Abstract
Opioid use by pregnant women results in neonatal opioid withdrawal syndrome (NOWS) and lifelong neurobehavioral deficits including language impairments. Animal models of NOWS show impaired performance in a two-tone auditory discrimination task, suggesting abnormalities in sensory processing in the auditory cortex. To investigate the consequences of perinatal opioid exposure on auditory cortex circuits, we administered fentanyl to mouse dams in their drinking water throughout gestation and until litters were weaned at postnatal day (P)21. We then used in vivo two-photon Ca2+ imaging in adult animals of both sexes to investigate how primary auditory cortex (A1) function was altered. Perinatally exposed animals showed fewer sound-responsive neurons in A1, and the remaining sound-responsive cells exhibited lower response amplitudes but normal frequency selectivity and stimulus-specific adaptation (SSA). Populations of nearby layer 2/3 (L2/3) cells in exposed animals showed reduced correlated activity, suggesting a reduction of shared inputs. We then investigated A1 microcircuits to L2/3 cells by performing laser-scanning photostimulation (LSPS) combined with whole-cell patch-clamp recordings from A1 L2/3 cells. L2/3 cells in exposed animals showed functional hypoconnectivity of excitatory circuits of ascending inputs from L4 and L5/6 to L2/3, while inhibitory connections were unchanged, leading to an altered excitatory/inhibitory balance. These results suggest a specific reduction in excitatory ascending interlaminar cortical circuits resulting in decreased activity correlations after fentanyl exposure. We speculate that these changes in cortical circuits contribute to the impaired auditory discrimination ability after perinatal opioid exposure.
SIGNIFICANCE STATEMENT This is the first study to investigate the functional effects of perinatal fentanyl exposure on the auditory cortex. Experiments show that perinatal fentanyl exposure results in decreased excitatory functional circuits and altered population activity in primary sensory areas in adult mice. These circuit changes might underlie the observed language and cognitive deficits in infants exposed to opioids.
Introduction
From 2018 to 2019, the ongoing opioid epidemic in the United States claimed nearly 50 000 lives from opioid overdose deaths, and of these lives, >36 000 deaths involved synthetic opioids like fentanyl (Mattson et al., 2021). Opioid use by pregnant women results in neonatal opioid withdrawal syndrome (NOWS) and long-lasting deficits that result from the prenatal and perinatal opioid exposure (Benninger et al., 2020; Patrick et al., 2020), including sensory-related deficits (Ornoy et al., 2001; Kivisto et al., 2015) such as language and cognitive deficits (Kocek et al., 2016; Benninger et al., 2020), suggesting sensory areas are affected by opioid exposure. The findings of NOWS in humans have been validated in rodent models, which allow the identification of neural circuit changes (Alipio et al., 2021a). Perinatal fentanyl exposure in mice results in impaired auditory discrimination and lower levels of engagement in auditory tasks (Alipio et al., 2021a). Perinatal fentanyl exposure also leads to an imbalance of spontaneous vesicle release with an overall reduction in excitation and slight increase in inhibition, changes that are echoed in evoked synaptic properties and diminished dendritic complexity in the somatosensory cortex (Alipio et al., 2021b). Prenatal morphine exposure also inhibits dendritic arborization of cortical pyramidal neurons (Mei et al., 2009). Together these changes are consistent with reduced excitatory cortical circuits and suggest abnormal cortical sensory processing after prenatal and neonatal opioid exposure. Given that perinatal fentanyl exposure results in deficits in auditory discrimination behavior (Alipio et al., 2021a), we speculated that perinatal fentanyl exposure caused altered sound processing in auditory cortex because of altered auditory cortical circuits.
To investigate this possibility, we used a model of perinatal fentanyl exposure (Alipio et al., 2021a,b) and studied the functional and circuit changes of the adult primary auditory cortex (A1). We administered fentanyl in the drinking water of pregnant mouse dams throughout gestation and until litters were weaned at postnatal day (P)21. Since we were interested in persistent effects of the early fentanyl exposure, we subsequently raised animals for approximately one year in our colony. We then investigated whether sound processing was altered in fentanyl exposed animals by using in vivo two-photon imaging of layer 2/3 (L2/3) neurons in A1. Imaging revealed fewer sound-responsive cells and that sound-evoked responses had reduced amplitude. Moreover, nearby L2/3 cells had reduced pairwise activity correlations. We next examined the functional circuits to A1 L2/3 pyramidal cells in vitro by using whole-cell patch-clamp recordings and laser-scanning photostimulation (LSPS). We found that L2/3 cells in exposed animals received fewer excitatory inputs from layers L4 and L5/6, indicating interlaminar hypoconnectivity. In contrast, inhibitory connections were normal. These results suggest a specific reduction in ascending excitatory interlaminar cortical circuits after fentanyl exposure, consistent with the observed reduction in responsiveness and pairwise correlation. We speculate that the reduced activity in auditory cortex contributes to the auditory impairments after perinatal opioid exposure.
Materials and Methods
All procedures were conducted in accordance with the Guide for the Care and Use of Laboratory Animals and approved by the Institutional Animal Care and Use Committees at the University of Maryland School of Medicine and University of Maryland, College Park.
Animals
Both male and female adult mice were used (12–15 months, N = 24 for in vitro, N = 10 for in vivo). We used adult offspring of male CBA (JAX: 000654) mice crossed with female Thy1-GCaMP6s (on C57BL/6J background, JAX: 024275) since CBA strains do not exhibit age-related auditory deficits (Frisina et al., 2011; Bowen et al., 2020). As performed in our prior studies, fentanyl was added to the water hydration pouches when copulatory plugs were identified (Alipio et al., 2021a,b). Pouches were replenished weekly until litters were weaned at P21. Offspring were housed two to five per cage, in single-sex groups, in a temperature-controlled and humidity-controlled vivarium. Food and water were available ad libitum, and lights were maintained on a 12/12 h light/dark cycle. For fentanyl citrate, as in our prior studies, we used 10 µg/ml fentanyl citrate (calculated as free base, Cayman Chemical) in 2% (w/v) saccharin, or 2% saccharin (vehicle control) in the drinking water (Alipio et al., 2021a,b), which does not influence maternal care behavior (Alipio et al., 2021a,b). We here chose this concentration since a concentration-response was conducted in previous work and the different concentrations we used (1, 10, and 100 µg/ml) each impaired circuit function and behavior without influencing maternal care behavior (Alipio et al., 2021a).
Slice preparation
Mice were deeply anesthetized with isoflurane (Halocarbon). A block of brain containing A1 and the medial geniculate nucleus (MGN) was removed, and slices (400 µm thick) were cut on a vibrating microtome in ice-cold artificial CSF (ACSF) containing (in mm) 130 NaCl, 3 KCl, 1.25 KH2PO4, 20 NaHCO3, 10 glucose, 1.3 MgSO4, and 2.5 CaCl2 (pH 7.35–7.4, in 95% O2/5% CO2). For A1 slices, the cutting angle is ∼15° from the horizontal plane to maintain the tonotopy (Zhao et al., 2009; Meng et al., 2017). Left hemisphere slices were incubated for 1 h in ACSF at 30°C and then kept at room temperature. For recording, slices were held in a chamber on a fixed-stage microscope (Olympus BX51) and superfused (2–4 ml/min) with high-Mg ACSF recording solution at room temperature to reduce spontaneous activity in the slice. The recording solution contained (in mm) 124 NaCl, 5 KCl, 1.23 NaH2PO4, 26 NaHCO3, 10 glucose, 4 MgCl2, and 4 CaCl2. The location of the recording site in A1 was identified by landmarks (Cruikshank et al., 2002; Zhao et al., 2009).
In vitro electrophysiology
Whole-cell recordings were performed with a patch-clamp amplifier (Multiclamp 700B, Molecular Devices) using pipettes with input resistance of 4–9 MΩ. Cells targeted for recording were located in an area of A1 overlying the rostral flexure of the hippocampus. Data acquisition was performed by National Instruments AD boards and custom software (Ephus; Suter et al., 2010), which is written in MATLAB (MathWorks) and adapted to our setup. Voltages were corrected for an estimated junction potential of 10 mV. Electrodes were filled with an internal solution containing (in mm): 115 cesium methanesulfonate (CsCH3SO3), 5 NaF, 10 EGTA, 10 HEPES, 15 CsCl, 3.5 MgATP, and 3 QX-314 (pH 7.25, 300 mOsm). Biocytin or neurobiotin (0.5%) was added to the electrode solution as needed. Series resistances were typically 20–25 MΩ.
LSPS
LSPS was performed as described previously (Meng et al., 2017, 2020). Caged glutamate (0.8 mm N-(6-nitro-7-coumarylmethyl)-L-glutamate; Muralidharan et al., 2016) was added to the high-divalent ACSF during recording. Laser stimulation (1 ms) was delivered through a 10× water immersion objective (Olympus). Laser power on the specimen was <25 mW and was held constant between recordings. For each map, an array of up to 30 × 30 sites with 30-µm spacing was stimulated once at 1 Hz in a pseudorandom order. This stimulation paradigm evokes an action potential at the stimulation sites with similar spatial resolution (∼100 µm) over cells in all cortical layers. Putative monosynaptic EPSCs were classified by the poststimulation latency of the evoked current. Evoked currents with latencies of <10 ms are likely to be the result of direct activation of glutamate receptors on the patched cell. Evoked currents with latencies between 10 and 50 ms were classified as monosynaptic evoked EPSCs. The first peak amplitude and the charge (the area of EPSC in the counting window) were quantified for each synaptic response. Recordings were performed at room temperature and in high-Mg2+ solution to reduce the probability of polysynaptic inputs. Cells that did not show any large (>100 pA) direct responses were excluded from the analysis, as these could be astrocytes. Excitatory and inhibitory inputs were recorded at −70 and 0 mV, respectively.
LSPS data analysis
Data were analyzed by custom software written in MATLAB. Cortical layer boundaries were identified by features in the bright-field image as described previously (Meng et al., 2017, 2020). Input area was calculated as the area within each layer that gave rise to PSCs. Integration distance refers to the distance that covered 80% of evoked PSCs along the rostro-caudal direction. Mean charge was the average charge of PSCs from each stimulus spot, and the mean peak amplitude was calculated as the average amplitude of PSCs. The balance of excitation and inhibition was calculated as E/I ratio, which is based on the number of inputs and the average input strength (Edensity/Idensity and Echarge/Icharge).
Cranial window surgery
Two hours before surgery, mice were prophylactically injected with dexamethasone (5 mg/kg) to prevent infection and cortical edema. At the time of surgery, mice were injected again with dexamethasone and atropine (0.1 mg/kg). For surgery, mice were anesthetized with isoflurane (3–4% for induction 1.5–2% for maintenance) and body temperature was maintained at 38°C using a heating pad and closed loop homeothermic monitoring system. Hair was removed by plucking and a chemical depilatory (Nair). The scalp was disinfected with three alternating swabs of betadine and 70% ethanol. The skin above the skull and temporal muscle was removed, and the temporal muscle was resected to expose the temporal bone. The headpost was attached to the skull with a combination of cyanoacrylate (Vetbond) and dental acrylic (C&B Metabond). A 3-mm circular section of bone above A1 was removed and the cranial window was implanted. The cranial window consisted of two 3-mm circular glass coverslips affixed to one 5-mm circular coverslip, the edges of which were filled with a clear silicone elastomer (Kwik-Sil). The window was affixed to the skull with dental acrylic. The dental acrylic and headpost were coated in iron oxide to prevent optical reflections. Mice were postoperatively given injections of meloxicam (0.5 mg/kg) and were allowed to recover for at least a week before experiments.
In vivo two-photon imaging
Imaging was performed as described previously (Liu and Kanold, 2021; Liu et al., 2019; Shilling-Scrivo et al., 2021). After recovery and acclimatization to the microscope, A1 location was functionally determined by its characteristic rostrocaudal tonotopic axis using widefield imaging (Liu et al., 2019). Imaging was performed on a rotatable microscope (Bergamo II series, B248, Thorlabs) using a pulsed femtosecond Ti:Sapphire 2P laser (Spectra-physics Insight X3) at 940-nm excitation wavelength. The imaging field size was ∼370 × 370 µm and was imaged at 30 frames per second.
Sound stimuli
All sound stimuli were presented with a free-field electrostatic speaker 10 cm away from the mouse's right ear (ES1 speaker with ED1 speaker driver, TDT Inc.). The speaker was calibrated by first recording a 70 dB SPL, 4–64 kHz white noise with a calibrated microphone (Bruel and Kjaer) to find the speaker's natural transfer function. The inverse of the function was calculated, which is the function, when added to the speaker's natural transfer function, will equalize the output of the speaker to give a flat frequency/dB curve. Next, this calibration was tested by recording pure tones at 70 dB SPL to ensure that the recorded sound level was <5 dB from the target for all tones played. To perform the frequency response analysis (FRAs), we presented a set of stimuli consisting of 33 tones logarithmically spaced from 4 to 64 kHz. The amplitudes of the tones were calibrated to 70 dB SPL and attenuated from 0 to 30 dB SPL with a step of 15 dB SPL (70, 55, 40 dB). Each tone was 100 ms in duration and had a 10-ms linear ramp at the onset and the offset of the tone. The interstimulus interval (ISI) was 1.5 s.
Stimulus-specific adaptation (SSA) was assessed using two frequencies (f1 and f2) presented in an “oddball” paradigm. We used three different sets of sound sequences, two of which included deviants. The first deviant sequence had f1 as deviant and f2 as standard while the second sequence had the frequencies switched. Each deviant sequence had 400 tones in total; 5% (20) of tones were deviant, and 95% (380) were standard. Since long ISIs will diminish SSA, we chose 1 s as a reasonable ISI in our paradigm. The choice of which sequence was selected first was random. The third sequence had f1 and f2 occurring with equal probability (80 tones in total, 40 f1 and 40 f2). We included 1 min of silence between each sequence. Deviant tones always followed at least 5 standard tone to make sure it is “deviant.” To make sure there is no adaptation to a particular frequency in the equal probability sequences, the same frequency occurred no more than four times in a row. All tones (100 ms, 1-ms ramp) were played at 70 dB SPL. The frequency of f1 and f2 were separated by 0.5 octaves where Δf = 0.37 is the normalized frequency difference [Δf = (f2 − f1)/(f2 × f1)1/2]. Because individual neurons within an imaging field differ in their frequency selectivity, we ensured that for each neuron we used an f1 and f2 that generated similar response and therefore used multiple pairs of stimuli for each imaging field. Thus, tone frequencies were between 4 and 64 kHz and f1/f2 pairs that show 0.5 octave difference were used (4/5.7 kHz, 4.4/6.2 kHz, 5.2/7.3 kHz, 5.7/8.0 kHz, …, 45.3/64 kHz). For each pair of stimuli, we selected neurons which generated similar responses (ΔF/F) at f1 and f2. For example, neurons tuned to low frequencies were picked when we use 5.2 and 7.3 kHz as f1 and f2, and neurons tuned to middle frequencies were picked when we use 12.3 and17.4 kHz as f1 and f2. Therefore, the neurons we used in our experiment are diverse and thus our conclusion apply broadly.
Imaging data analysis
Neuron fluorescence traces were extracted using custom MATLAB code (MathWorks, version 2016B) as described previously (Francis et al., 2018; Liu et al., 2019; Liu and Kanold, 2021; Shilling-Scrivo et al., 2021). Images were motion-corrected to subpixel precision using discrete time Fourier transforms (DFTs; Guizar-Sicairos et al., 2008). Cell bodies were then manually selected with cell bodies and neuropil divisions created automatically. Any pixels that overlapped multiple cells were excluded from analysis. For each neuron, all pixels within the cell body were averaged to create the baseline fluorescence value. To calculate the neuropil fluorescence, all pixels in a ring surrounding the labeled neuron were averaged, excluding pixels that corresponded to other neurons. The corrected neuropil fluorescence was calculated as follows:
ΔF/F was calculated by dividing fluorescence from each trial by the average F of the preceding silent baseline frames as follows:
To test whether neurons were responsive to sound, we determined the significance of the responses by using a similar approach to that used in our prior study (Liu and Kanold, 2021). A 10-frame window before stimulus onset was chosen for measuring baseline activities, while a 35-frame window after stimulus onset was chosen to measure evoked activity, and the 99.9% confidence interval (CI) of the mean of data points before and after stimulus was obtained. A response was deemed significant if the lower bound of the poststimulus CI was higher than the upper bound of the prestimulus CI. To classify the frequency response areas (FRAs), we performed computations as in our previous study (Liu and Kanold, 2021). We first aligned the FRAs of all responsive cells (pooling from all cell types) at the geometric center, calculated through weighted average of frequencies by significant responses, as follows:
SSA analysis
From the fluorescence responses of either f1 or f2 as standard or deviant, we calculated the SSA indices (SIs; Ulanovsky et al., 2004; Taaseh et al., 2011) for f1 and f2 as
Signal correlations and noise correlations were calculated as previously (Winkowski and Kanold, 2013). The signal correlation is defined as how similar the tuning curves are for two neurons. To obtain the signal correlations between two neurons (i and j), we obtained the correlation between the tuning curves by using the Pearson correlation equation as follows:
Neuronal activity has fluctuations which are independent of the sound stimulus. However, neurons which share common inputs or directly connect to each other will preserve correlated trial-to-trial activity. This trial-to-trial correlation is quantified as noise correlation. To calculate noise correlations, we subtracted the average fluorescence response to obtain only the variance between trials, which removes the effect of signal correlation. For each pair of neurons, we then calculated the trial-to-trial correlation with the same correlation equation above.
Statistics
Results are plotted as mean ± SEM unless otherwise indicated. Populations are compared with a rank sum or Student's t test (based on Lilliefors test for normality).
Results
Sound-evoked responses of single L2/3 neurons in fentanyl-exposed and control animals are similar, but population correlations are reduced
Because perinatal fentanyl exposure results in deficits in auditory discrimination behavior (Alipio et al., 2021a), we speculated that the function of auditory cortex was impaired after fentanyl exposure. Thus, we first examined the sound-evoked responses in A1 in adult mice that were perinatally exposed to fentanyl and in control animals (Fig. 1A).
In vivo imaging of A1 in awake adult mice perinatally exposed to fentanyl shows fewer sound responsive neurons. A, Schematic of experimental timeline. B, Schematic of in vivo two-photon imaging in awake mice. Image shows example imaging field. C, Fraction of sound responsive neurons in imaging field; *p < 0.05. D, Example fluorescence traces of four neurons to various frequency sound level combinations. Average trace in black. Sound-responsive stimulus combinations indicated in red.
To investigate sound responses, we performed in vivo two-photon imaging of A1 neurons in L2/3 in awake adult mice that were perinatally fentanyl-exposed (N = 5 mice, 1956 cells) and in control mice (N = 5 mice, 2237 cells; Fig. 1A). We presented sounds (4–64 kHz, eight frequencies/oct, 40–70 dB SPL) and measured fluorescence responses (Fig. 1B). Cells in both control and fentanyl-exposed mice responded to sounds with an increase in fluorescence, but fewer cells were responsive in animals exposed to fentanyl (p = 0.034, Wilcoxon rank sum; Fig. 1C,D). In mouse A1, L2/3 neurons in a local area can show a diversity of frequency preference (Bandyopadhyay et al., 2010; Rothschild et al., 2010; Winkowski and Kanold, 2013) and L2/3 cells can show a diverse set of sound-evoked receptive fields (e.g., V-shaped, I-shaped, etc.) based on the relative amount and spectral composition of excitatory and inhibitory inputs (Sutter et al., 1999; Schreiner et al., 2000; Tao et al., 2017; Liu et al., 2019; Liu and Kanold, 2021). These different receptive fields are thought to be because of both ascending thalamocortical inputs and intracortical signals (Liu and Kanold, 2021). We thus first tested whether the proportion of cells with distinct receptive field types was altered in fentanyl-exposed mice. We performed clustering on the receptive fields (Liu et al., 2019; Liu and Kanold, 2021; Fig. 2A,B) and found that cells in both control and fentanyl-exposed animals show receptive fields of all types and that the proportion of each receptive field type was similar (Fig. 2C). We next characterized properties of the sound responses of single neurons and measured the maximum response amplitude and the bandwidth of evoked responses. We found that the response amplitude for cells across the frequency spectrum was lower in fentanyl-exposed animals (p = 0.0061, Wilcoxon rank sum; Fig. 2D). However, we found that the frequency selectivity (bandwidth) of cells was not affected by fentanyl exposure (p = 0.5083, Wilcoxon rank sum; Fig. 2E). These results suggest that many aspects of sound-evoked responses in single neurons are unchanged in fentanyl exposed animals.
In vivo two-photon imaging shows normal tuning properties but reduced sound-evoked responses of single L2/3 neurons. A, Clustering of neurons based on FRA obtained with in vivo two-photon imaging. Neurons can be grouped into six distinct classes based on FRA shapes. B, FRA shapes of the six classes of neurons. Plotted are the average fluorescence response (ΔF/F) of neurons in each cluster to each stimulus combination (sound frequency and sound level). FRAs from each cell are aligned to center and frequency axis before averaging. Frequency axis shows octave distance from center. FRAs can show classic V shape of frequency selective responses, or W shape for wider tuning. Three groups of sparse S-shaped intensity tuned FRAs were preset with different intensity tuning. H-shaped FRAs had high threshold. C, Distribution of neurons with particular FRA shape in control and exposed animals. D, Amplitude of sound-evoked response; **p < 0.01. E, Bandwidth of FRA.
So far, our analysis has only considered the static responses to sound. Given that fentanyl alters synaptic properties and reduced adaptation to somatosensory stimuli (Alipio et al., 2021b), we next tested whether fentanyl affected dynamic properties of sound responses. In particular, sound-evoked responses adapt to the short-term and long-term sound statistics (Ulanovsky et al., 2004), enabling the binding of behaviorally relevant stimulus components. Therefore, we investigated whether such dynamic properties of sound processing are altered after fentanyl exposure. Neurons in the auditory cortex show SSA in that the response to a repeating sound is reduced while responses to a novel sound are enhanced. Mechanisms such as synaptic depression are thought to at least partially underlie this effect (Ulanovsky et al., 2004; Khouri and Nelken, 2015; Yarden and Nelken, 2017). To measure SSA in imaging fields containing diversely tuned neurons, we chose pairs of sound frequencies, f1 and f2, 0.5 octaves apart, that spanned the entire frequency range of 4–64 kHz. We designed sound sequences with either f1 or f2 as standard or deviant, respectively. Each sequence contained 400 tones; the standard was presented 380 times (95%) while the deviant occurred 20 times (5%). We then measured the neuronal responses to the standard and deviant. For each set of sequences, we only included in the analysis neurons that responded equally when standard and deviant were presented in equal probability (Fig. 3A). To quantify the effect of the varying occurrence probability on the tone response we calculated the SI, which measures the contrast between the responses to the same frequency when it was standard and when it was deviant (Ulanovsky et al., 2004; Taaseh et al., 2011). This index is zero if there is no adaptation. In deviant sequences we found that the amount of adaptation was similar in control and fentanyl-exposed animals [control: p = 0.21, fentanyl: p = 0.63, SI(f1): p = 0.30, SI(f2): p = 0.34, Wilcoxon rank sum; Fig. 3B–D]. We also calculated the CSI between the deviant and standard responses to characterize the average effect of adaptation for this pair of frequencies (Ulanovsky et al., 2003; Taaseh et al., 2011). The CSI was similar for cells in control and fentanyl-exposed animals (p = 0.52, Wilcoxon rank sum; Fig. 3E). Together, these results show that many aspects of static and dynamic sound-evoked responses in single neurons are unchanged in fentanyl-exposed animals.
Normal SSA of single L2/3 neurons in fentanyl-exposed animals. A, Schematic of the three sequences used. In each trial, f1 and f2 are presented pseudo-randomly according to their occurrence probability. B, Responses to f1 or f2 in equal-probability sequences. Cell from control and fentanyl-exposed animals showed similar responses; p > 0.1. C, SIs for f1 and f2. SIs were similar between control and fentanyl-exposed animals; p > 0.1. D, Scatter plots of SI for f1 versus f2 for control and fentanyl-exposed animals. E, Common contrast (CSI) for cells from control and fentanyl-exposed animals; p > 0.1.
While auditory tone detection is abnormal in fentanyl-exposed animals, our assessment of single-cell properties did not reveal large differences. Behaviorally relevant stimuli are encoded not by single neurons but by populations of neurons (Francis et al., 2018). We thus investigated the possibility that population responses to sound are abnormal consequent to fentanyl exposure. To test this possibility, we calculated the pairwise signal correlations between simultaneously imaged neurons (Rothschild et al., 2010; Winkowski and Kanold, 2013; Bowen et al., 2020; Rupasinghe et al., 2021). Prior work has shown that signal as well as noise correlations are highest for neighboring neurons and that both correlations decrease with increasing distance between neurons (Winkowski and Kanold, 2013; Bowen et al., 2020; Rupasinghe et al., 2021). Plotting the signal correlation as a function of pairwise distance (Fig. 4A), we found that, as expected, in control animals signal correlation are highest for nearby neurons and decrease with increasing distance, reflecting a high level of shared input in neighboring neurons. In contrast, in fentanyl-exposed animals signal correlations were lower for nearby neurons and show no distance dependence. We find a similar but more subtle decrease in noise correlations in fentanyl-exposed animals suggesting a decrease in local interconnectivity (Fig. 4B). The decreased response amplitude in fentanyl-exposed animals could potentially reduce the observed correlations. We thus subsampled each imaging field such that the amplitude distributions were similar (Fig. 4C). In the subsampled data we still observe lower signal correlation in fentanyl-exposed animals (Fig. 4D). Together, these results show that while single-cell response properties in L2/3 are essentially normal in fentanyl-exposed animals, pairwise signal and noise correlations are reduced, suggesting a decrease in shared ascending as well as local inputs to L2/3.
Reduced signal correlations of nearby neurons after fentanyl exposure. A, Plots show signal correlation of cell pairs as function of the distance between cells. Solid line is mean, shaded area is SEM. Correlation of neurons within 50 µm are reduced in fentanyl-exposed animals. B, Plots show noise correlation of cell pairs as function of the distance between cells. C, Histogram of response amplitude of subsampled neurons form both control and fentanyl groups with a bin size of 0.2 ΔF/F. D, Signal correlation of cell pairs as function of the distance between cells in subsampled data. Solid line is mean, shaded area is SEM. Correlation of neurons within 50 µm are reduced in fentanyl-exposed animals.
A1 L2/3 cells in fentanyl-exposed animals show fewer mESPCs but not mIPSCs
Because our in vivo data suggested a decrease in shared input to L2/3 neurons, we wanted to investigate the underlying circuit. Perinatal fentanyl exposure leads to reduced excitation in the somatosensory cortex (Alipio et al., 2021b). We thus first investigated whether such changes in intra-cortical circuits also occurred in A1 after perinatal fentanyl exposure. To obtain an unbiased readout of all synaptic inputs to a given cell, we cut thalamocortical slices of A1 and performed whole-cell recordings of L2/3 cells (N = 21 cells, N = 4 mice, fentanyl-exposed; N = 20 cells, N = 4 mice, control) in the presence of TTX to block action potential propagation and recorded spontaneous mPSCs. Spontaneous mEPSCs and mIPSCs were present in cells from both groups (Fig. 5A). We next tested whether the rate of mPSCs was altered in fentanyl-exposed animals and found that the mEPSC rate decreased in exposed animals while the mIPSC rate was unchanged (Fig. 5B). These data suggested that L2/3 cells received fewer excitatory inputs.
Spontaneous EPSC frequency is reduced in exposed mice. A, Example traces showing mEPSCs and mIPSCs from control (black) and fentanyl-exposed (red) mice. B, mEPSC frequency, amplitude and decay time; **p < 0.01. C, E/I ratio based on mPSC rate or amplitude; **p < 0.01.
We next investigated the properties of the individual mPSCs and found that the amplitudes and decay time constants of both mEPSCs and mIPSCs were similar in cells from control and fentanyl-exposed mice (mEPSC rate; amplitude; tau: p = 0.0027; 0.16; 0.79, mIPSC rate; amplitude; tau: p = 0.42; 0.15; 0.20, Wilcoxon rank sum; Fig. 5B), suggesting that individual synapses were unchanged after fentanyl. The balance of excitation and inhibition is critical for proper cortical function (Sohal and Rubenstein, 2019). The reduction in mEPSC but not mIPSC rate suggested that the balance of excitation and inhibition is altered. To directly investigate this hypothesis, we next calculated the E/I balance based on the frequency and amplitude of the mEPSCs and mIPSCs (E/I rate; amplitude: p = 0.0021; 0.56, Wilcoxon rank sum; Fig. 5C) and found that the E/I balance based on mPSC rate was reduced, consistent with a reduction in excitatory inputs. Together, our results suggest that fentanyl exposure results in a loss of excitatory connections to L2/3 neurons and a change in the balance of functional inputs to L2/3 neurons toward inhibition.
Fentanyl exposure does not alter photo- excitability of A1 neurons
We next wanted to identify which intracortical circuits were altered by fentanyl exposure. We performed LSPS in thalamocortical slices of A1 in a separate set of animals as in our prior studies (Meng et al., 2017, 2020). To reliably compare the spatial connection pattern of cells between groups of animals, we first confirmed that the spatial resolution of LSPS was similar across groups by performing cell-attached patch recordings in cells from L2/3, L4, and L5/6 and testing the ability of A1 neurons to fire action potentials in response to the photoreleased glutamate (N = 19 cells, N = 3 mice, fentanyl-exposed; N = 20 cells, N = 3 mice, control). Short UV laser pulses (1 ms) were targeted to multiple stimulus locations to focally release glutamate at the targeted location. The grid of stimulation spots (∼900 spots) covered the entire A1, resulting in a high-resolution 2D photoactivation pattern for a given cell (Fig. 6A). We then counted numbers of evoked action potentials at each stimulus location. Each effective stimulus site generated a single action potential on average, and this was similar between groups (L2/3: p = 0.31, L4: p = 0.47, L5/6: p = 0.83, Wilcoxon rank sum; Fig. 6B). We also measured the distance of the effective stimulation sites from the soma for each stimulus location that generated action potentials. Most action potentials were generated within 150 µm of the soma in both groups of animals (p > 0.05; Fig. 6C). These findings suggest that the sensitivity of cells to photoreleased glutamate and thus the spatial resolution of LSPS remains unchanged after fentanyl exposure.
Photo-excitability of A1 neurons is not affected by fentanyl exposure. A, Schematic of cell-attached mapping experiment. Solid triangles represent neurons. Traces on the right represent cell-attached recordings. Only stimulation close to recorded neuron leads to action potential generation. B, C, Number of action potentials and effective stimulation distance from cell-attached recordings of L2/3, L4, and L5/6 neurons. The numbers of evoked action potentials of neurons in all layers were similar (all p > 0.1), and most spikes were evoked within 150 μm (all p > 0.1).
Fentanyl exposure results in fewer interlaminar excitatory connections to L2/3 neurons
We next investigated whether intracortical circuits to L2/3 neurons change after fentanyl exposure. To visualize the locations of presynaptic cells impinging on L2/3 neurons, we combined LSPS with whole-cell recordings from A1 L2/3 cells. Recorded cells were located at similar laminar positions (p > 0.93; Fig. 7A) and held at –70 mV (EGABA) to isolate excitatory synaptic currents. Laser pulses were targeted in a grid to ∼900 distinct stimulus locations spanning all cortical layers around the recorded cell, and the resulting membrane currents were measured (Fig. 7B). Stimulus locations were sampled in a pseudo-random manner to avoid repeated stimulation of neighboring locations. We observed large, short-latency (<10 ms) inward currents when the stimulation location was close to the recorded neuron because of direct activation of the cell body and the proximal dendrites. In contrast, postsynaptic currents caused by activation of presynaptic neurons were long-latency (>10 ms) events (Meng et al., 2017, 2020).
Reduced interlaminar excitatory connections to L2/3 neurons after fentanyl exposure. A, Relative position of recorded neurons within L2/3. Plotted is the relative position within L2/3 with 10 referring to the border near L4 and 100 referring to the border with L1. Cells were sampled from the middle of L2/3 in fentanyl-exposed and control animals (p > 0.1). B, Schematic of recording EPSCs. Whole-cell voltage clamp recordings at holding potentials of –70 mV. If the presynaptic neuron synapses on the recorded neuron, an EPSC would be observed (yellow trace). Shown on the right are exemplar patch-clamp recordings of direct response (red), EPSC (yellow), and IPSC (green), acquired at holding potentials of −70, −70, and 0 mV, respectively. Dashed blue line indicates time of photostimulation; solid blue line marks 8 ms poststimulus, which is the minimal latency of synaptic responses. C, Schematic illustration of how the connection probability map is calculated. Binarized connection maps from recorded neurons are aligned to the soma location and averaged to create the connection probability map (right). D, Maps of connection probability for excitatory connections in control (top) and fentanyl-exposed (bottoms) mice. Soma location is indicated by the white circle. Connection probability is encoded according to the pseudocolor scale. White horizontal lines indicate averaged laminar borders and are 100 μm long. E, Distributions of the area, integration distance, and mean EPSC charge of excitatory input, originating from L2/3 (top), L4 (middle), and L5/6 (bottom) of control (black) or fentanyl-exposed (red) animals; *p < 0.05.
We mapped 52 L2/3 cells in A1 (N = 25 cells, N = 5 mice, fentanyl-exposed; N = 27 cells, N = 5 mice, control). For each cell, we identified stimulus locations that gave rise to an evoked EPSC and generated a binary input map showing locations that gave inputs or not. For each group of cells, we aligned all binary input maps to the soma position of the individual cells and averaged them, resulting in a spatial connection probability {P(connection)} map for excitatory inputs (Fig. 7C). These maps allowed us to qualitatively identify cortical locations that over the population gave rise to inputs to L2/3 neurons. Qualitatively comparing the connection probability maps, we observed that in fentanyl-exposed mice, fewer cortical locations provide inputs to L2/3, indicating hypoconnectivity (Fig. 7D) consistent with our observation of fewer mEPSCs.
To quantify these qualitative observations, we calculated the laminar inputs of each cell as in previous studies and then compared the populations (Meng et al., 2017, 2020). For each recorded cell, we first identified layer boundaries in images of the slice to map the stimulus locations onto cortical layers. From the binary input maps, we then calculated the total area within each layer from which EPSCs could be evoked. This measure thus reflects the amount of input convergence from each layer to the recorded L2/3 cell. A comparison of input areas between fentanyl-exposed and control mice indicates that the amount of intralaminar excitatory input from L4 and L5/6 was reduced in mice exposed to fentanyl (L2/3: p = 0.63 L4: p = 0.014 L5/6: p = 0.010, Wilcoxon rank sum; Fig. 7E, left). These results indicate that L2/3 neurons in mice exposed to fentanyl receive fewer inputs from thalamorecipient L4 and L5/6.
Because of its two-dimensional nature, our laminar areal measurement takes into account both changes in the input distribution within a layer, e.g., within L5 and L6, as well as changes in the orthogonal direction. Changes induced byfentanyl exposure could occur in either or both directions. Importantly, our thalamocortical slices preserve the macroscale rostro-caudally oriented tonotopic map in the orthogonal direction. Therefore, the spatial extent of the input distribution along the rostro-caudal axis is a proxy for the integration along the tonotopic axis, enabling us to probe whether the reduction of L2/3 inputs occurred along the tonotopic axis. We thus calculated the distance that includes 80% of the evoked EPSCs. We find that this intralaminar integration distance is reduced for inputs originating in both L4 and L5/6 in cells from fentanyl-exposed animals, indicating that L2/3 cells receive L4 and L5/6 inputs from a more restricted area in the tonotopic axis (L2/3: p = 0.27, L4: p = 0.016, L5/6: p = 0.016, Wilcoxon rank sum; Fig. 7E, middle).
Functional circuit changes can occur through alteration of connection probabilities but also through changes of connection strength. Therefore, we tested whether connection strength was altered by perinatal fentanyl exposure. We measured the average size (transferred charge) of the evoked EPSCs and found that connections strength from all layers was unchanged (L2/3: p = 0.69, L4: p = 0.91, L5/6: p = 0.46, Wilcoxon rank sum; Fig. 7E, right). Thus, these findings suggest that A1 L2/3 cells in animals that were exposed to fentanyl receive fewer interlaminar connections from L4 and L5/6 and that these inputs originate from a smaller area in the tonotopic axis. This suggests that the spatial distribution of inputs from L4 and L5/6 are narrower in exposed animals and that L2/3 cells within a local area receive inputs from differing pools of L4 and L5/6 neurons.
Fentanyl exposure does not alter inhibitory connection to L2/3 neurons in A1, leading to an imbalance toward inhibition in fentanyl-exposed animals
We next investigated inhibitory circuits by holding cells at 0 mV and performing LSPS. We recorded evoked IPSCs and derived inhibitory input maps as detailed above. Averaging these maps yielded inhibitory connection probability maps for both groups. In contrast to excitatory maps, the connection pattern of inhibitory circuits showed no qualitative differences (Fig. 8A). Laminar analysis quantitatively supported these observations. L2/3 neurons from treated and sham control mice received a similar amount of inhibitory input from each layer and inhibitory inputs were of similar strength (Fig. 8B, left, L2/3: p = 0.55, L4: p = 0.17, L5/6: p = 0.16; middle, L2/3: p = 0.42, L4: p = 0.66, L5/6: p = 0.20; right, L2/3: p = 0.37, L4: p = 0.96, L5/6: p = 0.26, Wilcoxon rank sum). Thus, our findings show that perinatal fentanyl exposure alters the maturation of excitatory but not inhibitory inputs to L2/3 cells.
Intralaminar inhibitory connections to A1 L2/3 neurons are not affected by fentanyl exposure. A, Average maps of connection probability for inhibitory connections in control (top) and fentanyl-exposed (bottom) animals. Maps are aligned to the soma location. Connection probability is encoded according to the pseudocolor scale. White horizontal lines indicate averaged laminar borders and are 100 μm long. B, Distributions of area, integration distance, mean IPSC charge of inhibitory inputs originating from L2/3 (top), L4(middle), and L5/6 (bottom) of control (black) and fentanyl-exposed (red) animals. C, E/I balance index of inputs originating from L2/3 (top), L4 (middle), and L5/6 (bottom) of cells from control and fentanyl-exposed animals; *p < 0.05.
We next asked whether the E/I balance of inputs from each layer is affected by fentanyl exposure. We calculated the E/I ratio of the inputs from each layer based on the area of PSCs as in our prior studies (Meng et al., 2015). As expected from the excitatory hypoconnectivity, we found that E/I ratios were decreased for inputs from L4 and L5/6 in fentanyl-exposed animals (L2/3: p = 0.79, L4: p = 0.045, L5/6: p = 0.022, Wilcoxon rank sum; Fig. 8C). These results indicate that the relative number of intralaminar excitatory inputs to L4 and L5/6 neurons is decreased.
Discussion
We investigated cortical circuits and function in adult mice perinatally exposed to fentanyl. Our results show a specific hypoconnectivity of interlaminar excitatory connections fromL4 and L5/6 to L2/3 neurons in auditory cortex after fentanyl exposure. In contrast, inhibitory connections were not affected. The excitatory inputs from L4 and L5/6 originated from a more restricted area across the tonotopic axis, indicating a reduced frequency integration. On the functional level our results show decreased activity correlations by nearby neurons in A1. These results indicate that consequent to fentanyl exposure there are distinct effects on specific excitatory circuits and an imbalance toward inhibition.
Animal models of NOWS have shown altered auditory function in that performance in a tone discrimination task was impaired, with exposed mice making fewer overall responses and correct choices (Alipio et al., 2021a). We show here that excitatory ascending processing from L4 as well as feedback circuits from L5/6 to L2/3 in the auditory cortex are impaired and that these changes are reflected in reduced correlated activity of nearby neuronal populations. Thus, the behavioral impairments after fentanyl are likely because of reduced ascending sensory auditory processing.
Studies of the effects of perinatal fentanyl in the somatosensory cortex in rodents have shown a reduction of excitatory and inhibitory synaptic connections to L5 pyramidal cells (Alipio et al., 2021b). We here show by LSPS and mPSC recordings that deficits in excitatory synaptic connections are also present in L2/3 and that these deficits are specific to inputs from L4 and L5/6. Moreover, our results show that while the rate of mEPSCs is reduced, the amplitude of EPSCs is not changed, indicating that there is a reduction in the number of connections and not a change of size in individual synaptic inputs. This is consistent with the reduced morphologic complexity and reduced spine number after prenatal morphine exposure (Mei et al., 2009).
Opioid exposure has been reported to affect inhibitory development. For example, reduced inhibitory synaptic strength in L5 neurons in the somatosensory cortex has been reported after perinatal fentanyl exposure (Alipio et al., 2021b), and prenatal morphine or methadone exposure can alter the pattern of parvalbumin immunoreactivity (Maharajan et al., 2000; Lum et al., 2021). In contrast, we did not detect any changes in inhibitory connections or synaptic strength after perinatal fentanyl either in LSPS or mIPSC recordings. The differences between our study and recordings from somatosensory cortex in the same animal model (Alipio et al., 2021b) might be because of cells being in different layers (L5 vs L2/3), differences between auditory and somatosensory cortex, and changes in PV immunoreactivity not leading to overt functional circuit changes in our assay. Regional differences are conceivable based on differences seen between frontal and somatosensory cortex (Alipio et al., 2021b). Nevertheless, all studies show that early fentanyl exposure can lead to persistent changes in synaptic circuits across the cerebral cortex.
Our results show decreased excitatory input to L2/3 neurons after fentanyl exposure. In normal brains excitation and inhibition are kept in balance by homeostatic plasticity mechanisms (Turrigiano, 1999; Lee and Kirkwood, 2019). Since we did not observe changes in inhibitory inputs, the balance of excitation and inhibition is shifted toward inhibition. This suggests that mechanisms of homeostatic plasticity might not operate normally after fentanyl exposure. This is consistent with the involvement of opioid receptors in synaptic homeostasis, at least in the hippocampus (Queenan et al., 2018), and a role for opioids in plasticity (Beltrán-Campos et al., 2015). Moreover, the altered excitatory inhibitory balance we observe is consistent with reduced network activity observed in vivo (Alipio et al., 2021b).
We here use LSPS to map synaptic circuits. While our whole-cell patch-clamp recordings reveal differences in synaptic currents, our cell-attached data show that the excitability of neurons to photostimulation is unaffected by fentanyl exposure. This suggests that basic neuronal maturation is robust to early fentanyl exposure or that potential changes in neuronal excitability are relatively modest to not affect photoactivation. In contrast to the excitatory hypoconnectivity we observe, prenatal methadone exposure increases cellular excitability and photostimulation-evoked EPSC amplitude in motor cortex (Grecco et al., 2021). While increased excitability could contribute to increased EPSC amplitudes, these data might also point to differences between experimental models or differences between cortical areas.
We here test the long-term persistent effects of prenatal and neonatal fentanyl exposure on cortical circuits. Since such circuits undergo extensive developmental remodeling including overexpansion and pruning (Meng et al., 2020), it is possible that fentanyl exposure interferes with one or both of these processes. Prenatal morphine exposure results in a reduction of spines on L2/3 and L4 neurons in juvenile animals (Mei et al., 2009). It is thus likely that the reduction in excitatory circuits we observe is because of a lack of stabilization of circuits. Nevertheless, our results show that the effects of early fentanyl exposure persist into adulthood.
We find that single-cell responses to sounds were essentially normal in fentanyl-exposed mice. This is somewhat surprising given that we observe reduced L4 to L2/3 circuits. These results might indicate that only a few of the L4 inputs to L2/3 neurons determine the functional tuning properties of L2/3 neurons. Moreover, in the present study, we only used simple tonal stimuli. Thus, it is possible that more complex stimuli might reveal changes in more complex receptive fields (Liu and Kanold, 2021).
We find that pairwise signal and noise correlations are reduced especially for nearby cells. This is consistent with reduced shared ascending connectivity. Human infants prenatally exposed to methadone show changes in event-related potentials (Paul et al., 2014) and EEG on a large scale (van Baar et al., 1989), and also differ in features of the EEG, such as a reduction in the duration of spontaneous activity transients (SATs; Malk et al., 2014). Our results suggest that these changes could be because of a reduction in neuronal population synchrony.
We here chose a fentanyl dose of 10 µg/ml based on prior work (Alipio et al., 2021a). In humans, the range of values likely to be found in pregnant women suspected of opioid use is unknown since physicians find out through voluntary admission from the mother, or observations of neonatal opioid withdrawal signs exhibited by the child. Therefore, doses, type of opioid/s, and levels that get to the child greatly vary in humans and what is modeled/tested in animals. A test that provides good face validity for our model or any model is looking at neonatal opioid withdrawal signs in the pups, which we did in previous studies (Alipio et al., 2021a,b).
In humans, developmental opioid or cocaine exposure is associated with deficits in speech and language processing (Bandstra et al., 2010; Benninger et al., 2020; Lee et al., 2020) while the overt function of the peripheral auditory system is normal (Grimmer et al., 1999). This is consistent with our observation that fentanyl exposure disrupts excitatory processing in the auditory cortex and the observations by others that frontal cortical areas show deficits as well (Alipio et al., 2021b). Since frontal and auditory areas are thought to be involved in complex auditory processing and can influence auditory cortex (Lee et al., 2009; Fritz et al., 2010; Bizley and Cohen, 2013; Francis et al., 2018; Winkowski et al., 2018), together this suggests that perinatal fentanyl exposure disrupts distributed central structures involved in complex sound processing.
In summary, our results show that prenatal and neonatal fentanyl exposure causes a persistent decrease in interlaminar excitatory circuits and population activity in the auditory cortex.
Footnotes
This work was supported by National Institutes of Health Grants RO1DC009607 (to P.O.K.) and R01GM056481 (to J.P.Y.K.) and the Opioid Use Disorders Initiative, MPowering The State, from the State of Maryland.
The authors declare no competing financial interests.
- Correspondence should be addressed to Patrick O. Kanold at pkanold{at}jhu.edu
