Maternal continuous oral oxycodone self-administration alters pup affective/social communication but not spatial learning or sensory-motor function

Abstract

Background:

The broad use/misuse of prescription opioids during pregnancy has resulted in a surge of infants with Neonatal Opioid Withdrawal Syndrome (NOWS). Short-term irritability and neurological complications are its hallmarks, but the long-term consequences are unknown.

Methods:

A newly-developed preclinical model of oxycodone self-administration enables adult female rats to drink oxycodone (~10/mg/kg/day) before and during pregnancy, and after delivery, and to maintain normal liquid intake, titrate dosing, and avoid withdrawal.

Results:

Oxycodone was detected in the serum of mothers and pups. Growth parameters in dams and pups and litter mass and size were similar to controls. There were no differences in paw retraction latency to a thermal stimulus between Oxycodone and Control pups at postnatal (PN) 2 or PN14. Oxycodone and Control pups had similar motor coordination, cliff avoidance, righting time, pivoting, and olfactory spatial learning from PN3 through PN13. Separation-induced ultrasonic vocalizations at PN8 revealed higher call frequency in Oxycodone pups relative to Control pups (p<0.031; Cohen’s d=1.026). Finally, Oxycodone pups displayed withdrawal behaviors (p’s<0.029; Cohen’s d’s>0.806), and Oxycodone males only vocalized more than Control pups in the first minute of testing (p’s<0.050; Cohen’s d’s>.866). Significant effects were corroborated by estimation plots.

Conclusions:

Our rat model of oral oxycodone self-administration in pregnancy shows exacerbated affect/social communication in pups in a sex-dependent manner but spared cognition and sensory-motor behaviors. This preclinical model reproduces selective aspects of human opioid use during pregnancy, enabling longitudinal analysis of how maternal oxycodone changes emotional behavior in the offspring.

1. Introduction

The growing opioid epidemic has led to increased opioid use during pregnancy and a nearly five-fold increase in newborns exposed to opioids (Kolodny et al., 2015). These newborns experience neonatal opioid withdrawal syndrome [NOWS], marked by irritability, high-pitched crying, gastrointestinal and autonomic dysregulation, and longer hospitalizations (Patrick et al., 2012; Sutter et al., 2014).

Preclinical opioid exposure animal models enable controlled, single-substance approaches to identifying opioid-related neurobehavioral alterations and causative mechanisms, overcoming confounding factors inherent in human studies. These models contribute substantially to our knowledge of NOWS by employing either daily opioid administration to pregnant females or continuous release via minipumps or pellets ensuring continuous maternal and fetal exposure. Both the minipump and pellet models inherently include fetal in utero exposure but also withdrawal from opioids due to spontaneous withdrawal or termination of treatment at parturition (Barr et al., 1998; Hutchings et al., 1992; Kunko et al., 1996; Robinson and Wallace, 2001; Tempel et al., 1995). These fetal exposure models are complemented by postnatal models in which pups are injected to mimic drug exposure during the last trimester of human pregnancy (Jones and Barr, 1995; Robinson et al., 2019).

Recently, we developed an oral oxycodone self-administration protocol in which rats have access to and freely drink oxycodone or water, escalate their oxycodone intake over months, are highly-motivated to continue intake, and become physically dependent, thus emulating the human profile of opioid use disorder [(Zanni et al., 2020); see (Harder and Murphy, 2019) for a recent topical review]. Since initiating opioid use during pregnancy activates a stress-response (Jězová et al., 1982; Martinez-Piñero et al., 1993; Pirnik et al., 2001), we initiated oxycodone intake prior to breeding. The effects of pre-natal oxycodone are well-documented to induce enduring effects in rodent models (Davis et al., 2010; Devarapalli et al., 2016; Sithisarn et al., 2013, 2017; Vassoler et al., 2018), recapitulating the long-term effects observed in humans (Conradt et al., 2019). Nonetheless, oral oxycodone self-administration has not been assessed in pregnant dams (Kokki et al., 2012; Seaton et al., 2007). Furthermore, in the current study, developing pups are exposed to the drug during all phases of fetal and infant development, making it valuable for modeling NOWS.

This approach of opioid exposure before, during, and after pregnancy allowed dams to titrate their intake and minimized confounds associated with withdrawal. Importantly, it also enabled us to ask how oxycodone impacted pup development from postnatal [PN] 2-PN7 (roughly comparable to human premature and term birth) to the point when pups crawl outside the nest and eat solid food [PN14; (Semple et al., 2013; Workman et al., 2013)]. Our results show that following perinatal oxycodone exposure, pups have typical sensory-motor development and spatial learning relative to Control pups; however, drug exposure altered affective states following maternal separation and precipitated drug withdrawal.

2. Methods

2.1. Ethical guidelines

All procedures followed the NIH Guide for the Care and Use of Laboratory Animals (Publications No. 80–23), the Society for Neuroscience and the International Society of Developmental Psychobiology guidelines and were approved in advance by the Children’s Hospital of Philadelphia IACUC. Humane endpoints for euthanizing animals were included and no animal reached these humane endpoints.

2.2. Subjects

Subjects were adult male and female Hooded Long-Evans rats, purchased from Envigo, and their offspring. Three separate cohorts were tested over the course of two years. There were 7 litters in cohort 1, 4 litters in cohort 2, and 6 litters in cohort 3. Rats for cohorts 1–2 arrived at 70–75 days of age; cohort 3 arrived at 50 days old. Because individual pups are not independent, the N’s represent a single litter data point averaged from multiple pups in the litter exposed to the same experimental conditions. This is a long-standing accepted practice for multiparous species (Abbey and Howard, 1973; Festing, 2006; Wainwright, 1998).

Following acclimation to water-only access, females were randomly assigned to the oxycodone only (0.06–0.12 mg/mL; Experimental) or water only (Control) group. Liquid intake was measured before breeding, throughout breeding, and during gestation (Fig. 1A and Supplemental Methods). This concentration of oxycodone elicited stable self-administered doses of ~10 mg/kg/day (Fig. 1B) prior and subsequent to breeding. Drug availability and intake continued throughout data collection.

Fig. 1. Maternal intake, growth, and behaviors.

A. Timeline depicting course of pup pre/postnatal oxycodone or control exposure and experimental procedures. E, embryonic day; G, gestational day; PN, postnatal day; USV, ultrasonic vocalizations.

B. Maternal liquid intake for Oxycodone and Control dams increased over time in both groups (p<0.001) with no difference between groups (p=0.707; larger graph). The oxycodone dose remained relatively constant around 10 mg/kg/day (smaller insert).

C. Dam weight increased similarly during gestation (G1-G21, p<0.001) and lactation (PN0-PN14, p<0.001) with no difference between groups (p=0.575).

D. Mean ± SEM maternal and self-care behavior total duration observed on PN1. No difference was observed between groups (all p’s>0.050).

E. Mean ± SEM rough behavior and milk ejection total frequency observed on PN1. No difference was observed between groups (all p’s>0.050).

F. Mean ± SEM number of pups born per litter to Oxycodone and Control litters. There was no difference in litter size (p=0.999).

G. Mean ± SEM pup weight from PN2-PN14 in Oxycodone and Control litters; weight increased significantly (p<0.001) but no difference in weight gain was observed between groups (p=0.605).

H. Mean blood levels of oxycodone in dams and pups at PN0, PN2, PN9, and PN14.

Breeding began one week after oxycodone/water administration started. Dams were weighed daily, and pups weighed prior to each behavioral test. The day of birth was designated postnatal day (PN) 0 with no cage changes until 2 days after delivery. On PN2, litters were culled to 8–10 pups with equal males and females when possible.

2.3. Maternal behavior

Maternal behavior of the dam was assessed using previously-validated behavioral observation protocols [(Robinson-Drummer et al., 2019); Table 1 and Supplemental Methods]. Blind scoring of video-recorded maternal behavior at PN1 was performed by 2 independent observers offline.

Table 1.

List and description of behaviors recorded and analyzed during maternal behavior observations.

	Behavior	Description
Maternal Care	Nursing on pups	Mom is hunched over actively, or laying on her side passively, with pups attached to nipples
	Grooming pups	Mom cleaning body of pup with her tongue
	Anogenital lick pups	Mom is using tongue to stimulate anogenital region of pup
Nesting Behavior	Pups all in nest	All pups gathered in nest huddle
	Mom in nest	Mom’s entire body is inside nest area
	Nest Building	Mom is actively manipulating bedding material in or around the nest (i.e. shredded paper)
Dam Activity	Mom Active	Any instance of activity where mom is moving inside or outside of the nest
	Walking	Mom is walking around the cage
	Eat/drink	Mom is eating food pellets or drinking from water container
	Rearing	Mom is sitting on back paws with or without her paws resting on the cage wall
	Self-Grooming	Mom is using her tongue to clean her body
Rough handling	Rough transport	Picking up and moving a pup by any area that is not the nap of the neck
	Stepping on pups	Mom tramples pups under foot
	Dragging pups	While pups are still nipple-attached, mom walks away from nest dragging pup underneath
Milk Ejections		While nipple-attached, milk is released causing pups to reflexively elongate their bodies and become momentarily active

Note: Behaviors were compared between moms exposed to oxycodone before, during, and after gestation and control moms exposed only to water.

2.4. Plantar thermal test

PN2 and PN14 pups were tested for the latency to withdraw both hindpaws from a thermal stimulus (Plantar test; IITC Inc. heated base) as previously described [Supplemental Methods; (Barr et al., 2016; Zanni et al., 2020)]. Cut-off time was set at 20 s to avoid tissue injury.

2.5. Neurodevelopmental tests

Pups were tested for the behaviors in the order listed in Table 2 between 1000 h and 1200 h at PN3 and PN7, except pivoting which was performed only at PN7. Each pup test took about 6–8 min and the litter was away from the dam for 45–60 min.

Table 2.

List and description of behaviors recorded and analyzed neurobehavioral assessment.

Behavior	Description
Negative Geotaxis	Pups were placed head down on a mesh surface tilted at a 20° angle. The latency to make a full 180° turn was measured. Trials were capped at 60 s.
Surface Righting	Pups were placed on their backs on a warmed heavy plexiglass sheet (30–32 °C) and the latency to turn over measured. Cutoff latencies were 30 s.
Cliff Avoidance	Pups snout and forepaws were extended over the edge of an elevated surface and latency to move their head and paws away from the edge within 30 s recorded.
Pivoting	Pups were placed on the warmed heavy Plexiglas sheet and the number of 90° turns executed within 60 s was recorded

2.6. Maternal separation ultrasonic vocalizations

Separation-induced ultrasonic vocalizations (USVs) were measured individually in temperature-controlled chambers (32 °C) at PN8 [(Barr and Wang, 1992; Carden et al., 1991); Supplemental Methods]. Using standard measures provided by Raven Pro (version 64 1.4), vocalizations were assessed for total call count, peak frequency (kHz), and peak power (dB). After five minutes of separation, pups were returned to the mom for two minutes then tested for a second five-minute separation period.

2.7. Precipitated withdrawal

At PN9 pups were injected intraperitoneally with naltrexone HCl (Sigma; 1 mg/kg; 0.1 mg/mL) or saline and immediately placed in the temperature-controlled chambers (32 °C). USV’s were recorded for five minutes. Video recordings of pup behavior were analyzed blindly for specific withdrawal behaviors (Table 3). Following testing, pups were sacrificed by rapid decapitation and trunk blood collected and promptly put on ice.

Table 3.

List and description of pup withdrawal behaviors.

Behavior	Description
Face Washing	Licking and/or swiping the front paws over the face and head
Walking	Coordinated front and hind paw movements with linear or circular locomotion
Moving Head and Paws	Simultaneous head and front paw movement without locomotion
Head Moves	Lateral or rotational head movement without paws or locomotion
Quiet	No locomotion or movement except as-needed for respiration

Note: Behaviors were adapted from (Ceger and Kuhn, 2000; Jones and Barr, 1995; Perez-Saad et al., 1996; Windh et al., 1995) and were recorded and analyzed following i.p. naltrexone injections on PN9. Behaviors were compared between Oxycodone and Control pups given either naltrexone or saline injections immediately prior to behavioral recording.

2.8. Spatial navigation

PN13 pups were tested for hippocampus-dependent spatial learning [Fig. 3B; (Nway et al., 2017; Wiedenmayer et al., 2000)] using odor cues. There were four training trials for pups to navigate to a goal box containing littermates (cutoff=150 s). After the fourth trial, the goal box and littermates were removed and replaced with a clean empty goal box and the latency to enter the former location of the littermate box recorded (cutoff=180 s). Navigation cues were four cotton applicators, one in each corner fixed 8 cm above the floor and infused with 100 u L of anise oil, coconut oil, orange extract, or peppermint extract (McCormick). There were two configurations: cues remained constant between trials (“Fixed”) to allow for use of the cues to identify the location of the goal box; or the four odors were randomly exchanged among the corners after each trial (“Alternating”) to eliminate the use of these odors as navigation cues (Supplemental Methods). Within a litter, half of the pups received the Fixed training and half received the Alternating training.

Fig. 3. Thermal nociception tests and spatial learning.

A. Mean ± SEM latency (seconds, s) to withdraw the hindpaw from a noxious thermal stimulus on PN2 (left graph) or PN14 (right graph). There were no differences in withdraw latencies observed between Oxycodone or Control litters at either timepoint (p’s>0.050).

B. Schematic depicting spatial learning paradigm. The “cloud-like” images show placement of the four odors.

C. Mean ± SEM latency (seconds, s) to return to the littermate goal box during training (trials 1–4) and at test when the goal box was removed. There was no difference in performance between Oxycodone and Control litters (p=0.767). For both groups, there was a main effect of trial (p=0.001) with pups in the fixed configuration performed significantly better at test than the alternating configuration (p=0.018).

C-1. Estimation Plot of the significant main effect of the training cues (Fixed vs. Alternating) on the test trial. For this and subsequent estimation plots, the left side of the vertical line show data from an each litter with the horizontal dashed lines shows group means. The right side of the vertical shows the difference between the means and its 95% CI. Note the lack of overlap of the 95% CI of the Fixed condition with the mean of the Alternating condition, providing confidence as to the strength of the effect. Cohen’s d effect size is presented in the upper left corner and is defined as “large” by the “rule of thumb” of Cohen as elaborated on by Sawilowsky (Cohen, 1988; Sawilowsky, 2009).

2.9. Dam and pup oxycodone blood levels

Blood was collected taken via rapid decapitation from dams (14 days after delivery) and pups (PN0, PN2, PN9, and PN14) in different litters to assay oxycodone levels (Supplemental Methods). We did not draw blood from the dam at earlier ages to avoid disrupting maternal behavior that might affect ongoing behavioral tests. Blood was sent to the Bio-analytical Core Center for Clinical Pharmacology at CHOP’s Research Institute for analysis, as described previously (Zanni et al., 2020).

2.10. Statistics

Table 4 presents the number of animals and litters for each dependent measure in all experiments.

Table 4.

The detailed listing of the number of animals and litters for each dependent measure in all experiments.

Experiment	Oxycodone			Saline			Mean/Median
	#M	#F	#Total N	#M	#F	#Total N	#Pups/Litter
Maternal Liquid intake	NA	NA	8	NA	NA	8	NA
Dam Weight	NA	NA	5	NA	NA	5	NA
Pup Weight (All ages)	9	9	9	8	8	8	7.3, 7
Maternal Behavior	NA	NA	7	NA	NA	4	NA
Dam time in nest	NA	NA	9	NA	NA	7	NA
Oxycodone Blood (Pup)	NA	NA	4–9*	NA	NA	NA	NA
Oxycodone Blood (Dam)	NA	NA	8	NA	NA	NA	NA
Thermal test (PN2)	7	7	7	7	7	8**	8.4, 8
Thermal test (PN14)	7	8	8	7	7	8**	5.5, 5
Neurodevelopment (PN3)	9	9	9	8	8	8	8.6, 8
Neurodevelopment (PN7)	8	8	8	8	8	8***	8.4, 8
Separation USV	5	5	10	5	5	10	8.2, 8#
Experiment	Oxycodone			Saline
	#Saline	#Naltx	#Total N	#Saline	#Naltx	#Total N	#Pups/Litter
	#M/#F	#M/#F
Precipitated Withdrawal (USV)	5/5	5/5	10	5/5	5/5	10	2##
	#Saline	#Naltx	#Total N	#Saline	#Naltx	#Total N	#Pups/Litter
Precipitated Withdrawal (behavior)	10	10	20	5	6	11	4###
Experiment	Oxycodone			Saline			Mean/Median
	#Alter	#Fixed	#Total N	#Alter	#Fixed	#Total N	#Pups/Litter
Spatial Navigation	4	4	8	6	5	11	5.22, 4

Notes: The entries under the heading #M, #F, #Saline, #Naltx, #Alter, and #Fixed, are the number of litters. #Total N is the number of data points for each dependent measure. #Pups/Litter represents the mean and median number of pups in a litter averaged to provide the litter score. Thus, for Neurodevelopment tests at PN3, males in 9 litters and females in the same 9 litters were tested with an average of 8.6 individual animals making up the litter mean (shown in the last column, #Pups/Litter). Thus, there was an N = 9 composed of multiple pups for each “N”.

NA means Not Applicable.

^*Depending on age at weighing/blood draw.

^**In some litters only one sex was tested, thus we have more litters than pups.

^***7–9 depending on the measure.

^#For Separation induced USV’s, data from multiple pups of the same sex in a litter were averaged and males and female averages were analyzed as a within litter effect.

^##For precipitated withdrawal induced USV’s, we also tested two males and two females from each litter. Here there were sex differences, and the number of pups in each litter for which data was combined was two of each sex.

^###For precipitated withdrawal behaviors, two males and two females were tested from each litter, one of each sex receiving naltrexone and one saline. There were no sex differences and the two same sex pups were combined to form litter means as described.

Analyses were planned a priori and performed using Prism GraphPad 8.0/9.0 software (Graphpad, San Diego, CA). N’s are the number of litters as data from littermates (~2–8 pups in the litter exposed to identical treatment) were averaged to produce a “litter mean”, ensuring independent measurements, reduced variability and increased power (Abbey and Howard, 1973; Festing, 2006; Wainwright, 1998). Independent sample Welch’s t-tests, which do not assume equal variance in the two groups and which result in corrected fractional degrees of freedom, were used to assess differences in litter size and maternal behavior. Mixed factor ANOVA’s were used to assess differences in all other variables with Sex is included as a variable in all statistics except spatial navigation, due to insufficient power from low N. Posthoc tests were conducted using Šidák’s method. Tests for outliers were conducted using the ROUT outlier test (Q=0.10 %) and identified outliers for the maternal intake data and the blood analyses. To assess the strength of the effects, estimation plots and Cohen’s d effect sizes were calculated for all significant main effects and interactions (Cohen, 1988; Ho et al., 2019).

3. Results

3.1. Dam liquid intake

Oxycodone did not alter dam liquid intake during the perinatal period; the intake dose was approximately 10 mg/kg throughout (Fig. 1B). We identified 4 outliers in the Control group out of approximately 475 measurements (ROUT test, Q<0.10%). When analyzed without the outliers using a mixed factor ANOVA, there was a significant increase (p<0.001) in liquid intake over days, driven by increased drinking during lactation relative to the pre-breeding period. Neither Drug treatment (p=0.707) nor the Day-X-Drug Treatment (p=0.253) were significant. We also analyzed gestational and postpartum intake separately, with and without outliers. In no case was there a significant Drug Treatment or Day-X-Drug treatment interaction (all p’s>0.310). Thus, Experimental dams self-titrated oxycodone intake throughout pregnancy and nursing, drinking the same total liquid amount as Control dams. These data are consistent with prior work when non-pregnant female rats and male rats were given a choice of oxycodone in their water versus water only (Zanni et al., 2020)

3.2. Dam weight

Experimental dams showed similar weight gain as Control dams (Fig. 1C). There was a significant increase (p<0.001) in dam weight in both groups between the first (G1) and last (G21) day of gestation that was maintained throughout lactation (PN0-PN14; p<0.001). There was no main effect of Drug Treatment (p=0.575) but there was a significant Drug Treatment-X-Day interaction (p=0.003). Posthoc tests did not find significant differences between Experimental and Controls on any day with a trend at G0 (p=0.089). Thus, maternal continuous oral oxycodone self-administration did not alter overall weight gain throughout pregnancy.

3.3. Maternal behavior

Within the limits of the small sample sizes, maternal behavior was unaffected by oxycodone intake (Figs. 1D and 1E). Independent samples Welch’s t-tests revealed no significant differences between Control and Oxycodone-exposed mothers in the duration or frequency of any observed behavior (p’s>0.085; Table 5). Hence, within the constraints of our protocol, we found no major maternal oral oxycodone effects on maternal behavior.

Table 5.

Results of Independent samples Welch’s t-test comparing maternal care, nesting behavior, dam activity, and rough handling in Oxycodone and Control dams.

	t ratio	df	p value
Maternal Care
Nursing on pups	0.838	3.765	0.452
Grooming pups	0.378	7.955	0.715
Anogenital lick pups	0.501	4.786	0.639
Nesting Behavior
Pups all in nest	1.000	3.000	0.391
Mom in nest (total time)	2.011	6.833	0.085
Mom in nest (mean time)	0.439	10.230	0.670
Nest Building	0.374	3.707	0.729
Dam Activity
Walking	0.500	3.448	0.647
Eat/drink	0.270	4.531	0.799
Rearing	0.467	3.245	0.670
Self-grooming	0.510	8.986	0.623
Rough Handling
Rough transport	0.622	3.767	0.569
Stepping on pups	0.914	7.283	0.390
Dragging pups	0.172	8.154	0.867
Milk Ejections	0.932	0.933	0.408

Note: Maternal oxycodone consumption during and after gestation did not change maternal behavior relative to water-only control dams.

3.4. Pup and litter measures

3.4.1. Litter size

Maternal oxycodone consumption did not alter litter size relative to Controls prior to culling (Fig. 1F; p>0.999).

3.4.2. Pup weight

There was no main effect of Sex (p=0.198) nor significant Sex interactions (p’s>0.291). Pups gained weight over time (p<0.001) between PN2 and PN14 with no significant effect of Drug Treatment (p=0.605) or Age-X-Drug Treatment interaction (p=0.978). Therefore, prenatal maternal continuous oral oxycodone self-administration did not affect the offspring growth rate (Fig. 1G).

3.5. Blood levels

Female and male pup data on oxycodone blood levels were combined. A ROUT outlier test (Q=0.10 %) identified two aberrant blood assay results: one dam (130 ng/mL) and one PN14 pup (57 ng/mL); these data are not shown. Oxycodone was detectable in blood at all ages except at PN14, when only two pups had detectable levels (Fig. 1H). These data suggest that oxycodone is passed from the lactating mom to the pups, albeit pup levels are lower than maternal levels at ages of significant milk intake.

3.6. Neurodevelopmental tests

There were no gross drug exposure-induced impairments in sensory-motor function due to maternal drug exposure throughout early development (all p’s>0.050; Fig. 2, Table 6).

Fig. 2. Neurodevelopmental tests performed on PN3 (top) or PN7 (bottom).

There was no difference observed between groups in any test (all p’s>0.050).

A/D. Geotaxis: Mean ± SEM seconds to reorient head and body on a 20° slope in 60 s.

B/E. Righting: Mean ± SEM seconds to rotate from supine to prone position within 30 s.

C/F. Cliff Avoidance: Mean ± SEM seconds to move back-wards from an elevated edge in 30 s.

G. Pivoting (PN7 only): Mean ± SEM total frequency of 90° turns in 60 s.

Table 6.

Mixed-Factor ANOVA analyses of neurodevelopmental and thermal response test between Oxycodone and Control pups.

		PN3			PN7
Behavior	Treatment Effect	DFn, DFd	F ratio	p value	DFn, DFd	F ratio	p value
Geotaxis	Drug-X-Sex	1, 15	0.04	0.853	1, 15	0.04	0.853
	Drug	1, 15	0.14	0.716	1, 15	0.14	0.716
	Sex	1, 15	0.01	0.920	1, 15	0.01	0.920
Righting	Drug-X-Sex	1, 15	0.71	0.414	1, 14	0.24	0.630
	Drug	1, 15	0.19	0.673	1, 14	0.48	0.498
	Sex	1, 15	0.27	0.608	1, 14	0.01	0.923
Cliff Avoidance	Drug-X-Sex	1, 15	0.05	0.823	1, 14	2.39	0.144
	Drug	1, 15	2.71	0.121	1, 14	0.00	0.971
	Sex	1, 15	0.15	0.702	1, 14	0.04	0.847
Pivoting	Drug-X-Sex	–	–	–	1, 14	0.01	0.916
	Drug	–	–	–	1, 12	3.95	0.070
	Sex				1, 12	0.04	0.850
		PN2			PN14
Behavior	Treatment Effect	DFn, DFd	F ratio	p value	DFn, DFd	F ratio	p value
Thermal Test	Drug-X-Sex	1, 12	0.41	0.533	1, 25	0.04	0.841
	Drug	1, 12	1.58	0.233	1, 25	2.08	0.162
	Sex	1, 12	0.35	0.563	1, 25	0.00	1.000

Note: Oxycodone exposure did not alter pup development or response to thermal stimuli. N’s are litter means, each litter consisting of 2–6 pups/sex.

3.7. Nociceptive thermal tests

Experimental and Control pups were similar in response to the thermal stimulus at PN2 or at PN14 (all p’s>0.160; Fig. 3A, Table 6). This suggests that oxycodone blood levels in pups were either not sufficient to produce analgesia or that tolerance developed.

3.8. Spatial navigation

When separated by Sex, N’s for spatial navigation were small (Experimental n=2–3, Control n=4–5) so analyses were collapsed across sex. Both Control and Experimental litters showed hippocampus-dependent spatial learning in the Fixed, but not Alternating, configuration (Fig. 3C). There was no effect of Drug Treatment (p=0.767) but there was an effect of Trial (p<0.001) and Configuration (Alternating vs. Fixed, p=0.018). Performance improved over training trials in the Fixed vs. Alternating configuration for both Control and Experimental litters (p’s=0.036). There were no interactions of Trial-X-Drug treatment (p=0.939), Trial-X-Configuration (p=0.154), Drug Treatment-X-Configuration (p=0.950) or three-way interaction (p=0.829). Thus, both Control and Experimental pups learned, and oxycodone exposure had no effect on this spatial navigation task at this age.

3.9. Maternal separation-induced USVs

There was no difference between pre- vs. post-separation USVs (all p’s>0.050), so all results were collapsed across separation (Fig. 4A).

Fig. 4. Separation-induced and Withdrawal-induced USVs. Maternal separation-induced USVs on PN8.

A. Mean ± SEM peak USV frequency (kHz) during maternal separation collapsed over pre- and post maternal reunion. Experimental pups emitted higher-frequency USVs overall relative to Control pups (p=0.002).

B. Mean ± SEM peak USV power (dB) emitted during separation periods. There were Minute-X-Sex (p=0.029) and Minute-X-Drug Treatment-X-Sex (p=0.031) interactions. Separate analyses of females and males found Minute-X-Drug Treatment interactions (p=0.023 and p=0.036, respectively), but neither reached significance at individual minutes (all p’s>0.050).

C. Mean ± SEM number of USVs emitted during separation periods. Number of calls decreased over time (p<0.001) for both groups with no difference observed between groups (p=0.495).

A-1. In the original ANOVA there was a significant main effect of perinatal treatment (Control vs Experimental) on the peak frequency of USV following separation from the mother, with Oxycodone pups crying at a higher frequencies. This main effect is plotted as an estimation graph. Note the lack of overlap between the 95% CI of the Experimental group with the mean of the Control group. The effect size determined by Cohen’s d was “large”.

Naltrexone-induced USVs on PN9.

D. In males, for both groups, cries decreased between the first and fifth minute (p<0.001). However, naltrexone-injected oxycodone male pups emitted more cries relative to all other groups during the first minute (*p’s<0.050,). In the second minute naltrexone increased crying in the Oxycodone pups compared to saline (^{^}p<0.050).

E. For females, cries decreased between the first and fifth minute (p<0.001) with no other effects observed (all p’s>0.124).

D-1, D-2, D-3. In the original ANOVA there was a significant increase in the number of USV’s precipitated by naltrexone for males in the first two minutes of the test (Šidák’s posthoc analysis) for the Control/Saline group compared to the Oxycodone/Naltrexone group (D-1, D-2) and for Oyxcodone pups given Saline or Naltrexone (D-3). For the first minute there was no overlap of the 95% CI with mean of the other group (D-1) However, for the second minute of the test, there was overlap, suggesting less confidence that this difference was reliable (D-2). In D-3, pups from the same litter were injected either with Saline or Naltrexone. Since they were littermates they were related and analyzed by a repeated measures analysis. Each line represents a single litter with Oxycodone pups given Saline or Naltrexone. For each litter there were more calls when pups were given Naltrexone than Saline. There was no overlap between the 95% CI of the two means. Cohen’s d values are in the upper left of each plot.

3.9.1. Peak frequency (Fig. 4A)

There was no effect of sex on USV peak frequency (all p’s>0.103) and the data were collapsed across sex and males and females considered separate data points. Peak frequency increased over the course of the five minute separation for all groups (p<0.002), but was higher in Oxycodone pups relative to Control pups (p=0.031), with no interaction between Drug Treatment-X-Minute (p=0.438).

3.9.2. Peak power (Fig. 4B)

There were significant Minute-X-Sex (p=0.029) and Minute-X-Drug Treatment-X-Sex (p=0.031) interactions. Post hoc analysis revealed no significant pairwise differences (all p’s>0.050). Subsequent analysis of males and females separately revealed significant Minute-X-Drug Treatment interactions (p=0.036 and p=0.023, respectively), but posthoc analyses did not show significant pairwise Minute differences between Experimental and Control groups for either sex (all p’s>0.376).

3.9.3. Number of USV’s (Fig. 4C)

There was no effect of Sex on number of calls emitted (p=0.342) so results were collapsed across sex. The number of calls emitted in both Oxycodone and Control animals decreased across time (p<.001) with no difference in the number of calls between groups (p=0.496) and no Time-X-Drug Treatment interaction (p=0.438).

These data show that overall oxycodone altered the acoustic properties of separation-induced vocalizations but did not alter USV number.

3.10. Precipitated withdrawal USVs

3.10.1. Number of vocalizations

Following naltrexone injection, USVs increased in male but not female pups in the Experimental group, although peak frequency and peak power were unchanged (Fig. 4). In males, the three-way ANOVA for the number of cries revealed a significant main effect of Minute (p<0.001), and interactions of Minute-X-Perinatal treatment (p=0.026) and Minute-X-Drug treatment (p=0.012). Although cries decreased generally between the first and fifth minute (p < 0.001), naltrexone-injected Oxycodone pups emitted more cries relative to all other groups during the first minute (p’s<0.050) and relative to Control-Naltrexone pups during the second minute (p=0.020), with no differences observed between other groups at any time (p’s>0.250). There were no other main effects or interactions (p’s>0.050). In females, there was a significant main effect of Minute (p<0.001) such that cries decreased between the first and fifth minute (p<0.001); however there were no other main effects or interactions observed (p’s>0.124).

3.10.2. Peak frequency and power

Peak power and peak frequency analyses showed no main or interaction effects of naltrexone injection in males or females regardless of perinatal drug exposure (all p’s>0.050 for power and all p’s>0.115 for frequency).

3.11. Precipitated withdrawal behaviors

Overall, Experimental pups increased their activity relative to Controls regardless of antagonist treatment (naltrexone or saline, Fig. 5, Table 7). There were no effects of sex (all p’s>0.050), so all analyses were collapsed across sex.

Fig. 5. Naltrexone-precipitated withdrawal behaviors.

A,F. Oxycodone increased concurrent head and paw movement total duration (p=0.008) and frequency (p=0.024; Figs. 5A, 5F).

B,G. Face washing total duration (p=0.029) was decreased in Oxycodone pups relative to Control pups (Fig. 5B; however, frequency did not differ significantly between groups (p=0.082; Fig. 5G).

C,H. Total duration of lying quietly was increased (p=0.008) in Control pups relative to Oxycodone pups, Frequency of quiet did not change (p=0.108; Fig. 5H).

D,I. Head movement did not differ between groups in total duration (p=0.806) or frequency (p=0.454).

E,H. Walking in Oxycodone pups showed non-significant trends toward increased total duration (p=0.076) and frequency (p=0.063).

Table 7.

Mixed factor ANOVA analyses of pup behavior in Oxycodone and Control pups following i.p. injections of naltrexone on PN9.

		Total Duration (s)			Total Frequency
		DFn, DFd	F ratio	p value	DFn, DFd	F ratio	p value
Moving Head and Paws	Perinatal Drug Exposure	1, 27	8.15	0.008	1, 27	5.67	0.024
	Naltrexone/Saline	1, 27	0.53	0.471	1, 27	1.43	0.242
	Perinatal Drug Exposure-X-Naltrexone/Saline	1, 27	0.46	0.503	1, 27	0.00	0.960
Face Washing	Perinatal Drug Exposure	1, 27	5.29	0.029	1, 27	3.25	0.082
	Naltrexone/Saline	1, 27	0.03	0.857	1, 27	1.47	0.237
	Perinatal Drug Exposure-X-Naltrexone/Saline	1, 27	1.09	0.307	1, 27	0.00	0.982
Quiet	Perinatal Drug Exposure	1, 27	8.10	0.008	1, 14	2.94	0.108
	Naltrexone/Saline	1, 27	0.02	0.878	1, 13	0.34	0.569
	Perinatal Drug Exposure-X-Naltrexone/Saline	1, 27	0.56	0.342	1, 13	0.28	0.607
Head Moves	Perinatal Drug Exposure	1, 27	0.29	0.596	1, 27	0.00	0.978
	Naltrexone/Saline	1, 27	0.06	0.806	1, 27	0.58	0.454
	Perinatal Drug Exposure-X-Naltrexone/Saline	1, 27	1.79	0.192	1, 27	0.68	0.415
Walking	Perinatal Drug Exposure	1, 27	3.42	0.076	1, 14	4.10	0.063
	Naltrexone/Saline	1, 27	0.00	0.979	1, 13	0.70	0.418
	Perinatal Drug Exposure-X-Naltrexone/Saline	1, 27	1.00	0.325	1, 13	0.21	0.652

Note: Analyses are presented for behavior total duration (seconds, s) and total frequency. Oxycodone exposure increased active behaviors and decreased quiet regardless of naltrexone injection. Significant effects are represented with bolded and italicized text. pw drug = precipitated withdrawal drug.

3.11.1. Head and Paw Movements

Oxycodone exposure increased the total duration (p<0.008) and frequency (p=0.024) of concurrent head and paw movements regardless of naltrexone versus saline treatment (Figs. 5A, 5F).

3.11.2. Face washing

Face washing duration was decreased in Oxycodone pups relative to Controls (p=0.029); its frequency did not differ (p=0.082; Figs. 5B, 5G).

3.11.3. Quiet

Oxycodone pups spent significantly less time quiet relative to Control pups (p=0.008) regardless of antagonist treatment, although the number of quiet bouts did not differ (p=0.108; Figs. 5C, 5H).

3.11.4. Head movements

There were no effects or interactions of perinatal treatment or drug on head movement duration (p’s>0.192) or frequency (p’s>0.415; Figs. 5D, 5I).

3.11.5. Walking

There was no significant differences in total duration (p=0.076) or frequency (p=0.063) of pup walking (Figs. 5E, 5J).

These data suggest that this model of maternal continuous oral oxycodone self-administration induces measurable levels of withdrawal behaviors in the offspring that are not further precipitated by opioid antagonist administration.

4. Discussion

Our data show pre- and postnatal exposure to the semi-synthetic opioid oxycodone disrupted pup affective behavior yet left neuro-development capabilities and spatial learning largely intact. This was not due to stunted growth, which was unaffected by perinatal oxycodone exposure. Within the constraints of the small sample size and limited sampling to one hour during the light cycle, there were also no major effects on maternal behavior.

4.1. Oxycodone did not alter pup growth and had minimal effects on dam weight gain

Pup weight, litter size, and dam weight between Control and Experimental litters did not differ throughout gestation and nursing. Thus, if there were differences in maternal care, they did not affect the growth of the pups. The current data largely corroborate previous oxycodone studies: prenatal oxycodone exposure produces little-to-no effects on fecundity, gestation, or neonatal growth, although some variability exists across exposure protocols (Davis et al., 2010; Devarapalli et al., 2016; Sithisarn et al., 2013, 2017). Similarly, prior work has shown that maternal intra-atrial/venous oxycodone did not change litter size, sex distribution, or pup weight, but did reduce maternal weight gain, which may be attributed to intermittent drug access and the confounding effects of withdrawal between drug exposures (Sithisarn et al., 2013, 2017; Vassoler et al., 2018). Oxycodone delivered by oral gavage and initiated prior to breeding reduced birth weight in Oxycodone pups that resolved by the first postnatal week (Davis et al., 2010). However, if gavage was initiated during gestation, weight reductions persisted into adolescence (Devarapalli et al., 2016). Likely when drug treatments are initiated prior to gestation and maternal withdrawal and stress are minimized, as done here, somatic effects on pups are minimized (Jězová et al., 1982; Lichtblau and Sparber, 1981; Pirnik et al., 2001).

4.2. Oxycodone was transferred from the dam to the pups

This study did not address pharmacokinetic mechanisms because of limited sampling throughout the time during which dams and pups were exposed to oxycodone. There was detectable oxycodone in the pups’ blood, but the levels were variable and low for dams and pups. The intake of the dams was about 10 mg/kg spreadout over the day and the variability of oxycodone in the blood of both dams, and by extension, the pups, is likely due to uncontrolled factors – for example, time and amount of the dam’s last oxycodone intake and time and amount of the pup’s last milk ingestion. Moreover, our data likely underestimate peak levels since blood was collected during the day when milk ingestion is low. Although there are few data on the relationship between oxycodone blood/brain levels and behavior, a dose of oxycodone (2.25 mg/kg s.c.) that was analgesic in adult male rats produced oxycodone blood levels of 340 ng/mL (Pravetoni et al., 2012), suggesting that the doses here may have been insufficient for behavioral effects. However, in an similar study with oral oxycodone intake, female rats drinking similar amounts of oxycodone, albeit for longer periods of time, did show precipitated withdrawal (Zanni et al., 2020). Human maternal oxycodone levels are also often variable. Colostrum concentrations, for example, range widely (from 0 to 168 μg/L), and correlated with maternal serum levels rather than dosage. Moreover, oxycodone milk levels in mothers range from below detection levels to 225 μg/L [Drugs and Lactation Database (LactMed): National Library of Medicine (US)]. Better insight into pharmacokinetic and drug disposition in both dams and their offspring, and the relationship of those parameters to behavioral effects, is clearly warranted but exceeded the goal of this study, which was to determine if there was transfer of drug from the dam to her pups.

4.3. Maternal continuous oral self-administration oxycodone and maternal behavior

Opioids acutely regulate the onset, modulation, and maintenance of maternal behavior (Cruz Ade et al., 2010; Grimm and Bridges, 1983; Slamberova et al., 2001; Yim et al., 2006). The effects of chronic exposure are less clear. Oxycodone IV self-administration prior to and during gestation increased pup retrieval latency (Vassoler et al., 2018). However, in that study, no other measures of maternal behavior were analyzed, raising the question as to what led to the increased pup retrieval time (e.g. dysfunctional motor control, reduced use of olfactory cues). Of the handful of reports that measured how maternal behavior is influenced by opioids (Chen et al., 2015; Davis et al., 2010; Robinson and Wallace, 2001; Sithisarn et al., 2013, 2017), most used morphine. Gestational morphine injections decreased pup-directed behaviors such as pup grooming, yet increased non-maternal activities such as dam self-care (Slamberova et al., 2001; Yim et al., 2006); however, maternal morphine effects on pup retrieval are not consistent (Cruz Ade et al., 2010; Yim et al., 2006). These behavioral changes are further confounded by increased maternal stress due to gestational or post-gestational drug initiation (Grimm and Bridges, 1983). Also, behavioral changes due to cycles of opioid withdrawal in studies of intermittent drug exposure muddle putative direct opioid effects. Indeed, dam self-grooming and licking is reduced by morphine exposure during gestation, but not during lactation (Cruz Ade et al., 2010; Yim et al., 2006). Moving forward, it will be important to control for withdrawal effects in such studies.

Clinical presentation and preclinical models of NOWS show impaired somatosensory development, altered affective behavior, and physiological abnormalities (Barr et al., 1998; Kunko et al., 1996; O’Callaghan and Holtzman, 1976; Sithisarn et al., 2017; Timar et al., 2010; Wallin et al., 2019). Rodent models using intermittent access expose pups to drug peaks and in utero withdrawal, which in some cases may be quite relevant clinically (Chen et al., 2015; Davis et al., 2010; Devarapalli et al., 2016; O’Callaghan and Holtzman, 1976; Sithisarn et al., 2013; Slamberova et al., 2001; Timar et al., 2010; Vassoler et al., 2018; Wallin et al., 2019), but may also cause poor fetal outcomes independent of specific drug effects (Lichtblau and Sparber, 1981). Chronic exposure models (minipumps, pellets) prevent in utero withdrawal, but if initiated during gestation cause maternal stress (Jězová et al., 1982; Martinez-Piñero et al., 1993; Pirnik et al., 2001). Moreover, pups experience withdrawal at birth if the maternal opioid is discontinued or pups are cross-fostered to drug-naïve surrogates (Hutchings et al., 1992; Kunko et al., 1996; Robinson and Wallace, 2001; Tempel et al., 1995). Here, using a continuous (24 h a day, 7 days a week) oral self-administration paradigm, dams had access to oxycodone prior to conception and through lactation and could titrate their intake, reducing the likelihood of spontaneous withdrawal (Zanni et al., 2020), thus dissociating the consequences or drug exposure from withdrawal for the offspring. These parameters more closely mimic human fetal exposure both during illicit maternal opioid use and clinically-regulated opioid assistance programs for avoiding adverse effects of newborn withdrawal (Kolodny et al., 2015; Patrick et al., 2012; Sutter et al., 2014).

4.4. Maternal continuous oral self-administration of oxycodone did not impair infant spatial learning

At PN13, both Control and Oxycodone pups demonstrated spatial learning. The task used here is sensitive to perinatal insults, at least in mice; perinatal exposure to environmental pollutants impaired olfactory-spatial learning in PN11 mice (Nway et al., 2017). Others report that prenatal opioid exposure impairs spatial learning from adolescence to adulthood (Davis et al., 2010). During spatial navigation tasks in that study, adult mice prenatally-exposed to oxycodone were impaired when long (40–50 minute) inter-trial intervals were used. This oxycodone-related deficit was not observed using short intervals between trials (15–30 minute), perhaps suggesting only a long-term memory retention deficit, implying hippocampal dysfunction. This lack of an oxycodone-related deficit with short intervals is consistent with our findings.

4.5. Maternal continuous oral self-administration oxycodone disrupted pup affective behavior

Pup isolation-induced USVs are considered a measure of distress, a proxy for maternal attachment, dampened by maternal presence, and critically-dependent on the developing opioid system (Carden et al., 1991; Carden and Hofer, 1990; Hofer et al., 2002). Prior results show that oxycodone exposure via maternal intravenous self-administration did not change total number of calls (Sithisarn et al., 2017), similar to our findings. Our novel findings are that perinatal oxycodone increased peak USV frequency and blunted the decline in power of USV’s during maternal separation relative to Control pups at PN8. Previous reports that show isolation- and stress-induced shifts in USV frequency are indicative of a negative affective state (Boulanger-Bertolus et al., 2017; Ise and Ohta, 2009). Indeed, non-opioid drugs (e.g. ethanol) also produce shifts in isolation-induced USVs toward higher frequencies (Shahrier and Wada, 2018). Together, these results suggest increased negative affect in pups following perinatal oxycodone exposure.

4.6. Withdrawal behavior is differentially-expressed in oxycodone-exposed pups

On PN9, both saline- and naltrexone-injected Oxycodone-exposed pups increased withdrawal behaviors relative to Controls; however, only naltrexone-Oxycodone male pups increased USVs. Pup withdrawal behaviors are distinct from those in adults, change from infancy through adulthood in the rodent (Ceger and Kuhn, 2000; Jones and Barr, 1995; Perez-Saad et al., 1996; Windh et al., 1995) and include increased numbers of USVs (Barr and Wang, 1992). In our work presented here, chronic oxycodone exposure increased infant withdrawal behaviors relative to Controls, but naltrexone- and saline-treated pups perinatally exposed to oxycodone did not differ. One possible explanation is that Oxycodone pups could not be further precipitated with naltrexone because they were already experiencing spontaneous withdrawal since they were tested in the day portion of the day-night light cycle when milk intake is presumably lower.

In contrast, USV numbers, an affective measure of withdrawal sensitive to naltrexone treatment, were elevated in males for the first two minutes after precipitated withdrawal. This is consistent with prior data where withdrawal was precipitated in pups injected twice daily with morphine (Barr and Wang, 1992) or born of mothers implanted with methadone pellets (Barr et al., 1998) and is a likely indication of increased distress. Interestingly, females did not increase USVs perhaps due to sex-dependent effects in oxycodone pharmacokinetics (Chan et al., 2008). Our previous study examining adult oxycodone oral self-administration showed that blood levels greatly differed between sexes (Zanni et al., 2020) with males showing lower blood levels when similar drug concentrations were ingested. It remains to be studied whether possible sex differences in blood levels account for these disparities in the USV response during withdrawal in the absence of sex-hormone differences in pups, or whether they are due to other mechanisms.

4.7. Limitations of the model

No preclinical model can fully reproduce the entirety of the human experience, including our attempt to model the clinical realities of maternal oxycodone use by allowing oral intake (Harder and Murphy, 2019). Thus, there are limitations to this work. First, dams were not given an alternative to ingesting oxycodone which was the sole source of liquid intake. In prior work, we found that in a continuous two-bottle choice paradigm (oxycodone or water), female non-pregnant rats drank almost exclusively from the oxycodone bottle, suggesting that the lack of a water option here might not have altered intake of oxycodone, and total liquid intake did not differ between Experimental and Control dams. A second constraint is that we did not explicitly explore the role of withdrawal in our outcomes (Sparber and Lichtblau, 1983), even though females strongly preferred drinking during the dark cycle. Although dependent on the unknown pharmacokinetics and disposition of oxycodone in lactating dams and pups, pups likely experienced some level of withdrawal during this experiment. The role of withdrawal in any outcome, however, remains to be directly assessed. Third, our goal was to determine if pups had detectable levels of oxycodone in blood derived from the mother’s milk, which they did, but not to conduct a more rigorous examination of blood concentrations. We did not want to do non-terminal blood draws from the female because of concern of stressing her and her pups. Also, we did not control tightly the time of day for the blood sampling; nor did we monitor the time from her last drinking bout or from the last milk ingestion for the pups. These were beyond the scope of this study. Controlling those variables with a range of oral intake concentrations is an important next step. Fourth, our rats were isolated and it would be important to know if different social environments would alter intake and outcomes. Fifth, although we did not find differences in maternal care, our efforts were limited in the time of day and scope and duration of our recordings, and the limited number of dams. It cannot be excluded that a more granular analysis with larger sample sizes would have uncovered subtle differences in maternal care in oxycodone-drinking mothers. Finally the kinetics of oxycodone differ in humans and rats, with therapeutic levels for pain in humans being higher than those found in dams here (Choi et al., 2017). Thus, our outcomes might have been different with higher doses of oxycodone.

4.8. Conclusion

Overall, our preclinical study recapitulated key aspects of chronic oral oxycodone self-administration in pregnant women and consequent neonatal withdrawal symptoms. Exposure to oxycodone as conducted here caused measurable detrimental effects on affective domains in developing pups. These effects are strikingly comparable to those observed clinically (i.e. high-pitched crying and irritability), suggesting that this route and mode of administration may be complementary to the existing opioid exposure models and can be utilized to investigate the immediate and long-term effects of oral self-administration on affect and cognition. Oral self-administration of oxycodone had fewer developmental effects than did studies giving other opioids through different routes and time windows of exposure, suggesting this route of oxycodone delivery targets developing affective systems more strongly than somatic, cognitive, and motor systems. The importance of our findings are two-fold: 1) We replicated selective aspects of NOWS and showed that maternal oral self-administration alters pup affective behavior, similar to what is observed in humans; and 2) Maternal oral self-administration of oxycodone may provide a suitable approach to reliably model the clinical scenario and further study the long-term trajectory of these altered affective components. For example, as withdrawal can occur in human oxycodone users, it could be assessed in in our model by withholding oxycodone at different points in time either pre- or postnatally. Various treatments, pharmacological or environmental, could be employed to assess if they reduce intake or alter outcomes. Thus, this model should allow for further exploration of maternal oxycodone use during pregnancy and the postnatal period.