Background
Angelman Syndrome (AS) is a rare neurodevelopmental disorder caused by the loss of functional ubiquitin protein ligase E3A [
1]. Specifically, AS results from deficient expression of the maternal allele, which leaves the entire brain deficient of UBE3A due to neuron-specific imprinting that silences the paternal allele [
2‐
6]. AS is characterized by developmental delay, intellectual disability, impaired communication, gross and fine motor deficits, as well as seizures [
7‐
12]. Since these symptoms are severe and persistent, and there is currently no effective therapeutic or cure for the disorder, those with AS require lifelong supportive care. It is therefore imperative that novel strategies to treat AS are developed.
Several in vivo models have been generated to aid in the pursuit of effective treatments, including a conventional germline mouse [
13] with a deletion of
Ube3a in exon 2, a conditional mouse with tamoxifen reactivation [
14], a larger deletion mouse [
15], and rat model with a full
Ube3a gene deletion [
16,
17]. Various models recapitulate phenotypes of AS and therefore provide useful systems in which to test candidate treatments. Lacking a functional level of UBE3A protein in the brain, models show hypo-locomotion, poor balance, impaired coordination, atypical gait, complex cognitive deficits, alongside communication deficits and aberrant social behavior. Since many of these behavioral deficits are not unique to AS, therapies that are effective for other disorders with shared symptomology, such as autism or other syndromic NDDs, may also be effective in treating AS [
18‐
22].
Insulin-like growth factors (IGFs), a family of proteins with similar structure to insulin, have recently emerged as potential treatments for the social deficits, communication impairments, and repetitive behaviors of genetic syndromes associated with autism spectrum disorder (ASD) [
19,
23‐
31]. IGF-1 is being evaluated as a novel treatment for core symptoms of syndromic autisms in one of the first clinical trials of its kind (NCT01970345) [
18‐
22,
29‐
34]. IGF-1 is an FDA approved, commercially available compound that crosses the blood–brain barrier and has beneficial effects on synaptic development by promoting neuronal cell survival, synaptic maturation, and synaptic plasticity. Since IGF-1 has shown efficacy in reversing deficits in mouse and neuronal models of three single gene causes of ASD (namely Rett syndrome [
23,
24,
27], Phelan McDermid syndrome [
28,
35], and Fragile X syndrome [
29]), it may therefore be effective in treating ASD more broadly.
IGF-2, which is important for normal cellular growth and development, tissue repair, and regeneration, has also shown promising effects on ASD-relevant behavioral domains in preclinical studies [
36‐
41]. Injections into the hippocampus have demonstrated that IGF-2 is crucial to the consolidation and enhancement of memories and may be effective in ameliorating memory impairments [
31,
42‐
44]. Since the chemical properties of IGF-2 allow it to exert action within the central nervous system after crossing the blood–brain barrier [
45,
46], systemic delivery of IGF-2 represents a highly translational route of treatment. A study in mice by Stern et al. (2014) found that following systemic administration of IGF-2 via subcutaneous injection, adult male C57BL/6J mice showed enhanced novel object recognition, social recognition, contextual fear and working memory [
43]. Moreover, in the BTBR mouse model of ASD, Steinmetz et al. (2018) found that IGF-2 treatment normalized behavior in the marble burying task, improved social interaction and social memory deficits, and enhanced novel object recognition along with other types of memory [
31].
Despite substantial biological and behavioral differences between the inbred strain BTBR, previously used as an idiopathic ASD model, and the
Ube3a maternal deletion model of AS, the
Ube3amat−/pat+ mouse model of AS was recently reported by Cruz et al. (2020) to exhibit behavioral rescue following acute systemic IGF-2 treatment [
47]. These encouraging results prompted us to i) investigate if the effects of IGF-2 would be rigorous, reproducible, and inter-laboratory reliable, ii) examine both the mouse and rat model of AS to evaluate phenotypes observed across species (i.e., motor impairment), iii) determine whether IGF-2 could ameliorate or reduce the severity of communication deficits unique to the rat model of AS [
16], and iv) extend the standard, albeit non-translational, rescue of performance in the cerebellar dependent rotarod assay to a rescue of nuanced impairments in gait, which are being utilized as outcome measures in both AS models and AS individuals [
48,
49].
Following a dose range investigation using intra-cranial electroencephalography (EEG) recordings, we employed a battery of behavioral assays to evaluate the effect of systemic IGF-2 on social communication and several motor and learning outcomes in the mouse and rat models of AS. A subcutaneous injection was used to deliver IGF-2 to mice and rats 20 min prior to the start of testing. We utilized the standard behavioral protocols of our laboratory and IDDRC behavioral core [
16,
50‐
57] as well as the published protocols of the Alberini laboratory [
47] to compare data directly, fairly, and congruently. A comprehensive battery of tests confirmed that IGF-2 did not affect basic functions including physical characteristics, general behavioral responses, and sensory reflexes, which indicated safety. Disappointingly, however, our data did not provide strong support for reproducibility or inter-laboratory reliability of IGF-2’s improvement on outcomes since we observed a general lack of effect of IGF-2 in several behavioral domains across two AS rodent models.
Discussion
Novel data uncovered by this work illustrated that acute systemic administration of IGF-2 reduced delta spectral power in EEG, a theorized biomarker in AS. This was a very promising initial finding, considering newly published data linking delta power to improvements in the Bayley Cognitive Assessment [
70], prompting the prediction that an effect of IGF-2 on EEG would translate to behavioral improvements. However, disappointingly, the overwhelming majority of metrics for motor abilities, coordination, and learning and memory were unaffected and IGF-2 did not improve social communication, seizure threshold, or cognition. Although our study returned mostly negative results regarding the potential for IGF-2 to improve behavioral deficits in AS, our findings are nevertheless important to disseminate, as they contrast other recent data [
47]. While we were aiming to corroborate the previous reports of IGF-2 efficacy, as inter-laboratory reproducibility is a long-standing goal of ours, we did establish strong reproducibility with other rat studies [
16,
71‐
73], EEG and sleep studies [
72,
74‐
77,
82], and other genetic mutant mouse models of neurodevelopmental disorders [
51,
56,
78]. Furthermore, we did reproduce a number of the
Ube3amat−/pat+ mouse phenotypes observed by Cruz et al. specifically hypolocomotion, fewer marbles buried, and poor rotarod performance [
47].
We observed a moderate effect of IGF-2 on day 1 of rotarod testing in
Ube3amat−/pat+ mice, but this did not extend across the rotarod time course that addresses motor learning and did not extend across species into rats. However, we were able to replicate all of the
Ube3amat−/pat+ mouse and rat model deficits previously reported by our groups [
16,
48,
59,
75] and discover significant reduction of the elevated delta power in
Ube3amat−/pat+ EEG by IGF-2 treatment. One potential explanation as to why we observed effects on EEG power spectral density (PSD) but no changes in behavioral performance is that the increase in delta power may not have substantial behavioral significance. However, we find this explanation unlikely in light of recent evidence from our laboratory illustrating reductions delta power with concomitant behavioral improvements [
74] and a new report in humans with Angelman Syndrome [
70]. A more likely explanation for the present data is that the IGF-2-induced delta power reduction was sub-threshold for behavioral change. This is a burgeoning area of study and while many laboratories hypothesize that PSDs are effective biomarkers [
72,
77,
79‐
82], there is still little data characterizing the strength of the relationship(s) between spectral power and behavioral outcomes.
Given that we were unable to reproduce, nor extend, the broad phenotypic rescue shown in earlier work, it is critical to highlight that our study employed standardized experimental protocols for behavioral testing [
54,
83,
84], which differed from those used by Cruz et al. (2020) [
47]. We had aimed to leverage these protocol differences to show that the effects of IGF-2 treatment were robust enough to carry across laboratories and therefore bode well for translation to the clinic. Inter-laboratory methodological discrepancies included rotarod inter-trial interval duration, open field lighting and duration, marble burying experimental design and analysis, as well as object exploration times and post-training delays. When observing latencies, we did not record scores that exceeded the duration of the test (e.g., Fig.
4, Cruz et al. 2020). Additionally, while our washout period was shorter compared to previous work, we do not suspect that this hindered our ability to detect effects of IGF-2 since we did not find evidence of IGF-2 having an effect greater than one day in duration. Furthermore, if our washout period had been inadequate, the compounding effects of IGF-2 would have been revealed in subsequent testing. However, this was not the case and for each cohort of animals tested, the final assay of the test battery revealed no effect of IGF-2. Arguably, one of the most crucial methodological details that sets our behavioral experiments apart from those conducted previously is our large sample sizes, which were upwards of 25 animals per group. Pooling data from small subgroups (e.g.,
n = 3–4/group as used by Cruz et al.) can artificially inflate error rates (i.e., produce false positives and negatives) due to the high risk of “testing until significance,” particularly when group sizes are not pre-determined [
53,
54,
57,
84]. Pooling subgroups also requires that all groups be subjected to the same exact conditions (e.g., same sequence of prior tests, identical test parameters) and that scores from the various subgroups (particularly wildtype) are confirmed to be similar to each other. Rather than subgroups, it is recommended practice in rodent behavioral testing to use full groups consisting of 10 to 20 animals for a given experiment [
54,
84]. We therefore only used small groups in the collection of initial pilot data and used large cohorts with enough subjects per group to achieve robust statistical power for collection of behavioral data. The novel object recognition findings in the prior report utilized a protocol which i) we used in congenic B6J mice but were unable to reproduce previous results (i.e., there was not recognition as defined by greater time spent with novel vs. familiar object) and ii) does not appear congruent with many of the best recommended practices disseminated by the IDDRC behavioral working group (e.g., maximizing experimenter consistency, ensuring no intrinsic object preference, and using new object pairs when re-testing animals) [
85].
Our study was thorough and unique, as we used two different model species and statistically powerful, large sample sizes, and we investigated the strongest reported phenotypes in the established models. Our dual species approach allowed us to measure social communication in the rat, which exhibits more nuanced social behavior and employs a more sophisticated communication system as compared to the mouse, and to leverage the mouse model for its strong motor phenotypes. Because our rotarod paradigm consisted of three consecutive days, we were able to assess motor learning and not just use it to test motor function. Having both of these metrics available in both species was key as wildtype mice exhibited a ceiling effect that impeded interpretation of a motor learning deficit, but we were able to accurately evaluate this outcome in rats since the performance of wildtype rats changed significantly across test days. By comparing results across species, and across tests within the same behavioral domain, we are able to provide a more thorough and convincing assessment of this IGF-2 treatment paradigm.
While we did see a few promising trends in EEG and rotarod, we also detected effects on gait in the opposite direction than desired (i.e., worsening the phenotype), and the overwhelming majority of our findings indicate that any effect of IGF-2 is minor and does not lead to robust, reliable, or reproducible behavioral changes in either genotype. Moreover, IGF-2 treatment did not lead to consistent phenotypes in the previous report by Cruz et al. (2020). For instance, IGF-2 was not found to affect motor activity in an open field but it did lead to increased marble burying, despite motor abilities playing a key role in marble burying behavior. We did not observe alterations in wildtype mice, which suggests that IGF-2 does not have motor, communication, or cognition enhancing properties in the time windows we assessed. Furthermore, we did not observe alteration in seizure threshold or susceptibility. An obvious difference was Cruz et al.’s utilization of 129 background mice for their audiogenic seizure procedure. AS model mice on the traditional B6J background do not exhibit spontaneous seizures nor susceptibility to audiogenic seizures [
69]. Because the 129 strain has a 70% reduction in corpus callosum volume which adds to their seizure susceptibility [
77,
81,
82], and sensory-dependent audiogenic seizures are triggered by divergent neural circuitry compared to chemo-induction [
74], we utilized the B6J background with a chemo-convulsant.
Therapeutic mimetics of the IGF pathway are being evaluated as small molecule therapy for AS. They activate PI3K-Akt-mTOR and Ras-MAPK-ERK pathways and have been shown to increase synapse number and synaptic plasticity [
86,
87]. Spine numbers have been shown to be reduced in AS mouse models [
88] and activity dependent ERK phosphorylation and synaptic plasticity are impaired [
89‐
92]. The therapeutic hypothesis is that through upregulating synaptic plasticity and synapse number, these compounds may have benefit in AS. We wanted to disseminate our mostly negative data as cautionary for interpreting IGF-2 data, as this ligand shows some non-specificity in binding both the IGF-1 and IGF-2 receptors. IGF-1 has been, and is currently being, pursued as a treatment for neurodevelopmental disorders via four clinical trials: clinical testing in Rett Syndrome (NCT01777542) revealed no significant improvements [
93]; pilot clinical studies of IGF-1 are actively being conducted in non-genetically specified autism (NCT01970345); and two clinical studies of IGF-1 are in process for Phelan McDermid Syndrome, which is a rare genetic neurodevelopmental disorder associated with mutations in
SHANK3 and one of the most common comorbid autism-associated syndromes (NCT01970345; NCT04003207), accounting for up to ~ 1 of all syndromic autism [
94,
95].
Limitations
The major limitation of the present study is that the results are confined to the three doses (10, 30, and 60 µg/kg) and one route of administration (acute subcutaneous injection) used. Particularly, our behavioral results are limited to a 30 µg/kg injection of IGF-2 delivered 20 min prior to behavioral testing. It remains possible that different doses, injection timing and/or frequency, post-administration interval, and/or routes of administration may show greater efficacy in improving the endpoints measured herein. For instance, our negative results using an acute systemic treatment of IGF-2 do not preclude the possibility that chronic delivery of IGF-2 could ameliorate behavioral deficits over longer periods of time. Additionally, our investigation of learning and memory phenotypes was relatively limited so future work would be required to comprehensively determine whether IGF-2 could ameliorate learning and memory deficits.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.