, M.J., Huang, W.C., Shih, Y.H., Lee, Y.Y., Cheng, H.,
925Huang, M.C., 2017. Data on the expression of leptin and leptin receptor in the dorsal root
926ganglion and spinal cord after preganglionic cervical root avulsion. Data Brief 15, 567-572.
927Cunningham, C.L., Martinez-Cerdeno, V., Noctor, S.C., 2013. Microglia regulate the number of
928neural precursor cells in the developing cerebral cortex. J Neurosci 33, 4216-4233.

929Elmore, M.R., Lee, R.J., West, B.L., Green, K.N., 2015. Characterizing newly repopulated
930microglia in the adult mouse: impacts on animal behavior, cell morphology, and
931neuroinflammation. PLoS One 10, e0122912.
932Elmore, M.R., Najafi, A.R., Koike, M.A., Dagher, N.N., Spangenberg, E.E., Rice, R.A.,
933Kitazawa, M., Matusow, B., Nguyen, H., West, B.L., Green, K.N., 2014. Colony-stimulating
934factor 1 receptor signaling is necessary for microglia viability, unmasking a microglia progenitor
935cell in the adult brain. Neuron 82, 380-397.
936Erblich, B., Zhu, L., Etgen, A.M., Dobrenis, K., Pollard, J.W., 2011. Absence of colony
937stimulation factor-1 receptor results in loss of microglia, disrupted brain development and
938olfactory deficits. PLoS One 6, e26317.
939Fourgeaud, L., Traves, P.G., Tufail, Y., Leal-Bailey, H., Lew, E.D., Burrola, P.G., Callaway, P.,
940Zagorska, A., Rothlin, C.V., Nimmerjahn, A., Lemke, G., 2016. TAM receptors regulate
941multiple features of microglial physiology. Nature 532, 240-244.
942Fuller, K., Owens, J.M., Jagger, C.J., Wilson, A., Moss, R., Chambers, T.J., 1993. Macrophage
943colony-stimulating factor stimulates survival and chemotactic behavior in isolated osteoclasts. J
944Exp Med 178, 1733-1744.
945Gao, Y., Vidal-Itriago, A., Kalsbeek, M.J., Layritz, C., Garcia-Caceres, C., Tom, R.Z.,
946Eichmann, T.O., Vaz, F.M., Houtkooper, R.H., van der Wel, N., Verhoeven, A.J., Yan, J.,
947Kalsbeek, A., Eckel, R.H., Hofmann, S.M., Yi, C.X., 2017a. Lipoprotein Lipase Maintains
948Microglial Innate Immunity in Obesity. Cell Rep 20, 3034-3042.
949Gao, Y., Vidal-Itriago, A., Milanova, I., Korpel, N.L., Kalsbeek, M.J., Tom, R.Z., Kalsbeek, A.,
950Hofmann, S.M., Yi, C.X., 2017b. Deficiency of leptin receptor in myeloid cells disrupts
951hypothalamic metabolic circuits and causes body weight increase. Mol Metab.
952Ginhoux, F., Greter, M., Leboeuf, M., Nandi, S., See, P., Gokhan, S., Mehler, M.F., Conway,
953S.J., Ng, L.G., Stanley, E.R., Samokhvalov, I.M., Merad, M., 2010. Fate mapping analysis
954reveals that adult microglia derive from primitive macrophages. Science 330, 841-845.
955Giulian, D., Chen, J., Ingeman, J.E., George, J.K., Noponen, M., 1989. The role of mononuclear
956phagocytes in wound healing after traumatic injury to adult mammalian brain. J Neurosci 9,
9574416-4429.
958Gomez Perdiguero, E., Klapproth, K., Schulz, C., Busch, K., Azzoni, E., Crozet, L., Garner, H.,
959Trouillet, C., de Bruijn, M.F., Geissmann, F., Rodewald, H.R., 2015. Tissue-resident
960macrophages originate from yolk-sac-derived erythro-myeloid progenitors. Nature 518, 547-551.
961Hanamsagar, R., Alter, M.D., Block, C.S., Sullivan, H., Bolton, J.L., Bilbo, S.D., 2018.
962Generation of a microglial developmental index in mice and in humans reveals a sex difference
963in maturation and immune reactivity. Glia 66, 460.
964Heinrich, J., Bsoul, S., Barnes, J., Woodruff, K., Abboud, S., 2005. CSF-1, RANKL and OPG
965regulate osteoclastogenesis during murine tooth eruption. Arch Oral Biol 50, 897-908.
966Hoeffel, G., Chen, J., Lavin, Y., Low, D., Almeida, F.F., See, P., Beaudin, A.E., Lum, J., Low,
967I., Forsberg, E.C., Poidinger, M., Zolezzi, F., Larbi, A., Ng, L.G., Chan, J.K., Greter, M., Becher,
968B., Samokhvalov, I.M., Merad, M., Ginhoux, F., 2015. C-Myb(+) erythro-myeloid progenitor-
969derived fetal monocytes give rise to adult tissue-resident macrophages. Immunity 42, 665-678.
970Huang, Y., Xu, Z., Xiong, S., Sun, F., Qin, G., Hu, G., Wang, J., Zhao, L., Liang, Y.X., Wu, T.,
971Lu, Z., Humayun, M.S., So, K.F., Pan, Y., Li, N., Yuan, T.F., Rao, Y., Peng, B., 2018.
972Repopulated microglia are solely derived from the proliferation of residual microglia after acute
973depletion. Nat Neurosci.

974Jin, W.N., Shi, S.X., Li, Z., Li, M., Wood, K., Gonzales, R.J., Liu, Q., 2017. Depletion of
975microglia exacerbates postischemic inflammation and brain injury. J Cereb Blood Flow Metab
97637, 2224-2236.
977Kierdorf, K., Erny, D., Goldmann, T., Sander, V., Schulz, C., Perdiguero, E.G., Wieghofer, P.,
978Heinrich, A., Riemke, P., Holscher, C., Muller, D.N., Luckow, B., Brocker, T., Debowski, K.,
979Fritz, G., Opdenakker, G., Diefenbach, A., Biber, K., Heikenwalder, M., Geissmann, F.,
980Rosenbauer, F., Prinz, M., 2013. Microglia emerge from erythromyeloid precursors via Pu.1- and
981Irf8-dependent pathways. Nat Neurosci 16, 273-280.
982Li, J., Chen, K., Zhu, L., Pollard, J.W., 2006. Conditional deletion of the colony stimulating
983factor-1 receptor (c-fms proto-oncogene) in mice. Genesis 44, 328-335.
984Li, M., Li, Z., Ren, H., Jin, W.N., Wood, K., Liu, Q., Sheth, K.N., Shi, F.D., 2017. Colony
985stimulating factor 1 receptor inhibition eliminates microglia and attenuates brain injury after
986intracerebral hemorrhage. J Cereb Blood Flow Metab 37, 2383-2395.
987Marsters, C.M., Rosin, J.M., Thornton, H.F., Aslanpour, S., Klenin, N., Wilkinson, G.,
988Schuurmans, C., Pittman, Q.J., Kurrasch, D.M., 2016. Oligodendrocyte development in the
989embryonic tuberal hypothalamus and the influence of Ascl1. Neural Dev 11, 20.
990Matcovitch-Natan, O., Winter, D.R., Giladi, A., Vargas Aguilar, S., Spinrad, A., Sarrazin, S.,
991Ben-Yehuda, H., David, E., Zelada Gonzalez, F., Perrin, P., Keren-Shaul, H., Gury, M., Lara-
992Astaiso, D., Thaiss, C.A., Cohen, M., Bahar Halpern, K., Baruch, K., Deczkowska, A., Lorenzo-
993Vivas, E., Itzkovitz, S., Elinav, E., Sieweke, M.H., Schwartz, M., Amit, I., 2016. Microglia
994development follows a stepwise program to regulate brain homeostasis. Science 353, aad8670.
995Mildner, A., Schmidt, H., Nitsche, M., Merkler, D., Hanisch, U.K., Mack, M., Heikenwalder,
996M., Bruck, W., Priller, J., Prinz, M., 2007. Microglia in the adult brain arise from Ly-
9976ChiCCR2+ monocytes only under defined host conditions. Nat Neurosci 10, 1544-1553.
998Miyamoto, A., Wake, H., Ishikawa, A.W., Eto, K., Shibata, K., Murakoshi, H., Koizumi, S.,
999Moorhouse, A.J., Yoshimura, Y., Nabekura, J., 2016. Microglia contact induces synapse
1000formation in developing somatosensory cortex. Nat Commun 7, 12540.
1001Nelson, L.H., Warden, S., Lenz, K.M., 2017. Sex differences in microglial phagocytosis in the
1002neonatal hippocampus. Brain Behav Immun 64, 11-22.
1003Nesan, D., Kurrasch, D.M., 2016. Genetic programs of the developing tuberal hypothalamus and
1004potential mechanisms of their disruption by environmental factors. Mol Cell Endocrinol 438, 3-
100517.
1006Paolicelli, R.C., Bolasco, G., Pagani, F., Maggi, L., Scianni, M., Panzanelli, P., Giustetto, M.,
1007Ferreira, T.A., Guiducci, E., Dumas, L., Ragozzino, D., Gross, C.T., 2011. Synaptic pruning by
1008microglia is necessary for normal brain development. Science 333, 1456-1458.
1009Parkhurst, C.N., Yang, G., Ninan, I., Savas, J.N., Yates, J.R., 3rd, Lafaille, J.J., Hempstead, B.L.,
1010Littman, D.R., Gan, W.B., 2013. Microglia promote learning-dependent synapse formation
1011through brain-derived neurotrophic factor. Cell 155, 1596-1609.
1012Patel, S., Player, M.R., 2009. Colony-stimulating factor-1 receptor inhibitors for the treatment of
1013cancer and inflammatory disease. Curr Top Med Chem 9, 599-610.
1014Rice, R.A., Pham, J., Lee, R.J., Najafi, A.R., West, B.L., Green, K.N., 2017. Microglial
1015repopulation resolves inflammation and promotes brain recovery after injury. Glia 65, 931-944.
1016Sakai, H., Chen, Y., Itokawa, T., Yu, K.P., Zhu, M.L., Insogna, K., 2006. Activated c-Fms
1017recruits Vav and Rac during CSF-1-induced cytoskeletal remodeling and spreading in
1018osteoclasts. Bone 39, 1290-1301.

1019Schafer, D.P., Lehrman, E.K., Kautzman, A.G., Koyama, R., Mardinly, A.R., Yamasaki, R.,
1020Ransohoff, R.M., Greenberg, M.E., Barres, B.A., Stevens, B., 2012. Microglia sculpt postnatal
1021neural circuits in an activity and complement-dependent manner. Neuron 74, 691-705.
1022Schwarz, J.M., Sholar, P.W., Bilbo, S.D., 2012. Sex differences in microglial colonization of the
1023developing rat brain. J Neurochem 120, 948-963.
1024Spangenberg, E.E., Lee, R.J., Najafi, A.R., Rice, R.A., Elmore, M.R., Blurton-Jones, M., West,
1025B.L., Green, K.N., 2016. Eliminating microglia in Alzheimer's mice prevents neuronal loss
1026without modulating amyloid-beta pathology. Brain 139, 1265-1281.
1027Squarzoni, P., Oller, G., Hoeffel, G., Pont-Lezica, L., Rostaing, P., Low, D., Bessis, A.,
1028Ginhoux, F., Garel, S., 2014. Microglia modulate wiring of the embryonic forebrain. Cell Rep 8,
10291271-1279.
1030Szalay, G., Martinecz, B., Lenart, N., Kornyei, Z., Orsolits, B., Judak, L., Csaszar, E., Fekete, R.,
1031West, B.L., Katona, G., Rozsa, B., Denes, A., 2016. Microglia protect against brain injury and
1032their selective elimination dysregulates neuronal network activity after stroke. Nat Commun 7,
103311499.
1034Thion, M.S., Low, D., Silvin, A., Chen, J., Grisel, P., Schulte-Schrepping, J., Blecher, R., Ulas,
1035T., Squarzoni, P., Hoeffel, G., Coulpier, F., Siopi, E., David, F.S., Scholz, C., Shihui, F., Lum, J.,
1036Amoyo, A.A., Larbi, A., Poidinger, M., Buttgereit, A., Lledo, P.M., Greter, M., Chan, J.K.Y.,
1037Amit, I., Beyer, M., Schultze, J.L., Schlitzer, A., Pettersson, S., Ginhoux, F., Garel, S., 2018.
1038Microbiome Influences Prenatal and Adult Microglia in a Sex-Specific Manner. Cell 172, 500-
1039516 e516.
1040Tremblay, M.E., Lowery, R.L., Majewska, A.K., 2010. Microglial interactions with synapses are
1041modulated by visual experience. PLoS Biol 8, e1000527.
1042Urabe, H., Kojima, H., Chan, L., Terashima, T., Ogawa, N., Katagi, M., Fujino, K., Kumagai,
1043A., Kawai, H., Asakawa, A., Inui, A., Yasuda, H., Eguchi, Y., Oka, K., Maegawa, H.,
1044Kashiwagi, A., Kimura, H., 2013. Haematopoietic cells produce BDNF and regulate appetite
1045upon migration to the hypothalamus. Nat Commun 4, 1526.
1046Valdearcos, M., Douglass, J.D., Robblee, M.M., Dorfman, M.D., Stifler, D.R., Bennett, M.L.,
1047Gerritse, I., Fasnacht, R., Barres, B.A., Thaler, J.P., Koliwad, S.K., 2017. Microglial
1048Inflammatory Signaling Orchestrates the Hypothalamic Immune Response to Dietary Excess and
1049Mediates Obesity Susceptibility. Cell Metab 26, 185-197 e183.
1050VanRyzin, J.W., Yu, S.J., Perez-Pouchoulen, M., McCarthy, M.M., 2016. Temporary Depletion
1051of Microglia during the Early Postnatal Period Induces Lasting Sex-Dependent and Sex-
1052Independent Effects on Behavior in Rats. eNeuro 3.
1053Wang, J., Xie, D., Liu, M., Gong, Y., Shi, X., Wei, J.Y., Quan, S., 2016. [Uterine macrophages
1054affect embryo implantation via regulating vascular endothelial growth factor A in mice]. Nan
1055Fang Yi Ke Da Xue Xue Bao 36, 909-914.
1056Yi, C.X., Walter, M., Gao, Y., Pitra, S., Legutko, B., Kalin, S., Layritz, C., Garcia-Caceres, C.,
1057Bielohuby, M., Bidlingmaier, M., Woods, S.C., Ghanem, A., Conzelmann, K.K., Stern, J.E.,
1058Jastroch, M., Tschop, M.H., 2017. TNFalpha drives mitochondrial stress in POMC neurons in
1059obesity. Nat Commun 8, 15143.
1060Zhan, C., Zhou, J., Feng, Q., Zhang, J.E., Lin, S., Bao, J., Wu, P., Luo, M., 2013. Acute and
1061long-term suppression of feeding behavior by POMC neurons in the brainstem and

1062
1063
hypothalamus, respectively. J Neurosci 33, 3624-3632.

1065Highlights
1066• The CSF1R inhibitor PLX5622 depleted ~99% of fetal microglia by E15.5
1067• Embryonic microglia depletion resulted in a decreased litter size
1068• Elimination of embryonic microglia resulted in a decrease in POMC neurons
1069• Accelerated weight gain is observed in microglia depleted animals

1070
1071
1072
• Depletion of microglia during embryogenesis had sex-specific effects on behaviour

PLX5622

45

Table 1