Divergent opioid-mediated suppression of inhibition between hippocampus and neocortex across species and development

μORs are selectively enriched in hippocampal PV-INs.

μORs are expressed throughout hippocampus (HPC) and neocortex (CTX)¹, and within the HPC are highly enriched in PV-INs^28,29. However, less is known about the specificity of μOR expression to PV-INs throughout CTX. To address this, we compared the expression of Oprm1 (encoding μORs) across HPC and CTX in GABAergic INs via single-nucleus RNA sequencing (snRNAseq) of postnatal day (P)28 mice, observing an enrichment of Oprm1 in HPC versus CTX within the delineated PV cell cluster (Fig. 1A–B). We next assessed the spatial colocalization of mRNA via RNAscope, observing a decreased percentage of Pvalb+ somata colocalized with Oprm1 in neocortical pyramidal cell layers (both supragranular layers 2/3 and infragranular layers 5/6) in primary motor (M1), somatosensory (S1) and visual (V1) cortex relative to hippocampal pyramidal cell layers in CA (CA1-CA3, Fig. 1C–D). Notably, in CTX we observed several high-expressing Oprm1+Pvalb- cells (Fig. 1C, open arrow), cells we did not encounter in HPC. To assess if transcript associated with the translational machinery was altered, we employed a RiboTag sequencing approach of all MGE-derived interneurons (MGE-INs, of which PV-INs comprise a substantial fraction⁴⁷), employing Nkx2.1^Cre/+:Rpl22(RiboTag)^HA/HA mice^48–50. We observed a selective enrichment of Oprm1 in hippocampal MGE-INs relative to neocortical MGE-INs and all bulk hippocampal/neocortical tissue (Fig. 1E).

At the protein level, we quantified PV-μOR colocalization via immunohistochemistry (IHC) by staining for PV in P70 μOR-mCherry mice (Fig. 1F). μOR-mCherry mice were employed as their expression colocalized with μOR antibodies but provided more robust somatic labeling (Fig. S1A-B). We observed an increased colocalization of PV and μOR somata across all layers of hippocampal proper (CA1, CA2, CA3) relative to primary cortical regions (M1, S1, V1) and higher-order prelimbic cortex (PrL) (Fig. 1G–H). Hippocampal dentate gyrus (DG) exhibited intermediate PV-μOR colocalization relative to hippocampus proper and neocortex.

As hippocampal MGE-INs are born earlier than neocortical MGE-INs (embryonic day (E)11.5⁴⁷ versus E13.5^51,52), we next sought to determine if embryonic birthdate of PV-INs contributed to regional enrichment of Oprm1. Through EdU proliferation labeling across E11–15 (Fig. S2A-C), we replicated prior observations that hippocampal PV-INs are born earlier than neocortical PV-INs (Fig. S2D). This large biological replicate RNAscope study also fully supported the above RNAscope evaluation of weak Oprm1 expression within cortical versus hippocampal PV-INs. However, later-born PV-INs were no more likely to express Oprm1 (Fig. S2E), suggesting factors other than embryonic birthdate establish the hippocampal-neocortical divergence in PV-IN μOR expression.

Hippocampal PV-INs are selectively hyperpolarized by opioids.

μORs hyperpolarize neurons and reduce neurotransmitter release via combined interactions with inward-rectifying K⁺ channels and closure of presynaptic Ca²⁺ channels^9,10. To determine whether a functional divergence existed between hippocampal and neocortical PV-INs, we recorded in a whole-cell configuration, currents elicited by the administration of μOR agonist DAMGO (100 nM) and antagonist CTAP (500 nM) in PV-tdTomato^+/− mice (Fig. 2A–B). Consistent with prior studies^25,26, we observed an outward hyperpolarizing current from DAMGO administration in CA1 PV-INs voltage-clamped to −50 mV (ΔV = +60 ± 12 pA, Fig. 2C, D₁-F₁). However, no significant change in holding current was observed in V1 PV-INs (+12 ± 5 pA, Fig. 2C, D₂-F₂). Of the CA1 PV-INs with sufficiently recovered morphology to assess axonal target (n_cell = 5 of 11), 3 exhibited perisomatic-targeting basket cell (BC) morphology, 1 bistratified (BSC) morphology targeting apical and basal dendrites, and 1 axo-axonic (AAC) morphology targeting the axon-initial segment (AIS) of PCs. The hyperpolarizing DAMGO-elicited current was observed in all PV-IN subpopulations (BC: +73 ± 37, BSC: +85, AAC: +101 pA), which is of interest as prior research has suggested AACs are less sensitive than BCs to DAMGO following carbachol administration⁴⁶. V1 PV-INs had somata spanning cortical layers 2 through 6, and of those with sufficiently labeled axons (n_cell = 6 of 15), 5 exhibited BC and 1 AAC morphology, with neither subpopulation responsive to DAMGO. In CA1 PV-INs, there was also a decrease in input resistance (Fig. 2G1), consistent with the opening of ion channels, which was not present in V1 PV-INs (Fig. 2G2). We next substituted GTP in the internal pipette solution with the non-hydrolyzable GTP analogue GTPγS, which binds to and occludes further GPCR activity⁵³, and observed no DAMGO-mediated change in holding current in either CA1 or V1 (Fig. 2H–J), indicating a direct mechanism of GPCRs expressed in the recorded PV-INs. Together, these findings point to a transcriptional, translational, and functional enrichment of μORs within hippocampal versus neocortical PV-INs.

Opioids selectively suppress hippocampal proper but not neocortical PV-IN mediated inhibition.

To determine if the selective enrichment of μORs within hippocampal PV-INs can reduce inhibitory synaptic transmission, we adopted an optogenetic approach to light-activate PV-INs and record their output in downstream pyramidal cells (PCs) in PV^Cre/+:ChR2^fl/+ mice (Fig. 3A–B). GABA_A-mediated currents were pharmacologically isolated with bath administration of AMPA, NMDA, and GABA_B antagonists (μM: 10 DNQX, 50 DL-APV, and 1 CGP55845 respectively), with drugs of interest (nM: 100 DAMGO, 500 CTAP) bath administered for at least 5 minutes. Consistent with prior hippocampal studies^25,26, light-evoked inhibitory post-synaptic currents (leIPSCs) were suppressed by DAMGO in CA1-PCs (Fig. 3C). To assess a pre- or post-synaptic mechanism, we analyzed the coefficient of variation (Fig. 3C), and in a subset of cells, the paired pulse ratio (Fig. 3D), observing an increase in both metrics (Fig. 3E–F), consistent with a presynaptic mechanism and prior studies^25,27. Notably however, PV-IN output was differentially modulated between HPC and CTX, with significant suppression (64 ± 5% of baseline) observed in CA1-PCs and no suppression (108 ± 6%) observed in M1-PCs (Fig. 3G). To broadly assess this across cortical regions, we recorded leIPSCs in three hippocampal (CA1, CA3, DG) and four neocortical regions (M1, PrL, S1, V1). Additionally, as differences in PV connectivity have been reported between radiatum (rad.)- adjacent superficial and oriens (or.)-adjacent deep PCs (sPCs and dPCs respectively)^54,55, we segregated recordings from these cell populations. DAMGO suppressed leIPSCs in both CA1-sPCs (55 ± 8%, Fig. 3H) and CA1-dPCs (72 ± 7%, Fig. 3I), with no difference between these cell populations (p = 0.135, Fig. S3A). Both in CA1-sPC and CA1-dPC recordings, CTAP reversed the DAMGO-mediated suppression, although this could be achieved through wash alone (data not shown). DAMGO also suppressed leIPSCs in CA3-PCs (69 ± 6%), although this did not fully reverse with either CTAP or wash (Fig. 3J). Interestingly, this long-lasting μOR-mediated suppression in CA3 is similar to observations of a long-lasting δOR-mediated suppression in CA2 that was not observed in CA1⁵⁶. In DG granule cells (DG-GCs), we observed a highly variable response, with no significant group effect of DAMGO or CTAP (95 ± 8%, Fig. 3K). Throughout neocortex, we also observed highly variable responses with no significant group effects of either DAMGO or CTAP in M1 (108 ± 6%, Fig. 3L), PrL (96 ± 8%, Fig. 3M), S1 (107 ± 9%, Fig. 3N), and V1 (89 ± 11%, Fig. 3O). Neocortical PCs were recorded in both supragranular (2/3) and infragranular (5/6) layers. To address the possibility that differing opioid sensitivity between cortical layers contributes to the increased variability in the observed leIPSC amplitude, we combined primary cortex (M1, S1, V1) PCs and segregated the combined data across L2/3 and L5/6, with no differences observed in the baseline normalized DAMGO response (p = 0.291, Fig. S3B). We also explored the increased variability of cortical responses by removing outliers (ROUT method) and all cells with greater than 150% increase from baseline, but still observed no significant group effects (data not shown).

We next examined the role of sex, as there have been reported differences in morphine response⁵⁷ and μOR trafficking in hippocampal PV-INs^58,59. Most regions were insufficiently powered to segregate along sex. However, by combining hippocampal proper (CA1, CA3) and primary neocortex (M1, S1, V1), we achieved sufficiently powered groups to analyze via 2-way ANOVA the effects of sex and region. A significant effect of region was found with sex not reaching significance (Fig. S3C). Thus, sex did not appear to be a driving confounder behind the observed hippocampal-neocortical differences.

The concentration of DAMGO was selected to be consistent with prior literature²⁵, but to determine if the lack of cortical effect was due to lower affinity in these regions, we increased the DAMGO concentration 10-fold (1 μM) and observed similar results, with suppression in CA1 (58 ± 8% of baseline) and none in DG (96 ± 11%) or M1 (108 ± 9%, Fig. S3D). These findings were not unique to the selective μOR agonist DAMGO, as the less selective and less potent μOR agonist morphine (10 μM) had similar effects in CA1 (56 ± 10%) and S1 (110 ± 12%, Fig. S3E). To test if endogenous opioids were tonically suppressing PV-IN release, we applied neutral antagonist CTAP preceding DAMGO, observing no increase from baseline in CA1 (83 ± 10%) or S1 (99 ± 13%), nor any effect from subsequent DAMGO administration with CTAP still applied (Fig. S3F). As μORs and δORs can become constitutively active, in which they activate G proteins even in the absence of agonist^60,61, we also tested the antagonist naloxone, which can function as an inverse agonist after opioid exposure^62,63. We observed no increase from baseline in PV-IN output from 1 μM naloxone in CA1 (110 ± 18%) or S1 (81 ± 11%, Fig. S3G). All drug conditions are summarized in Fig. S3H.

Although we previously observed a DAMGO-mediated hyperpolarizing current in one morphologically identified PV-AAC, we wished to more thoroughly explore this cell population due to prior reports of divergent opioid response amongst PV-BC and PV-AAC populations⁴⁶. To explore this, we employed the novel BiPVe4(AAC) enhancer virus (see Fig. S4 and⁶⁴ for validation), which labels AACs with closely matched features as other strategies to label this subpopulation^65,66. By injecting WT mice with pAAV-BiPVe4(AAC)-ChR2-mCherry, we were able to observe a robust DAMGO suppression of AAC-mediated inhibition in CA1 (47 ± 5%, Fig. 3P–R). Moreover, we confirmed DAMGO-mediated outward currents in a larger cohort of AACs (Fig. S4F-H). Critically, suppression of PV-IN inhibition was directly mediated by μORs expressed within PV-INs, as we observed no DAMGO-mediated suppression (92 ± 6%) following selective knockout of Oprm1 in PV-INs through breeding PV^Cre/+:Oprm1^fl/fl and probing output from these mice after injection with the red-shifted opsin C1V1 under control of Cre-recombinase (Fig. 3S–U and Fig. S5)

δOR activation and SST-IN inhibition exhibit similar functional HPC-CTX divergence.

Hippocampal PV-INs are also enriched in δORs, with μORs and δORs functioning through partially occlusive downstream pathways to hyperpolarize interneurons²⁷. In contrast to our observed lack of Oprm1 expression in neocortical PV-INs, we observed a high level of Oprd1 transcript within PV cells in our snRNAseq dataset (Fig. 4A–B), as well as high neocortical Pvalb-Oprd1 colocalization via RNAscope, albeit significantly less than hippocampal Pvalb-Oprd1 colocalization (Fig. 4C–D). Our RiboTag dataset revealed consistent findings; both hippocampal and neocortical MGE-INs were enriched in Oprd1, with higher enrichment in hippocampal versus neocortical MGE-INs (Fig. 4E). We next applied the δOR selective agonist DPDPE (500 nM) in PV^Cre/+:ChR2^fl/+ mice (Fig. 4F), observing a significant suppression in CA1 (76 ± 5%) that did not reach significance in V1 (88 ± 6%, Fig. 4G). However, when analyzing all normalized drug responses with a less conservative 1-way ANOVA rather than a repeated-measures ANOVA of the raw data, we found that across all drug conditions in CTX, only DPDPE resulted in a significant suppression (Fig. S3H). Thus, δOR activation may modestly suppress neocortical PV-IN output.

While we have thus far focused on PV-INs, Somatostatin-expressing interneurons (SST-INs) comprise a substantial portion of MGE-INs⁴⁷, and within the HPC also express μORs and δORs²⁹. Both in HPC and CTX, snRNAseq revealed an apparent enrichment of Oprm1 and Oprd1 within SST-INs (Fig. 1A, Fig. 4A,B). We undertook similar optogenetic experiments in SST^Cre/+:ChR2^fl/+ mice (Fig. 4H–I), observing that as with PV-INs, leIPSCs from SST-INs were reversibly suppressed by μOR agonist DAMGO in CA1 (49 ± 5%) and not in S1 (99 ± 12%, Fig. 4J). δOR agonist DPDPE resulted in a non-significant suppression in CA1 SST-IN output (62 ± 12%) with no observable suppression in S1 (101 ± 8%, Fig. 4K). Thus, hippocampal SST-INs appear to exhibit a similar functional specialization as PV-INs regarding opioid response, despite an apparent enrichment of μORs and δORs within neocortical SST-INs. The lack of a cortical effect suggests altered subcellular localization with limited opioid receptor expression at neocortical interneuron presynaptic sites to regulate synaptic transmission in a manner similar to hippocampal PV/SST-INs.

Prior research into the effects of systemic morphine administration on PrL observed microcircuit alterations distinct from our acute DAMGO findings (Fig. 3M), with morphine inducing weakened μOR-dependent PV-IN→PC and strengthened δOR-dependent SST-IN→PV-IN inhibition⁶⁷. The surprising δOR-mediated increase in SST-IN output was linked to an upregulation of Rac1 and Arhgef6 specifically in SST-INs. While it is unclear if these downstream plasticity mechanisms occur in our acute DAMGO/DPDPE administrations, the high expression of δORs within neocortical PV-INs/SST-INs and inhibitory interconnections between these populations could contribute to the highly variable neocortical opioid modulation that we observed.

Opioids selectively suppress hippocampal synaptic inhibition in nonhuman primates and resected human tissue.

To determine if μOR-mediated disinhibitory motifs are present in primate species, we turned to the study of rhesus macaques and resected human tissue. ISH RNAscope analysis of adult macaque tissue revealed that, consistent with mice, the percentage of Pvalb+ cells co-expressing Oprm1 was strongly enriched in CA1 (85.4 ± 4.7%), in contrast to neighboring temporal cortex (16.0 ± 5.3%, Fig. 5A–B). To functionally target PV-INs in macaques, we adopted a viral enhancer strategy. Adult macaques were injected with pAAV(PHP.eB)-S5E2-ChR2-mCherry in HPC or M1, an enhancer targeting PV-INs⁴⁴, which we found to be strongly colocalized with stained PV-INs (Fig. S6A-B). Consistent with our observed PV overlap, S5E2 leIPSCs in acute macaque slices were insensitive to the N-type Ca²⁺ channel blocker ω-conotoxin (95 ± 7%, Fig. S6C), robustly suppressed by the P/Q-type Ca²⁺ channel blocker ω-agatoxin (10 ± 4%, Fig. S6D), and unaffected by the synthetic cannabinoid agonist WIN 55212–2 (97 ± 5%, Fig. S6E), indicating that primate PV-INs utilize similar synaptic mechanisms as in mice. We next assessed the effect of μOR drugs on leIPSCs (Fig. 5C–D). As observed in mice, CA1 leIPSCs were reversibly suppressed by DAMGO (74 ± 3%, Fig. 5E). CA3 leIPSCs exhibited a long-lasting though non-significant DAMGO suppression (66 ± 9%), which was not easily reversed with CTAP or wash (Fig. 5F). DAMGO elicited a highly variable response with no significant suppression in both DG (102 ± 4%, Fig. 5G) and M1 (111 ± 7%, Fig. 5H).

We next assessed the effect of DAMGO/CTAP on resected human tissue from patients with drug-resistant epilepsy (Fig. 5I). We recorded spontaneous IPSCs (sIPSCs) in CA-PCs (CA1 & CA3) and temporal cortex PCs (Fig. 5J). Although sIPSCs include GABAergic currents from all inhibitory interneurons, perisomatic-targeting fast-spiking PV-INs are expected to be overly represented in this measure. Of note, several resections exhibited hippocampal sclerosis within CA1, observable as a marked reduction in the width of str. pyr. with few PCs available for patch clamp electrophysiology. Such sclerosis was likely related to the site of epileptogenesis, and sclerotic resections were not included for this study. Neighboring temporal cortical tissue was surgically removed to reach hippocampal structures and was not pathological. Notwithstanding these limitations, human sIPSCs exhibited a markedly similar regional divergence in DAMGO-mediated suppression, with a significant suppression in CA-PCs (63 ± 8%, Fig. 5K) and no suppression of CTX-PCs (111 ± 8%, Fig. 5L). Thus, although we cannot rule out that pathological activity was present in our human hippocampal resections, the remarkable similarity to healthy murine and nonhuman primate tissue would suggest that opioid-mediated suppression of inhibition was at least unimpaired. Comparing across species, this hippocampal-neocortical divergence was remarkably well-conserved across mice (Fig. 5M), macaques (Fig. 5N), and humans (Fig. 5O). An emerging model for this observed difference is that hippocampal PV-INs are enriched in μORs relative to neocortex, resulting in opioid-mediated hyperpolarization and presynaptic suppression of neurotransmission (Fig. 5P).

Tac1 cells, as a proxy for immature PV-INs, are suppressed by μOR agonists in HPC.

We next explored whether PV-INs exhibit regional opioid divergence in early development. PV-IN function has traditionally been difficult to study in early development as PV itself is not well-expressed until P10 in mice⁶⁸. We undertook an alternate approach from PV^Cre mice, with the observation that tachykinin precursor 1 (Tac1, which with post-translational modification produces substance P and neurokinin A) is co-expressed in and can be used as a marker for immature PV-INs^69–72. To confirm the utility of Tac1 as a marker for immature PV-INs, we examined via snRNAseq of hippocampal interneurons from P10 mice, the expression of principal markers delineating interneurons (Fig. 6A). MGE-INs were distinguished from CGE-INs via expression of Lhx6, while SST and Lamp5 (labeling neurogliaform/Ivy cells) labeled clusters of MGE-INs. Pvalb at this age was expressed at low levels with high overlap with Tac1. Amongst MGE subgroups at this developmental stage, the future PV cluster was enriched in Tac1 relative to SST and Lamp5 clusters (Fig. 6B). RiboTag evaluation confirmed that Tac1 associated with the translational machinery was enriched in both HPC and CTX relative to bulk tissue, and in HPC as early as P5 (Fig. 6C).

We next conducted IHC for PV in P5, P8, P12, and P45 Tac1^Cre/+:tdTom^fl/+ mice (Fig. 6D–E). As previously reported, PV was minimally expressed in P5 and P8 mice but was detectable within CA1 str. pyr. somata and perisomatic axonal terminals by P12. In contrast, Tac1 somata were readily detectable at all ages, with perisomatic axonal terminals detectable and highly colocalized with PV across hippocampal layers by P12 (Fig. S7A-B). We quantified somatic colocalization by counting PV+ and Tac1+ cells across hippocampal and neocortical (S1) layers, observing that the percentage of PV-INs co-expressing Tac1 rapidly increased to 86 ± 2% by P12 (Fig. 6G, left), coinciding with the emergence of labeled PV-INs (Fig. 6H, left). CA1 PV co-expression with Tac1 was maintained in adulthood at 81 ± 2% (Fig. 6G, middle). Across subregions and layers, the majority of PV-INs co-expressed Tac1 in HPC (70 ± 4%) and S1 (58 ± 6%). The inverse measure, the percentage of Tac1 cells that co-expressed PV, was somewhat lower, rising to ~50% throughout HPC and S1 in adulthood. Some putative Tac1+ PCs were observed and readily distinguished by morphology and bright fluorescent labeling of dendrites, perhaps representing a limitation of Tac1^Cre/+:tdTom^fl/+ mice as they appeared clustered in irregular patches in a minority of sections (20/49 across all ages). Putative Tac1_PCs represented a minority fraction, appearing in principal cell layers and comprising 5.9% of all Tac1+ labeled cells. However, their presence warrants caution if employing Tac1^Cre mice for functional studies of immature PV-INs without isolating GABAergic transmission or adopting an intersectional approach. Additionally, as there is a reported subpopulation of Tac1-expressing SST-INs⁷³, we also stained for SST (Fig. 6F). We observed only a small minority of SST cells co-expressing Tac1 and Tac1 cells co-expressing SST (5–8%, Fig. 6G,H, right), consistent with the limited expression of Tac1 in the SST cluster of our snRNAseq data and prior patch-seq assessment⁷⁴.

We also assessed Tac1-PV colocalization in adult Tac1^Cre/+:ChR2^fl/+ mice, observing that across all layers of CA1 the percentage of PV cells co-expressing Tac1 was 48 ± 3%, while the percentage of Tac1 cells co-expressing PV was 74 ± 3% (Fig. S7C-D). We did not observe putative Tac1_PCs in this cohort, indicating it may be a reporter-specific observation. These colocalization ratios were effectively inverted from the Tac1^Cre/+:tdTom^fl/+ colocalization, likely pointing to the limitations of IHC. ChR2 is a poor somatic label even with immunoboosting, and as PV levels themselves are dynamically regulated^75,76, several PV-INs likely express sub-threshold levels of PV. In a third IHC cohort, we confirmed that hippocampal Tac1 cells were enriched in μOR receptors by triple labeling Tac1^Cre/+:μOR-mCherry mice injected perinatally with Cre-dependent pAAV(AAV5)-pCAG-FLEX-EGFP-WPRE and staining against PV (Fig. S7E) at P13 and P45. Tac1+PV+ at both ages were >50% colocalized with μORs, while Tac1+PV- were non-significantly less colocalized with μORs (~40%). However, Tac1+ cells overall were enriched in μORs relative to Tac1- cells (Fig. S7F).

To more directly assess Tac1 cells as a method to target immature PV-INs, we recorded intrinsic parameters from 64 Tac1+ cells from P8–11 Tac1^Cre/+:tdTom^fl/+ mice (Fig. 6I), including from CA1 (33), CA2 (7), CA3 (13), DG (3), and CTX (7). Of these recorded Tac1+ cells, we were able to morphologically reconstruct 43. Of the hippocampal Tac1 cells with visible axon, 24/35 (69%) exhibited extensive perisomatic-targeting axonal morphology consistent with basket cells, 8/35 (25%) exhibited a bistratified morphology, 3/35 (9%) targeted only superficial dendritic layers rad. and l.m., and none exhibited PC morphology (Fig. 6J, Fig. S8). Targeted hippocampal Tac1-IN somata were primarily located within str. pyr., while neocortical Tac1-INs spanned pyramidal cell layers. For comparison of electrophysiological properties, we also recorded from 14 non-labeled putative PCs in HPC, observing striking differences in spiking properties of Tac1-INs versus PCs, consistent with a fast-spiking phenotype expected from PV-INs (Fig. 6K, Fig. S8A-D). We did not observe any regional differences between Tac1-INs (Table S2). To characterize the development of these cells, we also recorded from an additional 13 P16–18 Tac1-INs, observing maturation to a faster spiking phenotype approaching that of 10 recorded PV-INs from mature PV-tdTom mice (Fig. S8E-H). Overall, the developmental trajectory of recorded Tac1-IN physiology closely matched progression from immature to mature basket cell electrophysiological properties⁷⁷.

To assess the opioid sensitivity of Tac1 cells, we applied 100 nM DAMGO to 9 of the recorded hippocampal Tac1 cells in P8–11 mice, observing a reversable hyperpolarizing change in holding current (ΔV = +17 ± 4 pA, Fig. 6L–M). Next, we performed paired recordings of 7 CA1 and 3 CA3 Tac1-IN→PC pairs (Fig. 6N–O) in P10–12 mice, with connected pairs comprising 15% of attempted pairs, observing a strong suppression of transmission in both regions (CA1: 37 ± 5%, CA3: 37 ± 2%, Fig. 6P) and consistent with adult PV-IN→PC pairs25. Utilizing Tac1^Cre/+:ChR2^fl/+ mice (Fig. 6Q–R), we further explored the opioid modulation of inhibitory output across regions. In P5–7 mice we observed that GABAergic-isolated leIPSCs were reversibly suppressed by DAMGO in both CA1 (47 ± 4%) and CA3 (34 ± 4%, Fig. 6S). Within S1-L5, we observed that functional Tac1-IN→PC light-evoked synaptic responses were often small and unreliable before the age of ~P8, thus we included in our study slightly older mice with a full age range from P7–10. Consistent with adult PV-INs, we did not observe a significant suppression of Tac1 mediated inhibition (89 ± 10%, Fig. 6S). Thus, at the earliest timepoints just as immature PV-IN→PC synaptic connections are being established, there is opioid receptor dependent modulation within the hippocampus not found in neocortex.

Opioids and Tac1 cells regulate spontaneous activity of the developing hippocampus.

During early postnatal development between P5–10, the principal network signature is that of giant depolarizing potentials (GDPs). These spontaneous synchronous events occur during this critical developmental window while GABA is depolarizing⁷⁸ and are believed to play an important role in hippocampal synaptogenesis⁷⁹. To test the importance of opioids in regulating these events, we recorded spontaneous GDP associated currents (GDP-Is) intracellularly from CA3-PCs voltage-clamped to 0 mV to isolate GABAergic contributions, in which GDP-Is are readily detectable as outward events due to a barrage of GABA_A currents (Fig. 7A–B). GDP-I rate was highly variable across brain slices, so to ensure a sufficient number of events for averaging we only considered slices with ≥ 4 baseline GDPs (69/105 slices across all conditions). We next applied 100 nM DAMGO and observed a robust and reversible decrease in GDP-I event frequency (32 ± 6% of baseline), with no change in GDP-I amplitude (Fig. 7C). DAMGO also suppressed GDPs within S1 cortex (21 ± 9%), which we observed occurring at a lower rate than in CA3 (Fig. S10A-B). δOR agonist DPDPE could also suppress CA3 GDPs (55 ± 13%, Fig. S10C-D), while μOR antagonist CTAP had no effect (104 ± 16%, Fig. S10E-F). Thus, opioid agonists potently suppress spontaneous network activity of the developing brain, analogous to the effect of DAMGO on spontaneous sharp wave ripples (SWRs) in adulthood⁴⁵. To determine if Tac1 cells could subserve this suppressing role, we optogenetically silenced Tac1 cells with Tac1^Cre/+:ArchT^fl/+ mice (Fig. 7D–E), observing that GDP-Is were significantly suppressed (76 ± 8% of baseline, Fig. 7F), with no effect on sIPSCs (Fig. S10G). Moreover, Tac1 cell silencing occluded the effect of DAMGO, as DAMGO produced no further reduction in GDP-I frequency (97 ± 16% compared to light-on period, Fig. 7G–I). Prior work from our lab identified MGE-INs as key regulators of GDP activity; optogenetically silencing MGE-Ins reduces GDP-I frequency to 34 ± 6% of baseline⁸⁰. Several studies have attributed GDP regulation and generation to dendritic-targeting SST-INs^81,82. However, optogenetically silencing SST-INs results in only modest reductions of GDP-I frequency (71–80% of baseline)⁸³. Thus, to account for the entire effects of DAMGO-mediated suppression and optogenetic MGE-IN silencing, other IN subpopulations likely contribute. This study presents for the first-time evidence that Tac1+ immature PV-INs play a key role in primordial hippocampal rhythmogenesis, consistent with the extensive and functional perisomatic innervation already provided by these cells at this stage of development (Fig. 6I–P, Fig. S8) and the essential role adult PV-INs play in hippocampal rhythmogenesis. The limited information regarding immature PV-IN participation in developmental hippocampal rhythmogenesis likely reflects a prior lack of viable animal model to target these cells. Importantly, the suppression of immature PV-INs by opioids at these early developmental time-points has severe implications for the harmful aspects of opioid use on synaptogenesis and circuit development in the developing brain.

Morphine pre-treatment occludes acute DAMGO suppression.

Exposure to exogenous opioids can result in a shift in μOR and δOR function to become more constitutively active in chronic⁸⁴ and withdrawal conditions^63,85, which is believed to contribute to tolerance and dependence. Drug-induced alterations to the endogenous opioid system have not been well-studied in the hippocampus. One study found chronic morphine treatment results in altered RNA levels of Penk and Gnas (encoding the G_αs subunit of GPCRs)⁸⁶, although it is unclear if these changes are specific to distinct neuronal populations. Therefore, to determine if the opioid-mediated suppression of PV-INs is altered after morphine exposure, we injected morphine or saline in PV^Cre/+:ChR2^fl/+ mice prior to acute slice preparation and whole cell patch clamp recordings of leIPSCs. We used three treatment regimens (Fig. 8A), each with an independent saline control group, including a bolus injection (15 mg/kg), a chronic treatment with increasing doses over 6 days (15, 20, 25, 30, 40, 50 mg/kg), and a withdrawal group receiving the chronic treatment but with acute slice preparation 72 hours instead of 1 hour after the final injection as with other groups. Optogenetic leIPSC recordings were performed in CA1-PCs as previously described. Saline controls all exhibited significant DAMGO-mediated suppression of leIPSCs (Fig. 8B) similar to un-injected mice (cf. Fig. 3H–I). In morphine-injected mice, the acute DAMGO-mediated suppression was partially occluded, particularly in the withdrawal group for which DAMGO no longer significantly deviated from baseline (Fig. 8C). Comparing the normalized DAMGO response across all groups (Fig. 8D), the bolus and withdrawal groups exhibited a significant deviation from their saline control, which was somewhat moderated in the chronic group, potentially through long-term adaptations distinct from the withdrawal conditions where constitutive receptor activity has been described^63,85. Although the physiological role of the opioid-mediated suppression of hippocampal PV-INs is not yet fully understood, these data indicate that this disinhibitory motif becomes dysregulated after morphine use. Exogenous opioids, in addition to well-described roles in analgesic and stress response, can therefore interfere with neuromodulation of hippocampal PV-INs, leading to disruption of rhythmic activity such as GDPs (Fig. 7), SWRs⁴⁵, and gamma oscillations⁴⁶, activities essential for learning and memory in the developing and adult brain.

RESOURCE AVAILABILITY

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

METHOD DETAILS

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistical analysis was conducted in Graphpad Prism 10. All data were tested for normality and lognormality with Shapiro-Wilk tests. Parametric tests were selected if all groups were normal. If at least one group was non-normal and all groups were log-normal, data were log-transformed prior to parametric testing. If data were neither normal nor log-normal, non-parametric tests were used whenever available. Throughout, summary data are presented as mean ± SEM with symbols representing individual values. Additional descriptive statistics, normality tests, hypothesis testing, and post hoc comparisons are available for each figure in Table S1.

Cellular responses to drug treatments were analyzed with a 1-way repeated-measures design (ANOVA/Friedman, with Tukey’s/Dunn’s post hoc multiple comparisons). All statistical tests were conducted on raw non-normalized data, including plots showing responses normalized to baseline. Experiments missing at most one condition (CTAP/naltrindole) were included and analyzed with a mixed-effects model. Comparisons of the normalized DAMGO responses between regions were performed with a 1-sample t-test/Wilcoxon, comparing to 1 (baseline) and correcting sigma for the number of comparisons. Comparisons between different brain regions’ drug responses over time were analyzed with a 2-way design (region × time), with Šídák’s multiple comparisons restricted to comparisons between time-equivalent bins. In general, post hoc tests for 2-way analyses were restricted to comparisons of a priori interest, employing Šídák’s multiple comparisons tests, or if all comparisons of interest with Tukey’s multiple comparisons tests.