Converging lines of evidence across epidemiological and experimental studies have identified prenatal MIA-exposure and adolescent THC exposure as risk factors for neuropsychiatric illness. However, most individuals exposed to either risk factor alone do not develop psychiatric or neurodevelopmental disorders, which suggests that multiple exposures within an individual are required for disease onset (4,42). The work presented here is aimed at experimentally testing the multiple exposure hypothesis, by exposing MIA-primed offspring to a second hit: chronic adolescent THC exposure. Our results suggest that exposure to either prenatal MIA or adolescent THC were not sufficient to induce enduring neuroanatomical or behavioural alterations in adulthood, while combined exposure was, in some regions but not others. Behaviourally, only subtle, subthreshold alterations in anxiety-like and sensorimotor gating behaviours were observed in the group exposed to both risk factors. Interestingly, sex-differences were observed in patterns of brain-behaviour covariation, both for adult brain volume and for the within-subject brain volume change from post-treatment to adulthood, and adult behaviour. In several brain regions showing susceptibility to risk factor exposure (anterior cingulate and somatosensory cortices, striatum, hippocampus), striking decreases in the expression of CB1- and CB2-IR cells were observed in offspring exposed to either prenatal MIA, adolescent THC, or both, indicating that the effects of these risk factors may converge on the endocannabinoid system, amongst others. Our results show that a single risk factor may not be sufficient to cause enduring changes in gross neuroanatomy and behaviour, but that a combined exposure may be. Further, a single risk factor may affect the brain at finer scales of resolution.

Preclinical investigation of MIA-exposure has been critical to our understanding of neurodevelopmental disorders (6,43,44) identifying behavioural, neuroanatomical, and transcriptional alterations relevant to ASD and schizophrenia (44–49). Mice prenatally exposed to MIA had no statistically significant deviations in developmental trajectories, nor in anxiety-like, social, or sensorimotor gating behaviours. The lack of statistically significant effects was somewhat surprising, and not consistent with previous reports of significantly altered brain and behavioural development from our group (7) and others (37,49). However, exploration of the findings at an uncorrected (p<0.05) threshold identified altered trajectories in the striatum, hippocampus, lateral septum, and somatosensory cortex, wherein the MIA-exposed mice had an overshoot in the adolescent/early-adult period followed by a normalization in later adulthood (relative to SAL-SAL), in line with our previously published findings (7). The lack of behavioural alterations in adulthood is consistent with our previous findings; however, there are many reports of altered behaviour in adult MIA-exposed offspring across multiple domains (44). Some methodological differences may explain why we do not recapitulate our findings at the same level of statistical significance between this study and our previous one (7); most importantly, mice (lower n) were scanned only three times (compared to four in our previous work), which may have reduced our sensitivity to detect alterations in developmental trajectories (50). Additionally, the ages selected for MRI in this study did not perfectly match those of our previous work, which may have decreased our ability to capture the stages of greatest change across development. Significant brain structure alterations have been replicated by our group in adolescent/young adult (PND 35 and 60) MIA-exposed offspring indicating that we may have missed the vulnerable window in the current study (51). Finally, structural imaging was performed slightly differently between studies; here we used a cryogenically cooled surface coil with no contrast enhancement (manganese chloride administration 24 hours prior to scan), whereas in the previous study we used a quadrature volumetric coil, with contrast enhancement; perhaps differences in structural MRI acquisition could contribute to the lack of consistency in our findings.

Mice chronically exposed to THC in adolescence similarly had transient deviations in neurodevelopmental trajectories, with a peak difference in the adolescent/early-adult period. In various regions including the somatosensory cortex, medial preoptic area, globus pallidus, third ventricle, and reticular nucleus, THC exposed mice experienced a volume increase at the post-treatment timepoint (PND50), followed by a normalization at the later adult timepoint. In contrast, hippocampal volume decreased at the posttreatment timepoint, but also normalized. These results indicate that the chronic THC treatment may have induced neuroanatomical remodeling, but that given a few weeks without THC exposure, the brain normalizes relative to the control (SAL-SAL) exposed offspring. In agreement with the lack of neuroanatomical differences by the adult timepoint (PND85), we did not detect any behavioural alterations; had we tested these mice earlier in development or across other behavioural domains, we may have detected differences.

To our knowledge, no longitudinal rodent MRI studies exist examining the effects of chronic adolescent THC exposure on brain structure. Studies using other modalities do report neuroanatomical, neurochemical, and behavioural abnormalities relevant to the psychosis spectrum (36,38,52). Positron emission tomography studies of rats exposed to chronic THC report increased D2-like receptor availability in the dorsal striatum (53). Furthermore, chronic THC exposure in adolescence has been associated with decreased synaptic arborization in the rat hippocampus (36), and the prefrontal cortex (38). Impairments in cognitive flexibility, sensorimotor gating, and memory have all been identified in rodents chronically exposed to THC in adolescence (54–56).

Greater alterations to neurodevelopmental trajectories were observed in offspring exposed to both prenatal MIA and chronic adolescent THC. In the nucleus accumbens, anterior cingulate cortex, and entorhinal cortex, significant deviations in trajectory were observed, defined by enlarged volume in the POL-THC group relative to SAL-SAL at the post-treatment (PND50) scan. Similarly, to the SAL-THC group, these brain regions normalized in the absence of continued THC exposure by later adulthood (PND85). However, in the striatum and somatosensory cortex, the volume increase observed following the post-treatment scan was sustained into later adulthood, indicative of a more lasting neuroanatomical change. Similarly, in the subiculum, a volume decrease was observed following treatment, which was sustained into later adulthood. These regions may be more sensitive to the combined risk factor exposure; interestingly, they have been previously implicated in neurodevelopmental and neuropsychiatric disorders (57–62). Finally, investigation of the effects of THC in addition to MIA (POL-THC vs POL-SAL groups) revealed that mice exposed to both risk factors experience sustained growth of the bed nucleus of the stria terminalis, ventromedial thalamus, reticular nucleus and CA1, relative to mice exposed to MIA and adolescent SAL.

Behaviourally, we only observed subthreshold decreases in anxiety-like behaviours in our combined POL-THC group relative to SAL-SAL, potentially reflective of increased risk-taking behaviours. A subthreshold difference was also observed in sensorimotor gating behaviour, wherein the POL-THC mice performed worse at lower prepulse tones, but better at louder tones. Thus, although the combined risk factors were sufficient to induce neuroanatomical changes, they may not be sufficient to affect the behaviours we tested.

Strikingly, we observed dramatic decreases in CB-1 and CB-2 cell and fiber density in our regions of interest. Most pronounced were the decreases in the somatosensory cortex and ACC for both CB1 and 2. The somatosensory cortex showed consistent neuroanatomical alteration across the three risk factor groups, while the ACC showed sensitivity to both POL and POL-THC exposure. We observed a similar decrease in the fiber density of both CB1 and 2 staining in the striatum, another region affected by POL and POL-THC exposure. Finally, in the hippocampus we observed the most striking decreases in both CB1 and 2 cell and fiber density in the mice prenatally exposed to POL, while those exposed to adolescent THC (either prenatal SAL or POL) also had decreased density relative to SAL-SAL, but higher density relative to POL-SAL. Decreased CB1 receptor expression has been previously reported in the rodent brain (hippocampus and striatum) following both acute (63) and chronic THC exposure (64). Decreased sensitivity has also been observed in a number of rodent brain areas (cortex, hippocampus, and periaqueductal gray) following chronic use (31), as well as in chronic human users (temporal lobe, cingulate cortex, and nucleus accumbens (65). In contrast, increased expression has also been reported following short exposure to THC (66), indicating that the duration of exposure may modulate the directionality of changes to CB1 expression. Interestingly, previous work using PET imaging to longitudinally asses CB1 receptor expression in the offspring of MIA-exposed offspring (GD 15) found that MIA-exposed rats had lower CB1 receptor standard uptake values in the globus pallidus in adolescence, and lower values in the sensory cortex and hypothalamus in adulthood, relative to controls; these findings indicate that MIA may alter the cannabinoid system in adolescence and adulthood, potentially leading to greater sensitivity to drug use in this period (67).

The endocannabinoid system plays a crucial role in early brain development (68), and undergoes dynamic changes during the adolescent period, indicating that these two developmental windows may be particularly vulnerable to risk factors affecting this system (69). THC activates the endocannabinoid system precisely through the CB1 and CB2 receptors. The CB1 receptors are expressed in high concentrations in the hippocampus, amygdala, basal ganglia, many regions of the cerebral cortex and cerebellum (70,71). Activation of presynaptic CB1 receptors (via THC or endogenous endocannabinoids) suppresses GABA release (72), which can disrupt the development of pyramidal and parvalbumin-containing basket cell circuitry, creating an imbalance in excitatory-inhibitory signaling in the brain. Similar imbalances are thought to underlie psychosis (73), and have been identified in MIA-exposed offspring (74). Cannabis has also been shown to increase dopamine release acutely, which may underlie positive psychotic symptoms (hallucinations, delusions) (8). CB-1 receptors may also play a critical role in synaptic plasticity (75), and neuroinflammation, as CB1 receptors have been identified at low levels within microglial populations (76), and have been observed to affect glial cell function in response to cellular injury (77). CB2 receptors are highly expressed in the peripheral immune system and thought to have potent anti-inflammatory effects, making them interesting candidates for managing neuroinflammation as well (78). More recently, CB2 receptor expression has been identified in the cerebral cortex, hippocampus, striatum, amygdala, thalamic nuclei, periaqueductal grey, cerebellum and several brain stem nuclei of the rodent brain (40,78). These receptors may also play a functionally relevant role in the central nervous system, primarily through microglia (76). These may be interesting avenues for further investigation, as microglial activation has been implicated in many of the downstream neurodevelopmental aberrations due to MIA-exposure.

Few studies exist investigating the combined effects of the two risk factors studied here on offspring development. Previous work has identified decreased CB1 receptor availability due to prenatal MIA-exposure and adolescent cannabinoid exposure in the hypothalamus and sensory cortex, which are regions where we detect sustained alterations due to MIA and THC (79). Alterations to small non-coding microRNAs in the entorhinal cortex, associated with neurotransmission, cellular signaling, and schizophrenia, have been identified in rats exposed to both risk factors (80,81). Finally, alterations to the serotonergic (15) and dopaminergic systems (16) have both been identified in rats exposed to both risk factors; these neurotransmitter systems are relevant to the disease pathology of a number of neuropsychiatric disorders (82,83). These findings suggest that pathological changes may be occurring in response to the combined risk factors in some of the regions where we observe volume alterations. MIA-exposure has been shown to potentiate the effects of other environmental risk factors such as adolescent stress (84) or exposure to drugs of abuse (15), as well as genetic risk factors such as DISC1 (85,86), NRG1 (87), NR4A2, TSC2 (88) causing greater deficits than either exposure alone (5,6).

Finally, our investigation of brain-behaviour covariation identified some underlying sex differences, not clearly uncovered in our univariate analyses. The pattern identifying in adulthood differentiated the male SAL-SAL offspring from all other male offspring exposed to either a single or a combined risk factor, whereas it did not differentiate the female offspring. This was defined by increased cortical, thalamic and cerebellar volume associated with decreased social interactions and sensorimotor gating. The assessment of within subject volume change post-treatment to adulthood with behaviour identified a similar behavioural pattern, associated with the somatosensory cortex striatum, hippocampus, and cerebellum; in this case, male mice were more similar in their expression of the pattern, whereas female SAL-THC offspring different from the other three groups. Overall, this analysis suggests that there may be interesting sex differences underlying the neurodevelopmental abnormalities associated MIA and THC exposure. This dimensional approach may parse variability better than categorical approaches (89) and has proven successful in our previous work (7,17,28)

Sex differences in response to MIA-exposure have been identified in human birth cohorts, showing that males are more likely to develop neurodevelopmental disorders following MIA-exposure to bacterial infection (90). Rodent studies also provide evidence for increased susceptibility in males, with reports of earlier deviations in neuroanatomy and behaviour based on longitudinal studies (49), as well as male-biased deficits in spatial working memory, PPI, and locomotion (91). Finally, altered synaptic function in the hippocampus and abnormal microglia activation have been reported in male MIA-exposed offspring, at a greater level than in females (30,92). Sex-biases have also been observed in response to cannabinoids; some evidence from human studies suggest that males may be more susceptible to the effects of cannabis use if they carry genetic risk for schizophrenia (12), and are also more likely to initiate use earlier than females (93). Rodent studies suggest that females may be more likely to develop behavioural despair, anhedonia, and catalepsy in response to THC treatment (94,95); these behavioural differences may, in part, be explained by differences in THC metabolism (96). Higher CB1 receptor density in males relative to females has been reported in regions such as the striatum, limbic system, and pituitary (97). Underlying sex-differences in several neurobiological systems including microglia and synaptic function associated with MIA-exposures, as well as CB1 receptor density and THC metabolism, may result in the sex-differences in the brain-behaviour alterations we observe.

Based on our multivariate results, we believe that future studies should be performed to specifically investigate sex differences in response to both risk factors, ensuring sufficient power to detect these complex statistical interactions. Further, acquiring MRI and behavioural data more frequently throughout the development of the offspring may allow us to fully investigate subtle changes. This may, in part, explain why we did not detect significant alteration in the MIA-exposed mice, and only detected subtle changes due to THC exposure. Furthermore, performing behavioural assays more proximal to the chronic THC exposure would allow us to uncover more acute behavioral alteration in response to treatment.

We find that exposure to a single risk factor, either prenatal MIA-exposure or chronic adolescent THC exposure are not sufficient to induce lasting neuroanatomical or behavioural changes in adulthood, although transient alterations were observed in adolescent/early-adult neuroanatomy. In contrast, exposure to both risk factors may induce lasting neuroanatomical deviations in some regions, but not others, while only inducing subthreshold behavioural alterations. Interestingly, exposure to either prenatal MIA or adolescent THC, or both, significantly decreases the expression of CB1 and 2 receptors in the brain. This points to the central role of the endocannabinoid system in both prenatal and adolescent brain development, as well as its sensitivity to disruptions due to risk factor exposure. In conclusion, our findings show that exposure to multiple risk factors may have greater effects on offspring brain volume throughout development than exposure to a single risk factor, but that CB1 and 2 receptor density is similarly decreased following exposure to a single or multiple risk factors. Finally, exposure to either one or two risk factors did not alter adult behaviour. These findings may further our understanding of how exposure to risk factors throughout development could increase the likelihood of developing neuropsychiatric illnesses.