Aberrant Maturation of the Uncinate Fasciculus Follows Exposure to Unpredictable Patterns of Maternal Signals

Research Articles: Systems/Circuits Aberrant Maturation of the Uncinate Fasciculus Follows Exposure to Unpredictable Patterns of Maternal Signals https://doi.org/10.1523/JNEUROSCI.0374-20.2020

Cite as: J. Neurosci 2020; 10.1523/JNEUROSCI.0374-20.2020 Received: 17 February 2020 Revised: 25 November 2020 Accepted: 2 December 2020

This Early Release article has been peer-reviewed and accepted, but has not been through the composition and copyediting processes. The final version may differ slightly in style or formatting and will contain links to any extended data.

Alerts: Sign up at www.jneurosci.org/alerts to receive customized email alerts when the fully formatted version of this article is published.

1 Manuscript submission for Journal of Neuroscience

2 Aberrant Maturation of the Uncinate Fasciculus Follows Exposure to Unpredictable 3 Patterns of Maternal Signals

5 Abbreviated Title: Uncinate and Patterns of Maternal Sensory Signals

6 Steven J. Grangera,b, Laura M. Glynnc, Curt A. Sandmand, Steven L. Smalle, Andre Obenausf, 7 David B. Keatord, Tallie Z. Barama,f,g, Hal Sternh, Michael A. Yassaa,b,d ᅒ , Elysia Poggi Davisd,i 8 ᅒ

9 a Center for the Neurobiology of Learning and Memory, University of California, Irvine 92697 10 b Department of Neurobiology and Behavior, University of California, Irvine 92697 11 c Department of Psychology, Chapman University, Orange, CA 92866 12 d Department of Psychiatry and Human Behavior, University of California, Irvine, CA 92697 13 e School of Behavioral and Brain Sciences, University of Texas at Dallas, Richardson, TX, 14 75080 15 f Department of Pediatrics, University of California, Irvine, CA 92697 16 g Department of Anatomy and Neurobiology, University of California, Irvine, CA 92697 17 h Department of Statistics, University of California, Irvine 92697 18 i Department of Psychology, University of Denver, Denver, CO 80208 19 20 ᅒ Corresponding Authors

21 Michael A. Yassa Elysia Poggi Davis 22 UC Irvine University of Denver 23 [email protected] [email protected] 24 25 26 This PDF file includes: 27 43 pages 28 4 Figures, 3 tables 29 Number of words for abstract: 245 30 Number of words for introduction: 648 31 Number of words for discussion: 1,403 32 33

34 Acknowledgments: This work was supported by the National Institute of Health (Grants 35 P50MH096889 – PI: Baram; R01MH102392 – PI: Yassa; R03MH086062 – PI: Davis; NS-41298 36 – PI: Sandman, HD-51852 – PI: Sandman, and HD-28413 – PI: Sandman).

37 Conflict of Interest: The authors declare no conflicts of interest.

Page 1 of 29

39 Abstract

40 Across species, unpredictable patterns of maternal behavior are emerging as novel predictors of 41 aberrant cognitive and emotional outcomes later in life. In animal models, exposure to 42 unpredictable patterns of maternal behavior alters brain circuit maturation and cognitive and 43 emotional outcomes. However, whether exposure to such signals in humans alters the 44 development of brain pathways is unknown. In mother-child dyads we tested the hypothesis that 45 exposure to more unpredictable maternal signals in infancy associates with aberrant maturation 46 of corticolimbic pathways. We focused on the uncinate fasciculus, the primary fiber bundle 47 connecting the amygdala to the orbitofrontal cortex and a key component of the medial temporal 48 lobe - prefrontal cortex circuit. Infant exposure to unpredictable maternal sensory signals was 49 assessed at 6 and 12 months. Using high angular resolution diffusion imaging, we quantified the 50 integrity of the uncinate fasciculus using generalized fractional anisotropy (GFA). Higher 51 maternal unpredictability during infancy presaged greater uncinate fasciculus GFA in children 52 aged 9-11 (n=69, 29 female). In contrast to the uncinate, GFA of a second corticolimbic 53 projection, the hippocampal cingulum was not associated with maternal unpredictability. 54 Addressing the overall functional significance of the uncinate and cingulum relationships, we 55 found that the resulting imbalance of MTL-PFC connectivity partially mediated the association 56 between unpredictable maternal sensory signals and impaired episodic memory function. These 57 results suggest that unbalanced maturation of corticolimbic circuits are a mechanism by which 58 early unpredictable sensory signals may impact cognition later in life.

Page 2 of 29

67 Significance Statement

68 Our prior work across species demonstrated that unpredictable patterns of maternal care are 69 associated with compromised memory function. However, the neurobiological mechanisms by 70 which this occurs in humans remain unknown. Here, we identify an association of exposure to 71 unpredictable patterns of maternal sensory signals with the integrity of corticolimbic circuits 72 involved in emotion and cognition using state-of-the-art diffusion imaging techniques and 73 analyses. We find that exposure to early unpredictability is associated with higher integrity of the 74 uncinate fasciculus with no effect on a second corticolimbic pathway, the cingulum. The 75 resulting imbalance of corticolimbic circuit development is a novel mediator of the association 76 between unpredictable patterns of maternal care and poorer episodic memory.

90 Introduction

Page 3 of 29

91 Experiences occurring during sensitive periods in early life are powerful factors influencing brain 92 development and cognition (Hasler et al., 2004; Andersen et al., 2008; Short & Baram, 2019). 93 Although it is clear that the quality of maternal care affects the risk for offspring psychological 94 disorders and neurobiological changes later in life (Glynn & Baram, 2019; Bowlby 1950, Lebel et 95 al., 2015), less is known about how patterns of maternal behavior impact human development 96 (Baram et al., 2012; Davis et al., 2017). Recent work has indicated that unpredictable patterns 97 of maternal sensory signals during infancy are a novel contributor to impaired cognitive and 98 emotional functions later in life (Chen & Baram, 2016, Walker et al., 2017, Glynn and Baram, 99 2019; Noraña et al., 2020; Molet et al., 2016b).

100 We previously reported in both humans and rodents that exposure to greater unpredictable 101 sensory signals during infancy is associated with memory deficits later in life (Davis et al., 2017; 102 Molet et al., 2016a; Ivy et al., 2008; Ivy et al., 2010, Bath et al., 2017). In rodents, pup exposure 103 to unpredictable maternal behavior leads to abnormal maturation of brain circuits involved with 104 memory (Brunson 2005; Guadagno et al., 2016, Walker et al., 2017; Molet et al., 2016a, Ivy et 105 al., 2010). Whether the same cortico-limbic circuits are impacted by unpredictable maternal 106 sensory signals in humans remains unknown.

107 Across species, aberrant maturation of amygdala-prefrontal circuits is well recognized as a key 108 outcome of exposure to stress (Burgos-Robles et al., 2017; Chen & Baram, 2016, Burghy et al., 109 2012, Tottenham & Galván, 2016; Gee et al., 2013). Accelerated maturation of amygdala- 110 prefrontal cortex functional connectivity may be an adaptive response to early life adversity, 111 which reprioritizes developmental goals to match the demands of adverse early life 112 environments (Gee et al., 2013). In humans, the predominant anatomical connection between 113 these two regions is the uncinate fasciculus (Ebeling and Von Cramon, 1992). The uncinate is 114 thought to be important for episodic memory and has recently been hypothesized to play an 115 important role in adjudicating among competing memory representations during retrieval (Alm et 116 al., 2016; Von der Heide et al., 2013). Abnormal development of the uncinate could lead to 117 impairments in cognitive and emotional functions (Vilgis et al., 2017; Carballedo et al., 2012; 118 Yang et al., 2017; Bhatia et al., 2018; Zhang et al., 2012). Abnormal development of the 119 uncinate has been associated with early life adversity including a history of institutionalization 120 during childhood and paternal depressive symptoms (Govindan et al., 2010; Eluvathingal et al., 121 2006, Hanson et al., 2015; Ho et al., 2017; Marroun et al., 2018).

Page 4 of 29

122 Here, we evaluated whether exposure to unpredictable maternal signals in infancy is associated 123 with uncinate integrity, measured between the ages of 9 and 11. To determine whether the 124 observed effects were specific to the uncinate or are general features of corticolimbic white 125 matter, we conducted similar analyses in the hippocampal cingulum, which shares the 126 uncinate’s frontotemporal connectivity but is spatially distinct (Bubb et al., 2018). As the 127 functional output of brain circuits is a combinatorial sum of the activity of each of their 128 components (Redish and Gordon, 2017), we also conducted analyses using the ratio of 129 uncinate to hippocampal cingulum connectivity as a measure of imbalance across these two 130 pathways.

131 To determine the behavioral relevance of changes in white matter connectivity associated with 132 unpredictable maternal sensory signals, we focused on episodic memory function which we 133 assessed by using the Mnemonic Similarity Task (MST), a well-validated task (Stark, Kirwan 134 and Stark, 2019) that is sensitive to the hippocampus’ ability to discriminate among highly 135 similar experiences (i.e. pattern separation), which is thought to be a core facet of episodic 136 memory (Marr, 1971; Yassa and Stark, 2011; Leal and Yassa, 2018). This task was chosen as 137 we have previously shown across species that greater unpredictability of maternal sensory 138 signals in infancy is associated with differences in memory function (Davis et al., 2017). Further, 139 MST performance is dependent on successful adjudication among competing memory 140 representations during retrieval, which is thought to be a core function of the uncinate fasciculus 141 (Von Der Heide et al., 2013; Alm et al. 2016).

142

143 Materials and Methods

144 Participants

145 A sample of 73 mother/offspring dyads (n=30 female) from a prospective, longitudinal, prenatal 146 cohort participated. Initial maternal recruitment criteria included: English-speaking, over 18 147 years old, non-smokers, no evidence of drug/alcohol use, singleton pregnancy. Unpredictable 148 maternal sensory signals were assessed when the infants were 6 and 12 months of age and the 149 children participated in a single MRI scan between 9 and 11 years of age. Demographic 150 information for the full sample of participants is summarized in Table 1.

151 Measurement of Unpredictability of Maternal Sensory Signals

Page 5 of 29

152 Coding maternal sensory signals

153 Unpredictable maternal sensory signals were coded from a free play dyadic interaction between 154 a mother and her child as described in Davis et al., 2019. Briefly, mothers were video-recorded 155 interacting with their child in a semi-structured 10-minute play episode at 6 and 12 months of 156 age. Mothers were given a standard set of age-appropriate toys and instructed to interact with 157 their infants naturally as if they were at home. Behaviors from the mother that provided auditory, 158 visual, or tactile sensory signals to the child were coded on a moment-to-moment basis from the 159 video recordings using The Observer XT 11 (Noldus). Auditory signals included all maternal 160 vocalizations, visual signals included maternal manipulation of toy or object while the infant was 161 visually attending, and tactile signals involved all instances of physical contact 162 (holding/touching) initiated by the mother. Coders were blind to all other information on study 163 participants. Interrater reliability was calculated based on 20% of videos and averaged 89% 164 (Davis et al., 2017).

165 Quantifying Unpredictability of Maternal Sensory Signals

166 Unpredictability of maternal sensory signals were quantified by calculating the entropy rate of 167 the sequences of visual, tactile and auditory maternal signals assessed during the two 10- 168 minute naturalistic observations periods described above. The entropy rate is determined as 169 follows. We first identify all transitions between visual, tactile, and auditory maternal signals, 170 accounting for all possible combinations of the different types of sensory signals. As an 171 example, if a mother was speaking (auditory) to her child and then shows her a toy (visual) 172 while labeling it (auditory), this would be classified as a transition from auditory only to auditory 173 and visual. The resulting transitions are then modeled as changes in the state of a discrete-state 174 first-order Markov process and the entropy rate of that process calculated as described in Davis 175 et al. (2017). We considered alternative strategies for calculating the entropy rate (e.g., based 176 on other Markov chain models; 2nd order and 3rd order) as well as a non-parametric approach 177 based on theoretical results relating data compression to unpredictability of maternal sensory 178 signals and found that results were highly consistent (Spearman’s rank correlations for resulting 179 unpredictability measures ranging from 0.91 to 0.98) (Vegetabile et al., 2019). The entropy rate 180 measurement can be thought of as capturing how random or uncertain a mother’s next behavior 181 would seem to someone making a guess based on previous observed behavior. Entropy rate 182 can range from 0 (transitions are perfectly predictable, e.g. speech is always followed by touch) 183 to a maximum of 2.807 (unpredictable, the next observed behavior among the appears to occur

Page 6 of 29

184 at random). The resulting unpredictability rates were calculated separately for each visit (6 and 185 12 month) and averaged. A comprehensive description of the calculation regarding maternal 186 unpredictability and the entropy rate quantification is described in Davis et al., 2019 and an R 187 package for the calculation of entropy is available at https://contecenter.uci.edu/shared- 188 resources.

189 Episodic memory: Mnemonic Similarity Task

190 The Mnemonic Similarity Task (MST) was administered on average 99 (r12.13) days from the 191 date of the MRI scan. The MST consists of two phases: an incidental encoding phase and a 192 surprise memory test. During the incidental encoding phase, children are presented with a 193 series of objects appearing on the screen for 2.0 seconds (ISI 0.5 seconds). At this time, the 194 children were asked to rate the objects as being found “Indoor” or “Outdoor” via computer 195 keyboard presses. A total of 64 images were presented during the testing phase. During the 196 retrieval (testing) portion, children were presented with 32 target images (identical images that 197 appeared during encoding), 32 foil images (images that did not appear during encoding), and 32 198 lure images (variants of images presented during encoding). During test, children were asked to 199 indicate if the presented object was “Old” (they had seen the image before), “Similar” (the 200 images were slightly different than images presented during test) or “New” (images were novel 201 and not seen during test). Lures items were varied by similarity ratings on a scale from 1 to 5 202 with 1 being the most similar (bin 1: most difficult to discriminate) and 5 being the least similar 203 (bin 5: easier to discriminate). Lure items were distributed across bin similarity in the following 204 way: bin 1 (6 items), bin 2 (7 items), bin 3 (7 items), bin 4 (6 items), bin 5 (6 items). As 205 measures of performance, we calculated a Lure Discrimination Index (LDI) which is defined as 206 the probability of indicating “Similar” given a Lure minus the probability of indicating “Similar” 207 given a Foil. In this fashion, this measure controls for response bias (indicating “Similar” for 208 every item). The LDI is used as the index of episodic memory. We analyzed the low similarity 209 conditions (average L4 and L5) as previous research using the MST demonstrated that children 210 this age exhibit chance level performance on the more difficult high similarity conditions (Rollins 211 & Cloude, 2018). A significant advantage of the MST task is the concurrent ability to assess 212 traditional recognition memory by evaluating recognition of target items vs. novel foils. We used 213 performance on this recognition condition as a control to determine the specificity of the 214 relationship between hippocampal pattern separation and unpredictable maternal sensory 215 signals.

Page 7 of 29

216 Measurement of Possible Confounding Variables

217 To assess whether associations with unpredictable maternal sensory signals are influenced by 218 confounding factors we evaluated maternal mental health, quality of maternal care, and 219 household income. These variables are further described below.

220 Maternal Depression/Anxiety

221 Symptoms of maternal depression and anxiety were assessed at the two postnatal visits (at 6 222 and 12 months). The Edinburgh Postnatal Depression Scale (EPDS; Cox et al., 1987) and State 223 Anxiety Inventory (STAI; Spielberger et al., 1983) were administered. Scores were standardized 224 and then averaged as an index of maternal mental health (Davis et al., 2019).

225 Quality of maternal care: Maternal Sensitivity

226 In order to determine if unpredictability of maternal signals was associated with brain outcomes 227 beyond more standard measures of quality of maternal care we evaluated a measure of 228 maternal sensitivity during the same observation period as the quantification of maternal 229 unpredictability. Maternal sensitivity was coded using a protocol developed for the National 230 Institute of Child Health and Human Development (NICHD) Study of Early Child Care and Youth 231 Development (NICHD) Study of Early Child Care and Youth Development (NICHD, 1999). 232 Maternal behavior was evaluated for sensitivity to non-distress, intrusiveness (reverse scored), 233 and positive regard (1 = not at all characteristic to 4 = highly characteristic). A sum of ratings of 234 sensitivity were calculated and used according to NICHD (1999).

235 Income-to-Needs Ratio (INR)

236 INR was determined by comparing household income to the federal poverty line for a given year 237 and a given household size. Ratios below 1.00 indicate that household income is below the 238 federal poverty line, in contrast a ratio of 1.00 or greater indicates income above the federal 239 poverty level (Grieger et al., 2009).

240 Diffusion Imaging Protocol and Processing

241 MRI were collected on a 3.0 Tesla Philips Achieva scanner at the Neuroscience Imaging Center 242 at the University of California, Irvine using a 32-channel head coil. The diffusion weighted 243 imaging scheme consisted of 6 separate runs of 11 volumes (including one b=0 volume) each.

Page 8 of 29

244 The first two runs were collected using a b- value of 500 s/mm2, the third and fourth runs were 245 collected using a b-value of 1,000 s/mm2, the fifth and sixth runs were collected using a b-value 246 of 2000 s/mm2. Thus, the final scan consisted of a multi-shell sequence with 60 noncollinear 247 directions and 6 B0 volumes. The following scanning parameters were used: direction of 248 acquisition was anterior to posterior, the field of view was 216 (ap), 216 (fh), 151.2 (rl), the scan 249 resolution was 108 (x), 127 (y), number of slices = 84, voxel size 1.69x1.68x1.68mm, TR = 250 ~11502.76, TE = 89. Raw data were corrected for motion and eddy currents using FSLs eddy 251 program (Andersson & Sotiropoulos, 2016). Data were analyzed using DSI-Studio (http://dsi- 252 studio.labsolver.org; July 26, 2017 build). Corrected data were then reconstructed using DSI- 253 Studio’s Q-spin Diffeomorphic Reconstruction (QSDR) function which uses a powerful 254 diffeomorphic algorithm to warp model-free orientation distribution functions (ODFs) to Montreal 255 Neurological Institute (MNI) space (Yeh & Tseng, 2011). ODF’s were reconstructed with the 256 default diffusion-sampling-length ratio of 1.25 which was sufficient to model crossing fibers at 257 the intersection of the corticospinal tract and corpus callosum. This value was additionally 258 validated by varying the diffusion sampling length ratio at intervals of 0.2 and visually examining 259 this intersection using Generalized q-sampling Imaging (GQI). Other reconstruction parameters 260 included registration method: norm 7-9-7, 8-fold ODF tessellation, number of fibers resolved: 5. 261 Output resolution was increased to 1mm. Quality control for fit to the template space was 262 implemented by setting a criterion on R2 at >0.6. Four subjects of the total 73 were eliminated 263 based on these criteria leaving a final sample size of 69 subjects (29 female). Subject head 264 motion was assessed by the eddy_movement_rms file exported from eddy (movement relative 265 to the previous volume) and included as a nuisance regressor. Subject T1-weighted images 266 were used for the quantification of intracranial brain volume. Structural MPRAGE scans were 267 collected at the time of the diffusion scan with the following parameters: 1mm isotropic 268 resolution, 208 sagittal slices, field of view = 208 x 256 x 256mm, flip angle = 8°, TR/TE = 8/3.7 269 (ms), matrix size = 256 x 210 (mm). Intracranial volume was calculated using Freesurfer version 270 6.0 and used to rule out if differences in GFA were due to intracranial volume.

271 Region-of-Interest (ROI) Based Approach

272 Masks of ROIs were obtained through the Johns Hopkins white matter atlas (in MNI space) 273 which is available within DSI Studio. Regions included limbic white matter regions, namely the 274 uncus of the uncinate fasciculus and the hippocampal cingulum bilaterally. Averaged 275 generalized fractional anisotropy (GFA) values were extracted for each ROI and averaged 276 across hemispheres. GFA is one of several model-free diffusion measures and is known to

Page 9 of 29

277 correlate with fractional anisotropy of the tensor model. GFA was used to assess structural 278 integrity of complex tissues in a clinical setting particularly when there are heterogeneous fiber 279 tissues (Koh et al., 2018; Yamada et al., 2018) and has been used in studies of affective 280 disorders (Lo et al., 2017; Chiang et al., 2016). However, it should be noted that the accuracy of 281 GFA depends on the type of tissue being evaluated as well as the b-value at which the raw data 282 are collected (Tuch et al., 2002; Tuch, 2004; Fritzche et al., 2010; Gorczewski et al., 2008).

283 Statistical Analysis

284 Covariates were identified based on theoretical links to either maternal behavior or child 285 outcomes. Covariates included in all analyses were child sex, maternal depressive and anxiety 286 symptoms, maternal sensitivity, and income-to-needs ratio. Correlational analysis was 287 conducted using Prism GraphPad 7 using two tailed tests of Pearson correlation coefficients. 288 Multiple linear regression models were implemented using R-Studio to assess the association 289 between unpredictable maternal sensory signals and integrity of white matter tract (uncinate 290 fasciculus and hippocampal cingulum) after consideration of covariates. Mediation analyses 291 were conducted using the “mediation” package in R-Studio. Each mediation model was tested 292 using bias corrected and accelerated procedures with 10000 simulations. One case of missing 293 data was imputed for the measurement of maternal depression and anxiety using all possible 294 covariates and uncinate fasciculus GFA (Sinharay et al., 2001). Three cases of missing data 295 were imputed for income to needs ratio in a similar manner using all possible covariates 296 (including maternal depression/anxiety and uncinate fasciculus GFA).

297 298 Results 299 300 Exposure to unpredictable maternal sensory signals during infancy is associated with 301 higher uncinate fasciculus integrity at age 9-11 302 303 First, we tested the hypothesis that exposure to unpredictable maternal sensory signals in 304 infancy is associated with the integrity of the uncinate fasciculus in 9-11year old children using 305 GFA as our primary outcome measure. Exposure to unpredictable maternal sensory signals 306 during infancy predicted greater UF GFA (r = 0.37, P = 0.0017; Figure 2a). Multiple regressions 307 with covariates (maternal depression/anxiety, maternal sensitivity, income-to-needs ratio, and 308 sex) did not change the association between unpredictable maternal sensory signals and

Page 10 of 29

309 uncinate fasciculus GFA (β = 0.26, R2 = 0.35, P = 0.020; Table 2). Sex was significant as an 310 independent predictor of UF GFA. However, the sex by unpredictability interaction was 311 nonsignificant.

312 In order to rule out the possibility that age at scan and in-scanner head motion (derived from 313 eddy) were responsible for differences in UF GFA, we included them as nuisance regressors in 314 an additional model with all previously mentioned covariates. The association between 315 unpredictable maternal sensory signals and UF GFA (β = 0.26, R2 = 0.38, P = 0.017) as 316 remained significant after accounting for age and head motion. Additionally, neither age, subject 317 head motion, nor their interaction predicted UF GFA. Additionally, intracranial volume did not 318 predict UF GFA (r = -0.023, P = 0.85). Finally, including age, and intracranial volume (calculated 319 from T1-weighted images) as covariates in our analyses did not alter the statistical significance 320 of the relationships between unpredictable maternal sensory signals and UF GFA.

321 Because the functional output of brain circuits is a combinatorial sum of the activity of each of 322 their components, we conducted similar analyses in the hippocampal cingulum bundle, a 323 second component of the corticolimbic circuit that connects the connects the MTL with the PFC 324 via a posterior pathway through the retrosplenial cortex (Bubb et al. 2018). While unpredictable 325 maternal sensory signals did not significantly predict hippocampal cingulum GFA (r = -0.0026, P 326 = 0.98, Figure 2b), the sex by unpredictability interaction was marginally significant (Table 3; β = 327 -1.22, R2 = 0.052, P = 0.086). The directionality of the associations identified in the UF and 328 marginal interaction with sex in the cingulum prompted us to ask whether the two findings are 329 related to an imbalance in the integrity or maturation of these two circuits. We calculated a 330 simple ratio measure of UF:Cingulum GFA to capture this imbalance. We found a significant 331 association between unpredictability of maternal sensory signals and the ratio of UF GFA to 332 hippocampal cingulum GFA (r = 0.28, P = 0.018; Figure 3a).

333 Corticolimbic circuit imbalance is associated with impaired episodic memory 334 performance

335 Based on prior work across species (Davis et al., 2017), we hypothesized that unpredictability of 336 maternal sensory signals would be associated with worse performance on cognitive tasks. In 337 the current study, we used the LDI to assess episodic memory performance. Interestingly, we 338 observed that unpredictable maternal sensory signals did not significantly predict impaired 339 episodic memory performance (r = 0.022, P = 0.86). We then probed whether the observed

Page 11 of 29

340 circuit changes related to unpredictable maternal sensory signals were associated with episodic 341 memory performance.

342 UF GFA was not significantly associated with episodic memory performance assessed with the 343 LDI (r = -0.091, P = 0.46). However, we found that decreased hippocampal cingulum GFA was 344 marginally associated with lower LDI (r = 0.22, P = 0.072). A higher UF:Cingulum GFA ratio was 345 also marginally associated with lower LDI (r = -0.23, P = 0.058; Figure 3b). The interaction 346 between sex and the UF:Cingulum ratio in association with LDI did not reach statistical 347 significance (β =1.28, P = 0.13).

348 We observed a significant indirect association through which higher levels of unpredictable 349 maternal sensory signals was associated with a higher UF:Cingulum GFA ratio which, in turn, 350 was associated with worse LDI. In separate mediation models, neither UF GFA (indirect effect = 351 -0.061, C.I. = -0.24 to 0.097, P = 0.44) nor cingulum GFA (indirect effect = -0.0008, C.I. = -0.088 352 to 0.11, P = 0.99) were significant mediators of the association between unpredictable maternal 353 sensory signals and LDI, however the UF:Cingulum GFA ratio measure was a marginally 354 significant mediator of the association between unpredictable maternal sensory signals and LDI 355 (indirect effect = -0.10, P = 0.049, CI = -0.28 to -0.0081; Figure 3c). Critically, this association 356 was specific to lure discrimination and did not generalize to general recognition memory; the 357 indirect association was not statistically significant with the recognition memory outcome 358 (indirect effect = -0.063, P = 0.31, CI = -0.27 to 0.034; Figure 3d).

359 Discussion

360 There are three key findings of this study. First, we show that infant exposure to unpredictable 361 patterns of maternal-derived sensory signals is associated with greater integrity of the uncinate 362 fasciculus in late childhood. Second, this finding is specific to the uncinate fasciculus and is not 363 present in the hippocampal cingulum. Lastly, we find that the ratio of uncinate fasciculus 364 (anterior MTL-PFC connectivity) to hippocampal cingulum GFA (posterior MTL-PFC 365 connectivity) mediates the association between experiencing greater unpredictability of maternal 366 sensory signals and impaired performance on an episodic memory task. Together, these data 367 suggest that one possible consequence of early life unpredictability is de-synchronized 368 maturation of two key corticolimbic pathways resulting in selective impairments in episodic 369 memory tasks requiring nuanced and fine-tuned MTL-PFC connectivity. Such findings are

Page 12 of 29

370 consistent with other studies suggesting these connections are vulnerable to early life adversity 371 (Gee et al., 2013, Tottenham & Galván, 2016).

372 Unpredictable maternal sensory signals have been identified as a novel type of early life 373 adversity, associated with reduced memory and executive functioning (Davis 2017, 2019) an 374 association that has been observed in independent cohorts, such as FinnBrain (Davis, 2019; 375 Karlsson et al., 2018). Evidence that patterns of sensory input contribute to the maturation of 376 brain pathways has been previously shown in rodents (Bolton et al., 2018; Molet et al., 2016a), 377 and affects emotional (Walker et al., 2017) as well as cognitive circuits (Chen & Baram, 2016). 378 Here we provide evidence that corticolimbic circuits are associated with infant exposure to 379 unpredictability from the mother in early life and imbalance among these circuits serves as a 380 possible mechanism explaining differences in cognitive function observed here and in previous 381 studies (Davis 2017, 2019).

382 Past literature on the uncinate fasciculus’ association with various types of early life adversity 383 report mixed results. For example, several studies have found reductions in fractional anisotropy 384 associated with early life maltreatment and deprivation (Govindan et al., 2010; Eluvathingal et 385 al., 2006; Hanson et al., 2015; Ho et al., 2017). In contrast, other studies reported increased 386 diffusion properties of the uncinate fasciculus in premature infants exposed to prenatal maternal 387 stress (Lautarescu et al., 2019), and among adults exposed to childhood trauma (Tatham et al., 388 2016). Discrepancies may be due to differences in type of adversity and developmental stage at 389 assessment and none of these prior studies evaluated unpredictability. We focus on 390 unpredictability as a type of early adversity that has been associated with long term 391 consequences for development (Glynn et al., 2019; Glynn et al., 2018; Howland et al., 2020; 392 Noraña et al. 2020). Patterns of sensory signals are known to shape the development of neural 393 circuits involved in sensory systems and we show that circuits involved in cognitive functions 394 also may be affected. Further, almost all previous studies have quantified integrity of the 395 uncinate fasciculus using tensor-based measures such as fractional anisotropy, and axial and 396 radial diffusivity. These measures rely on the tensor model which assumes a single fiber 397 orientation per voxel, and does not properly account for fiber crossing, bending, or twisting 398 (Alexander et al., 2001). A strength of our approach of modeling multiple fiber orientations using 399 orientation distribution functions is possibly more likely to approximate microstructural properties 400 of complex neural tissue than the tensor model (Fritzsche et al., 2010; Yamada et al., 2018; 401 Gorczewski et al., 2009).

Page 13 of 29

402 Anatomically, the uncinate fasciculus is a U-shaped, fanning white matter bundle with complex 403 structure (Ebeling & von Cramon, 1992; Bhatia et al., 2012, 2017, 2018; Riva-Posse et al., 404 2014; Vergani et al., 2016) and a number of termination points in the medial temporal lobes 405 (Ebeling & Von Cramon, 1992; Von Der Heide et al., 2013; Thiebaut et al., 2012). It has only 406 recently been visualized using Q-space imaging methods that model the U-shaped uncus and 407 orbitofrontal cortex branching and turning patterns (Leng et al., 2016; Bhatia et al., 2017). 408 Developmental studies have shown that white matter rapidly develops during early infancy, with 409 MTL-PFC pathways like the uncinate (as well as cingulum) continuing to develop until the age of 410 35 (Hermoye et al., 2006; Asato et al., 2010; Olson et al., 2015; Von der Heide et al., 2013; 411 Simmonds et al., 2014). In animal models, a very similar pattern exists as the amygdala- 412 prefrontal connection is relatively slow to develop and continues into adolescence with very 413 rapid growth during the first 10 postnatal days of life followed by synaptic pruning in 414 adolescence (Cunningham et al., 2002; Johnson et al., 2016). Although histological evidence is 415 needed to truly determine the significance of age-related increased anisotropy across 416 maturation, evidence points to very rapid limbic white matter change in the first two years of life 417 possibly indicating a period of sensitivity in which environmental stimuli could more drastically 418 affect the development of limbic white matter (Yu et al., 2020). In contrast to the association with 419 the uncinate fasciculus, early life exposure to unpredictability was not related to integrity in the 420 cingulum bundle which is in contrast with previous reports suggesting that this pathway is 421 vulnerable to early life adversity (Marroun et al., 2017; Huang et al., 2012; Choi et al., 2009).

422 The function of the uncinate fasciculus is still not well understood, although recent studies have 423 suggested that it plays a role in episodic memory, and in particular the adjudication among 424 similar mnemonic representations during retrieval (Von Der Heide et al., 2013; Alm et al. 2016). 425 It is well established that episodic memory involves the MTL, the PFC, as well as their 426 interactions (Eichenbaum, 2017; Preston & Eichenbaum, 2013; Jones & Wilson, 2005; Brincat & 427 Miller, 2015; Van Kesteren et al., 2010). Further, the ability to adjudicate among competing 428 memory representation is thought to rely on hippocampal pattern separation. With this in mind, 429 our primary behavioral outcome measure (LDI) was based on the Mnemonic Similarity Task, 430 which is designed to assess hippocampal pattern separation. Our results provide evidence that 431 an imbalanced ratio of uncinate connectivity to cingulum connectivity is related to impaired 432 performance on the Mnemonic Similarity Task. Additionally, our mediation analyses suggest 433 that this imbalanced connectivity, a possible indicator of imbalanced maturation, is a putative

Page 14 of 29

434 mechanism that links the experience of unpredictable maternal sensory signals early in infancy 435 to impaired memory function in childhood.

436 The ratio of integrity parameters of the uncinate fasciculus to cingulum was a mediator in the 437 association between unpredictable maternal sensory signals and performance in our task. This 438 should not be surprising, as both of these projections comprise components of the corticolimbic 439 brain circuitry, which is critical for executing complex behaviors such as memory. Notably, the 440 overall function of the circuit can be thought of as the combinatorial sum of the function and 441 connectivity of its distinct components (Redish and Gordon, 2017). Thus, whereas augmented 442 maturation of a single component or a deficit in another in and by itself may not suffice to distort 443 circuit function, their combined existence may be synergistic, leading to unbalanced functional 444 output.

445 In conclusion, our findings demonstrate that infant exposure to unpredictable patterns of 446 maternal care is associated with imbalanced maturation of the uncinate fasciculus and the 447 cingulum, detected during childhood and that this imbalance may be associated with deficits in 448 memory function. This is consistent with our prior work in rodents showing that early life 449 unpredictability leads to increased structural connectivity of the amygdala to prefrontal cortex 450 pathway in rats (Bolton et al., 2018) and to deficits in hippocampal memory (Molet et al. 2014). 451 This study, along with future investigations will yield novel insight into neurobiological 452 mechanisms of vulnerability to cognitive deficits as a result of unpredictable early environments.

453

454 References

455 Alexander AL, Hasan KM, Lazar M,Tsuruda JS, Parker DL. (2001) Analysis of partial volume 456 effects in diffusion-tensor MRI. Magn Reson Med 45:770–780.

457 Alm, K. H., Rolheiser, T., Olson, I. R. (2016) Inter-Individual Variation in Fronto-Temporal 458 Connectivity Predicts the Ability to Learn Different Types of Associations. Neuroimage 459 132: 213–224.

460 Anderson, S. L., Teicher, M. H. (2008) Stress, sensitive periods and maturational events in 461 adolescent depression. Trends in Neurosciences 31: 183–191.

462 Andersson, J. L. R., Sotiropoulo S. N. (2016). An integrated approach to correction for off-

Page 15 of 29

463 resonance effects and subject movement in diffusion MR imaging. NeuroImage, 464 125:1063-1078

465 Asato, M.R., Terwilliger, R., Woo, J., Luna, B. (2010) White matter development in adolescence: 466 A DTI study. Cereb. Cortex 20: 2122–2131.

467 Babenko, O., Kovalchuk, I., Metz, G.A. (2015) Stress-induced perinatal and transgenerational 468 epigenetic programming of brain development and mental health. Neurosci. Biobehav. 469 Rev 48: 70–91.

470 Bath, K.G., Nitenson, A. S., Lichtman, E., Lopez, C., Chen, W., Gallo, M., Goodwill, H., 471 Manzano-Nieves, G. (2017). Early life stress leads to developmental and sex selective 472 effects on performance in a novel object placement task. Neurobiology of Stress 7: 57– 473 67.

474 Bhatia, K., Henderson, L., Yim, M., Hsu, E., Dhaliwal, R. (2017) Diffusion tensor imaging 475 investigation of uncinate fasciculus anatomy in healthy controls: description of a 476 subgenual stem. Neuropsychobiology 75: 132–140.

477 Bhatia, K.D., Henderson, L., Ramsey-Stewart, G., May, J. (2012) Diffusion tensor imaging to aid 478 subgenual cingulum target selection for deep brain stimulation in depression. 479 Stereotactic Funct. Neurosurg 90: 225–232.

480 Bhatia, K.D., Henderson, L.A., Hsu, E., Yim, M. (2018) Reduced integrity of the uncinate 481 fasciculus and cingulum in depression: A stem-by-stem analysis. Journal of Affective 482 Disorders 235: 220–228.

483 Bolton, J. L., Molet, J., Regev, L., Chen, Y., Rismachni, N., Haddad, E., Yang, D.Z., Obenaus, 484 A., Baram, T.Z. (2018) Anhedonia Following Early-Life Adversity Involves Aberrant 485 Interaction of Reward and Anxiety Circuits and Is Reversed by Partial Silencing of 486 Amygdala Corticotropin-Releasing Hormone Gene. Biological Psychiatry 83: 137-147.

487 Bowlby, J. (1950) Research into the origins of delinquent behaviour. Br Med J 1: 570–573.

488 Bracht, T., Linden, D., & Keedwell, P. (2015). A review of white matter microstructure alterations 489 of pathways of the reward circuit in depression. Journal of Affective Disorders 187: 45– 490 53.

Page 16 of 29

491 Brincat S. L., & Miller E. K. (2015). Frequency-specific hippocampal-prefrontal interactions 492 during associative learning. Nat. Neurosci 18: 576–581.

493 Brunson, K. L., Kramár, E., Lin, B., Chen, Y., Colgin, L. L., Yanagihara, T. K., Lynch, G., Baram, 494 T. Z. (2005). Mechanisms of late-onset cognitive decline after early-life stress. The 495 Journal of neuroscience: the official journal of the Society for Neuroscience 25(41): 496 9328–9338.

497 Bubb, E. J., Metzler-baddeley, C., Aggleton, J.P. (2018) The cingulum bundle: Anatomy, 498 function, and dysfunction, Neuroscience and Biobehavioral Reviews 92: 104–127.

499 Burghy, C.A., Stodola, D.E., Ruttle, P.L., Molloy, E.K., Armstrong, J.M., Oler, J.A., Fox, M.E., 500 Hayes, A.S., Kalin, N.H., Essex, M.J., Davidson, R.J., Birn, R.M. (2012) Developmental 501 pathways to amygdala- prefrontal function and internalizing symptoms in adolescence. 502 Nat Neurosci 15: 1736-1741.

503 Burgos-Robles, A., Kimchi, E. Y., Izadmehr, E. M., Porzenheim, M. J., Ramos-Guasp, W. A., 504 Nieh, E. H., Felix- Ortiz, A.C., Namburi, P., Leppla, C.A., Presbrey, K.N., Anandalingam, 505 K.K., Pagan-Rivera, P.A., Anahtar, M., Beyeler, A., Tye, K. M. (2017) Amygdala inputs to 506 prefrontal cortex guide behavior amid conflicting cues of reward and punishment. Nature 507 Neuroscience 20: 824–835.

508 Carballedo, A., Amico, F., Ugwu, I., Fagan, A. J., Fahey, C., Morris, D., Meaney, J.F., Leemans, 509 A., Frodl, T. (2012) Reduced fractional anisotropy in the uncinate fasciculus in patients 510 with major depression carrying the met-allele of the Val66Met brain-derived neurotrophic 511 factor genotype. American Journal of Medical Genetics, Part B: Neuropsychiatric 512 Genetics 159: 537–548.

513 Chen, Y., Baram, T. Z. (2016) Toward understanding how early-life stress reprograms cognitive 514 and emotional brain networks. Neuropsychopharmacology 41: 197–206.

515 Chiang, H. L., Chen, Y. J., Shang, C. Y., Tseng, W. Y. I., & Gau, S. S. F. (2016). Different 516 neural substrates for executive functions in youths with ADHD: A diffusion spectrum 517 imaging tractography study. Psychological Medicine 46(6): 1225–1238.

518 Choi, J., Jeong, B., Rohan, M. L., Polcari, A. M., & Teicher, M. H. (2009). Preliminary evidence 519 for white matter tract abnormalities in young adults exposed to parental verbal

Page 17 of 29

520 abuse. Biological psychiatry 65(3): 227–234.

521 Cox, J.L., Holden, J.M., and Sagovsky, R. (1987) Detection of postnatal depression: 522 Development of the 10-item Edinburgh Postnatal Depression Scale. British Journal of 523 Psychiatry 150: 782-786.

524 Cunningham, M.G., Bhattacharyya, S., Benes, F.M. (2002) Amygdalo-cortical sprouting 525 continues into early adulthood: implications for the development of normal and abnormal 526 function during adolescence. J. Comp. Neurol 453: 116–130.

527 Davis, E. P., Korja, R., Karlsson, L., Glynn, L. M., Sandman, C. A., Vegetabile, B., Kataja, E., 528 Nolvi, S., Sinerva, E., Pelto, J., Karlsson, H., Stern, H.S., Baram, T. Z. (2019). Across 529 continents and demographics, unpredictable maternal signals are associated with 530 children’s cognitive function. EBioMedicine 46, 256–263.

531 Davis, E.P., Stout, S.A., Molet, J., Vegetabile, B., Glynn, L.M., Sandman, C.A., Hein, K., Stern, 532 H., Baram, T.Z. (2017) Early life exposure to unpredictable maternal sensory signals 533 influences cognitive development: a cross-species approach. Proc. Natl. Acad. Sci. USA 534 14: 10390-10395

535 Ebeling, U., von Cramon, D. (1992) Topography of the uncinate fascicle and adjacent temporal 536 fiber tracts. Acta Neurochir (Wien) 115:143–148.

537 Eichenbaum, H. (2017). Prefrontal–hippocampal interactions in episodic memory. Nature 538 Reviews Neuroscience 18, 547.

539 Eluvathingal, T. J., Chugani, H. T., Behen, M. E., Juha ́sz, C., Muzik, O., Maqbool, M., Chugani, 540 D.C., Makki, M. (2006) Abnormal brain connectivity in children after early severe 541 socioemotional deprivation: A diffusion tensor imaging study. Pediatrics 117: 2093– 542 2100.

543 Fritzsche, K.H., Laun, F.B., Meinzer, H.P., Stieltjes, B. (2010) Opportunities and pitfalls in the 544 quantification of fiber integrity: What can we gain from Q-ball imaging? NeuroImage 51: 545 242–251.

546 Gardner CO, Kendler K. Sex differences in the pathways to major depression: a study of 547 opposite-sex twin pairs. (2014) Am J Psychiatry 171:426–35.

Page 18 of 29

548 Gee, D.G., Gabard-durnam, L.J., Flannery, J., Goff, B., Humphreys, K.L., Telzer, E.H. , Hare͒ 549 T.A., Bookheimer, S.Y., Tottenham, N. (2013) Early developmental emergence of 550 human amygdala – prefrontal connectivity after maternal deprivation. Proc Natl Acad Sci 551 USA 110: 15638-43

552 Gilles, E.E., Schultz, L., Baram, T.Z. (1996) Abnormal corticosterone regulation in an immature 553 rat model of continuous chronic stress. Pediatric neurology 15: 114–119.

554 Glynn, L.M., Baram, T.Z. (2019) The Influence of Unpredictable, Fragmented Parental Signals 555 on the Developing Brain. Frontiers in neuroendocrinology 53, 100736.

556 Glynn, L. M., Howland, M. A., Sandman, C. A., Davis, E. P., Phelan, M., Baram, T. Z., & Stern, 557 H. S. (2018). Prenatal maternal mood patterns predict child temperament and 558 adolescent mental health. Journal of Affective Disorders 228, 83–90.

559 Glynn, L. M., Stern, H. S., Howland, M. A., Risbrough, V. B., Baker, D. G., Nievergelt, C. M., 560 Baram, T.Z., Davis, E. P. (2019). Measuring novel antecedents of mental illness: the 561 Questionnaire of Unpredictability in Childhood. Neuropsychopharmacology 44(5), 876– 562 882.

563 Gorczewski, K., Mang, S., Klose, U. (2009) Reproducibility and consistency of evaluation 564 techniques for HARDI data. MAGMA 22: 63–70.

565 Govindan, R.M., Behen, M.E., Helder, E., Makki, M.I., Chugani, H.T. (2010) Altered water 566 diffusivity in cortical association tracts in children with early deprivation identified with 567 tract-based spatial statistics (TBSS). Cerebral Cortex 20: 561–569.

568 Grieger, L.D., Schoeni, R.F., Danziger, S., (2009) Accurately Measuring the Trend in Poverty In 569 the United States Using The Panel Study of Income Dynamics. J Econ Soc Meas 34, 570 105.

571 Groh, A.M., Roisman, G.I., van Ijzendoorn, M.H., Bakermans-Kranenburg, M.J., Fearon, R.P. ͒ 572 (2012) The significance of insecure and disorganized attachment for children’s in- ͒ 573 ternalizing symptoms: A meta-analytic study. Child Dev 83:591–610. ͒

574 Guadagno, A., Verlezza, S., Long, H., Wong, T.P., Walker, C.D. (2016) The effects of chronic 575 early-life stress on amygdala morphology and function in neonatal rats. Paper presented

Page 19 of 29

576 at the Neurobiology of Stress Workshop, Irvine, April 12–15.

577 Guadagno, A., Wong, T.P., Walker, C.D. (2017) Morphological and functional changes in the 578 preweaning basolateral amygdala induced by early chronic stress associate with anxiety 579 and fear behavior in adult male, but not female rats. Prog Neuropsychopharmacol Biol 580 Psychiatry 81: 25-37.

581 Hanson, J.L., Knodt, A.R., Brigidi, B.D., Hariri, A.R. (2015) Lower structural integrity of the 582 uncinate fasciculus is associated with a history of child maltreatment and future 583 psychological vulnerability to stress. Development and Psychopathology 27: 1611–9.

584 Hasler, G., Drevets, W.C., Manji, H.K., Charney, D.S. (2004) Discovering Endophenotypes for 585 Major Depression. Neuropsychopharmacology 1765–1781.

586 Henderson, S.E., Johnson, A.R., Vallejo, A.I., Katz, L., Wong, E., Gabbay, V. (2013) A 587 preliminary study of white matter in adolescent depression: relationships with illness 588 severity, anhedonia, and irritability. Front. Psychiatry 4: 152.

589 Hermoye, L., Saint-Martin, C., Cosnard, G., Lee, S.K., Nassogne, M.C., Menten, R., Clapuyt, P., 590 Donohue, P.K., Hua, K, Wakana, S., Jiang, H., van Zijl, P.C., Mori, S. (2006) Pediatric 591 diffusion tensor imaging: normal database and observation of the white matter 592 maturation in early childhood. Neuroimage 29: 493-- 504.

593 Ho, T.C., King, L.S., Leong, J.K., Colich, N.L., Humphreys, K.L., Ordaz, S.J., Gotlib, I.H. (2017). 594 Effects of Sensitivity to Life Stress on Uncinate Fasciculus Segments in Early 595 Adolescence. Social Cognitive and Affective Neuroscience October 2016: 1–9.

596 Howland, M.A., Sandman, C.A., Davis, E.P., Stern, H., Phelan, M., Baram, T.Z., & Glynn, L.M. 597 (2020). Prenatal maternal mood entropy is associated with child 598 neurodevelopment. Emotion. In Press

599 Huang, H., Gundapuneedi, T., & Rao, U. (2012). White matter disruptions in adolescents 600 exposed to childhood maltreatment and vulnerability to psychopathology. 601 Neuropsychopharmacology 37(12), 2693–2701.

602 Ivy, A.S., Brunson, K.L., Sandman, C., Baram, T.Z. (2008) Dysfunctional nurturing behavior in 603 rat dams with limited access to nesting material: a clinically relevant model for early-life

Page 20 of 29

604 stress. Neuroscience 154: 1132–1142.

605 Ivy, A.S., Rex, C.S., Chen, Y., Dube, C., Maras, P.M., Grigoriadis, D.E., Gall, C.M., Lynch, G., 606 Baram, T.Z. (2010) Hippocampal dysfunction and cognitive impairments provoked by 607 chronic early-life stress involve excessive activation of CRH receptors. J. Neurosci. 30: 608 13005–13015.

609 Johnson, C.M., Loucks, A., Peckler, H., Thomas, A., Janak, P., Wilbrecht, L. (2016) Long-range 610 orbitofrontal and amygdala axons show divergent patterns of maturation in the frontal 611 cortex across adolescence. Developmental Cognitive Neuroscience 18:113–120.

612 Jones M. W., & Wilson M. A. (2005). Theta rhythms coordinate hippocampal–prefrontal 613 interactions in a spatial memory task. PLoS Biol. 3: 402.

614 Karlsson, L., Tolvanen, M., Scheinin, N. M., Uusitupa, H. M., Korja, R., Ekholm, E., Tuulari, J., 615 Pajulo, M., Huotilainen, M., Paunio, T., Karlsson, H. (2018). Cohort Profile: The 616 FinnBrain Birth Cohort Study (FinnBrain). International Journal of Epidemiology 47(1), 617 15-16.

618 Keedwell, P.A., Chapman, R., Christiansen, K., Richardson, H., Evans, J., Jones, D.K. (2012) 619 Cingulum white matter in young women at risk of depression: The effect of family history 620 and anhedonia. Biol. Psychiatry 72: 296–302.

621 Kessler RC. Epidemiology of women and depression. (2003) J Affect Disord 74:5–13

622 Koh, C., Tang, P., Chen, H., Hsu, Y., Hsieh, C., Tseng, W.I. (2018) Impaired Callosal Motor 623 Fiber Integrity and Upper Extremity Motor Impairment Are Associated With Stroke 624 Lesion Location. Neurorehabil Neural Repair 32: 602-612.

625 Lautarescu, A., Pecheva, D., Nosarti, C., Nihouarn, J., Zhang, H., Victor, S., Craig, M., 626 Edwards, D., Counsell, S. J. (2019). Maternal Prenatal Stress Is Associated with Altered 627 Uncinate Fasciculus Microstructure in Premature Neonates. Biological Psychiatry 1–11.

628 Lebel, C., Walton, M., Letourneau, N., Giesbrecht, G.F., Kaplan, B.J., Dewey, D. (2015) 629 Prepartum and postpartum maternal depressive symptoms are related to children's brain 630 structure in preschool. Biol. Psychiatry 80: 859-868.

631 Leng, B., Han, S., Bao, Y., Zhang, H., Wang, Y., Wu, Y., Wang, Y. (2016). The uncinate

Page 21 of 29

632 fasciculus as observed using diffusion spectrum imaging in the human brain. 633 Neuroradiology 58: 595–606

634 Lo, Y. C., Chen, Y. J., Hsu, Y. C., Tseng, W. Y. I., & Gau, S. S. F. (2017). Reduced tract 635 integrity of the model for social communication is a neural substrate of social 636 communication deficits in autism spectrum disorder. Journal of Child Psychology and 637 Psychiatry and Allied Disciplines 58(5), 576–585.

638 Lu, S., Wei, Z., Gao, W., Wu, W., Liao, M., Zhang, Y., Li, W., Li, Z., Li, L. (2013) White matter 639 integrity alterations in young healthy adults reporting childhood trauma: A diffusion 640 tensor imaging study. Aust N Z J Psychiatry 47(139).

641 Leal, S.L., Yassa, M.A. (2018) Integrating new findings and examining clinical applications of 642 pattern separation. Nature Neuroscience 21: 163-173.

643 Leutgeb, J. K., Leutgeb, S., Moser, M.-B. & Moser, E. I. (2007) Pattern separation in the dentate 644 gyrus and CA3 of the hippocampus. Science 315, 961–966.

645 Marr, D. (1971) Simple memory: a theory for archicortex. Phil. Trans. R. Soc. Lond. B 262: 23– 646 81

647 Marroun, H. El, Jaddoe, V.W., White, T., Zou, R., Muetzel, R.L., Verhulst, F.C., Tiemeier, H. 648 (2018) Prenatal exposure to maternal and paternal depressive symptoms and white 649 matter microstructure in children. Depress Anxiety 35: 321–329.

650 McClelland, J. L., McNaughton, B. L. & O’Reilly, R. C. (1995) Why there are complementary 651 learning systems in the hippocampus and neocortex: insights from the successes and 652 failures of connectionist models of learning and memory. Psychol. Rev. 102, 419–457.

653 Molet, J., Heins, K., Zhuo, X., Mei, Y.T., Regev, L., Baram, T.Z., Stern, H. (2016a) 654 Fragmentation and high entropy of neonatal experience predict adolescent emotional 655 outcome. Translational Psychiatry 702.

656 Molet, J., Maras, P.M., Kinney-Lang, E., Harris, N.G., Rashid, F., Ivy, A.S., Solodkin, A., 657 Obenaus, A., Baram, T.Z. (2016b) MRI uncovers disrupted hippocampal microstructure 658 that underlies memory impairments after early-life adversity. Hippocampus 26: 1618– 659 1632.

Page 22 of 29

660 NICHD Early Child Care Research Network (1999) Chronicity of maternal depressive 661 symptoms, maternal sensitivity, and child functioning at 36 months. Dev Psychol 662 35:1297–1310.

663 Noraña, A., Morgan, A., Glynn, L.M., Sandman, C.A., Baram, T.Z., Stern, H.S., Davis, E.P. 664 (2020) Unpredictable maternal behavior is associated with a blunted infant cortisol 665 response. Developmental Psychobiology. In Press

666 Olson, I.R., Heide, R.J. Von Der, Alm, K.H., Vyas, G. (2015) Development of the uncinate 667 fasciculus: Implications for theory and developmental disorders. Developmental 668 Cognitive Neuroscience 14: 50–61.

669 Perlman, W.R. Webster, M.J., Herman, M.M., Kleinman, J.E., Weickert, C.S. (2007) Age-related 670 differences in glucocorticoid receptor mRNA levels in the human brain. Neurobiol. Aging 671 28: 447– 458

672 Preston, A. R., & Eichenbaum, H. (2013). Interplay of hippocampus and prefrontal cortex in 673 memory. Current Biology 23(17), 764–773.

674 Raineki, C., Cortes, M.R., Belnoue, L., Sullivan, R.M. (2012) Effects of early-life abuse differ 675 across development: Infant social behavior deficits are followed by adolescent 676 depressive-like behaviors mediated by the amygdala. Journal of Neuroscience 32: 677 7758–7765.

678 Redish, D.A., Gordon, J.A. (2017) Computational Psychiatry: A New Perspective on Mental 679 Illness. Cambridge, MA. The MIT Press pp 12.

680 Reynolds K, Pietrzak RH, El-Gabalawy R, Mackenzie CS, Sareen J. (2015) Prevalence of 681 psychiatric disorders in U. S. older adults: findings from a nationally representative 682 survey. World Psychiatry 14: 74–81.

683 Riva-Posse, P., Choi, K.S., Holtzheimer, P.E., McIntyre, C.C., Gross, R.E., Chaturvedi, A., 684 Crowell, A.L., Garlow, S.J., Rajendra, J.K., Mayberg, H.S. (2014) Defining critical white 685 matter pathways mediating successful subcallosal cingulate deep brain stimulation for 686 treatment-resistant depression. Biol. Psychiatry 76: 963–969.

687 Rollins, L., & Cloude, E. B. (2018). Development of mnemonic discrimination during childhood.

Page 23 of 29

688 Learning and Memory 25(6), 294–297.

689 Sandman, C. A., Glynn, L. M., & Davis, E. P. (2013). Is there a viability-vulnerability tradeoff? 690 Sex differences in fetal programming. Journal of Psychosomatic Research 75(4), 327– 691 335.

692 Short, A. K., & Baram, T. Z. (2019). Early-life adversity and neurological disease: age-old 693 questions and novel answers. Nature Reviews Neurology 15: 657-669

694 Simmonds, D.J., Hallquist, M.N., Asato, M., Luna, B. (2014) Developmental stages and sex 695 differences of white matter and behavioral development through adolescence: a 696 longitudinal diffusion tensor imaging (DTI) study. NeuroImage 92: 356–368.

697 Sinharay, S., Stern, H. S., & Russell, D. (2001). The use of multiple imputation for the analysis 698 of missing data. Psychological Methods 6(4), 317–329.

699 Spielberger CD, Gorsuch RL, Lushene R, Vagg PR, Jacobs GA. (1983) Manual for the State- 700 Trait Anxiety Inventory. Consulting Psychologists Press.

701 Sroufe LA (2005) Attachment and development: A prospective, longitudinal study ͒from birth to 702 adulthood. Attach Hum Dev 7:349–367. ͒

703 Tatham, E.L., Ramasubbu, R., Gaxiola-Valdez, I., Cortese, F., Clark, D., Goodyear, B., Foster, 704 J., Hall, G.B. (2016) White matter integrity in major depressive disorder: Implications of 705 childhood trauma, 5-HTTLPR and BDNF polymorphisms. Psychiatry Research – 706 Neuroimaging 253: 15–25.

707 Thiebaut de Schotten, M., Dell'Acqua, F., Valabregue, R., Catani, M. (2012) Monkey to human 708 comparative anatomy of the frontal lobe association tracts. Cortex 48: 82–96.

709 Tottenham, N., Galván, A. (2016) Neuroscience and Biobehavioral Reviews Stress and the 710 adolescent brain Amygdala-prefrontal cortex circuitry and ventral striatum as 711 developmental targets. Neuroscience and Biobehavioral Reviews 70: 217–227.

712 Treves, A. & Rolls, E. T. (1994) Computational analysis of the role of the hippocampus in 713 memory. Hippocampus 4, 374–391.

714 Tuch, D.S. (2004). Q-ball imaging. Magnetic Resonance in Medicine 52: 1358–1372.

Page 24 of 29

715 Tuch, D.S., Reese, T.G., Wiegell, M.R., Makris, N., Belliveau, J.W., Wedeen, V.J. (2002) High 716 angular resolution diffusion imaging reveals intravoxel white matter fiber heterogeneity. 717 Magn. Reson. Med. 48: 577–582.

718 Van Kesteren, M. T., Fernández, G., Norris, D. G., Hermans, E. J. (2010). Persistent schema- 719 dependent hippocampal-neocortical connectivity during memory encoding and 720 postencoding rest in humans. Proc Natl AcadSci 107:7550-5.

721 Vegetabile, B.G., Stout, S.A., Davis, E.P., Stern, H.S. (2019) Estimating the Entropy Rate of 722 Finite Markov Chains with Application to Behavior Studies. Journal of Educational and 723 Behavioral Statistics 44: 282-308.

724 Vergani, F., Martino, J., Morris, C., Attems, J., Ashkan, K., Dell'Acqua, F. (2016) Anatomic 725 connections of the subgenual cingulate region. Neurosurgery 79: 465–472.

726 Vilgis, V., Vance, A., Cunnington, R., Silk, T.J. (2017) White matter microstructure in boys with 727 persistent depressive disorder. J Affect Disorders 221: 11-16.

728 Von Der Heide, R.J., Skipper, L.M., Klobusicky, E., Olson, I.R. (2013) Dissecting the uncinate 729 fasciculus: Disorders, controversies and a hypothesis. Brain 136: 1692–1707.

730 Walker, C.D., Bath, K.G., Joels, M., Korosi, A., Larauche, M., Lucassen, P. J., Morris, M.J., 731 Raineki, C., Roth, T.L., Sullivan, R.M., Taché, Y., Baram, T.Z. (2017) Chronic early life 732 stress induced by limited bedding and nesting (LBN) material in rodents: critical 733 considerations of methodology, outcomes and translational potential. Stress 421-448.

734 Wise, T., Radua, J., Nortje, G., Cleare, A. J., Young, A. H. (2016) Archival Report Voxel-Based 735 Meta-Analytical Evidence of Structural Disconnectivity in Major Depression and Bipolar 736 Disorder. Biological Psychiatry 79: 293–302.

737 Yamada, N., Ueda, R., Kakuda, W., Momosaki, R., Kondo, T., Hada, T., Sasaki, N., Hara, T., 738 Senoo, A., Abo, M. (2018) Diffusion Tensor Imaging Evaluation of Neural Network 739 Development in Patients Undergoing Therapeutic Repetitive Transcranial Magnetic 740 Stimulation following Stroke. Neural Plast 3901016.

741 Yan, C.G., Rincón-Cortés, M., Raineki, C., Sarro, E., Colcombe, S., Guilfoyle, D.N., Yang, Z., 742 Gerum, S., Biswal, B.B., Milham, M.P., Sullivan, R.M., Castellanos, F.X. (2017) Aberrant

Page 25 of 29

743 development of intrinsic brain activity in a rat model of caregiver maltreatment of 744 offspring. Translational Psychiatry 7, e1005.

745 Yang, X. hua, Wang, Y., Wang, D. fang, Tian, K., Cheung, E. F. C., Xie, G. rong, Chan, R. C. K. 746 (2017) White matter microstructural abnormalities and their association with anticipatory 747 anhedonia in depression. Psychiatry Research - Neuroimaging 264: 29–34.

748 Yassa, M. A., Stark, C. E. L. (2011) Pattern separation in the hippocampus. Trends in 749 Neurosciences 34: 515– 525.

750 Yeh, F.C., Tseng, W.Y.I. (2011) NTU-90: A high angular resolution brain atlas constructed by q- 751 space diffeomorphic reconstruction. NeuroImage 58: 91–99.

752 Yeh, F.C., Verstynen, T.D., Wang, Y., Fernández-Miranda, J.C., Tseng, W.Y.I. (2013) 753 Deterministic diffusion fiber tracking improved by quantitative anisotropy. PLoS ONE 8: 754 1–16.

755 Yu, Q., Peng, Y., Kang, H., Peng, Q., Ouyang, M., Slinger, M., Hu, D., Shou, H., Fang, F., 756 Huang, H. (2020) Differential White Matter Maturation from Birth to 8 Years of 757 Age, Cerebral Cortex 30(4): 2673–2689.

758 Zhang, A., Leow, A., Ajilore, O., Lamar, M., Yang, S., Joseph, J., Medina, J., Zhan, L., Kumar, 759 A. (2012) Quantitative tract-specific measures of uncinate and cingulum in major 760 depression using diffusion tensor imaging. Neuropsychopharmacology 37: 959–967.

761

762

763

764

765

766

767

Page 26 of 29

768 Figures

769 Figure 1: Mnemonic Similarity Task (MST). The MST consist of two phases. An incidental 770 encoding phase and a retrieval (testing) phase. The incidental encoding phase consists of a 771 series of 64 images where children are asked to indicate if the items would appear “indoor” or 772 “outdoor”. Upon completion of the incidental encoding phase, children completed the retrieval or 773 testing portion of the MST. During test, children were presented with 32 target images (identical 774 images that appeared during encoding), 32 foil images (images that did not appear during 775 encoding), and 32 lure images (variants of images presented during encoding). Children were 776 asked to indicate if the presented object was “Old” (they had seen the image before), “Similar” 777 (the images were slightly different than images presented during test) or “New” (images were 778 novel and not seen during test). Lures items were varied by similarity ratings on a scale from 1 779 to 5 with 1 being the most similar (bin 1: most difficult to discriminate) and 5 being the least 780 similar (bin 5: easier to discriminate).

781 Figure 2. Generalized fractional anisotropy results. (a) A significant positive association 782 between exposure to unpredictable maternal sensory signals during infancy and mean uncinate 783 fasciculus GFA primarily in girls (r = 0.44, P = 0.016) ages of 9 to 11. This association was not 784 significant in boys (r = 0.30, P = 0.057). (b) No association between exposure to unpredictable 785 maternal sensory signals during infancy and mean hippocampal cingulum GFA in girls (r = - 786 0.29, P = 0.13) or boys (r = 0.15, P = 0.35) ages of 9 to 11.

787 Figure 3. Results of UF:Cingulum ratio analyses. (a) Shows the statistically significant 788 positive association between unpredictable maternal sensory signals and the ratio of anterior 789 MTL-PFC connectivity (uncinate fasciculus GFA) to posterior MTL-PFC connectivity 790 (hippocampal cingulum GFA) (r = 0.28, P = 0.018). (b) Shows the marginally significant 791 association between greater ratio of anterior MTL-PFC connectivity to posterior MTL-PFC 792 connectivity and impaired lure discrimination on low similarity lure items (r = -0.23, P = 793 0.058). (c-d) Show the results of the mediation analyses. β values represent the coefficient for 794 the arrows effect. c' represents the direct effect and c represents the total effect of unpredictable 795 maternal sensory signals on low similarity lure discrimination. (c) Shows the significant indirect 796 effect of the Ratio of uncinate fasciculus GFA to hippocampal cingulum GFA mediating the 797 association between unpredictable maternal sensory signals and low similarity lure 798 discrimination. (d) Shows the non-significant indirect effect of the Ratio of uncinated fasciculus

Page 27 of 29

799 GFA to hippocampal cingulum GFA mediating the association between unpredictable maternal 800 sensory signals and recognition performance.

801 Figure 4. Anterior and posterior MTL-PFC circuits in humans. The circuit schematic 802 summarizes our results demonstrating that (1) the anterior MTL-PFC pathway (i.e. the uncinate 803 fasciculus in green lines), which connects the amygdala and rhinal cortex directly to the 804 orbitofrontal cortex is enhanced in relationship to increased unpredictability of maternal sensory 805 signals and (2) the posterior MTL-PFC pathway (i.e. the cingulum in red and white), which 806 connects the hippocampus and rhinal cortex with the retrosplenial cortex as well as medial 807 prefrontal and orbitofrontal cortex is not related to increased unpredictability of maternal sensory 808 signals. These aberrations in circuitry are consistent with a developmental scenario primarily in 809 female children aged 9-11 in which unpredictable maternal sensory signals bias connectivity in 810 favor of the anterior pathway at the expense of the posterior pathway and possibly lead to 811 episodic memory dysfunction.

812 Table 1. Sample characteristics for mother/infant dyads (N=73)

Maternal Characteristics

Cohabitation Status (% Married or Cohabitating) 84.9 Education in years 16 (.4) Education (%) High School or Less 11.1 Some College, Associates or Vocational Degree 28.7 4-Year College Degree 28.8 Graduate Degree 23.3 Maternal Race/Ethnicity (%) White/European/North African/Middle Eastern 49.3 Asian 12.3 Hispanic White 30.1 Multi-Ethnic/Other 8.2 Maternal Depression/Anxiety (Mean r SE) State Anxiety Inventory (STAI) 17.2222 r 0.641 Edinburgh Postnatal Depression Scale (EPDS) 4.95 r 0.612 Maternal Sensitivity (Mean r SE) 9.85 r 0.12 Income to Needs Ratio (Mean r SE) 509.36 r 59.5 Maternal Unpredictable Sensory Signals (Mean r SE) 0.81 r 0.018 Child Characteristics Age at MRI Scan (Mean + SE) 9.92 r 0.078 Sex 43 males, 30 females Days between Scan and MST Completion 99 r 12.14 813

Page 28 of 29

814

815

816 Table 2. Unpredictability of maternal sensory signals related to increased uncinate 817 fasciculus GFA accounting for possible covariates.

Regression Model R2 F β Standardized SE(β) P-value β Overall Model 0.35 6.73 Unpredictability of 1.34 x 10-2 0.26 5.62x10-3 0.020 Maternal Sensory Signals Maternal Depression 5.054x10-4 0.065 7.96x10-4 0.53 Anxiety Maternal Sensitivity -1.56x10-3 -0.21 8.041x10-4 0.057 Income-to-Needs Ratio -1.01x10-6 -0.066 1.60x10-6 0.53 Sex -6.30x10-3 -0.39 1.67x10-3 0.00036

818

819 Table 3: Results of interaction model with unpredictability of hippocampal cingulum GFA 820 predicted by unpredictability of maternal sensory signals, sex and their interaction.

Regression Model R2 F β Standardized SE(β) P- β value Overall Model 0.052 1.18 Unpredictability of Maternal 0.025 0.59 0.015 0.11 Sensory Signals Sex 0.013 1.024 0.0084 0.12

Sex*Unpredictability -0.018 -1.22 0.010 0.086

821

822

823

824

Page 29 of 29