Nowadays, brain stimulation devices have gained significant interest in the scientific community and have received European Medicine Agency (EMA) and US Food and Drug Administration (FDA) approvals for different therapeutic purposes. An approach to achieve brain stimulation is through vagus nerve afferents. Indeed, vagus nerve stimulation (VNS) has become an interesting strategy to help handle drug-resistant epilepsy and depression (Yuan and Silberstein, 2015). In this context, transcutaneous VNS (tVNS), given its non-invasiveness, has received substantial attention. tVNS can be applied to different locations, such as in the neck (Brock et al., 2017) or in the cymba conchae of the external ear (Peuker and Filler, 2002), and, similar to the invasive form of VNS, has already been approved as an adjuvant in the clinical applications, such as for the treatment of drug-resistant epilepsy and depression (Hein et al., 2013; Bauer et al., 2016). Surprisingly, the direct effect of tVNS over brain function is far from understood. Renewed attention to this electrostimulation technique derives from its neuromodulatory effect on cognitive processes. The vagus nerve afferent fibers in the brainstem end in the nucleus of the solitary tract (NTS) (Ruffoli et al., 2011), a relevant relay area for visceral information. From there, afferent information is distributed to many brain regions, including the locus coeruleus (LC) (Grimonprez et al., 2015). The LC provides a widespread innervation to the hippocampus, amygdala, and prefrontal cortex, among other regions (Ruffoli et al., 2011). In this regard, it has been postulated that the LC could regulate cognition through the release of norepinephrine and dopamine in memory processing areas (Ruffoli et al., 2011; Duszkiewicz et al., 2019), mimicking those physiological processes involved in attention-driven cognition (Mello-Carpes and Izquierdo, 2013; Mather and Harley, 2016; Zerbi et al., 2019), but whether such interaction results in the modulation of brain networks is not completely understood. Among those brain networks supporting brain activity, the default mode network (DMN), a brain network predominantly active when the brain is not engaged in an attention-driven task (Raichle et al., 2001; Stafford et al., 2014), has received much consideration due to the reproducibility in its detection in the clinical and preclinical settings (Stafford et al., 2014). The DMN is therefore disengaged during periods of active brain function, as during attention-associated periods, and it has been described to be modulated by tVNS in patients with mild or moderate depressive symptoms (Fang et al., 2016). Focusing on the cognitive functions associated with memory, our group already reported that auricular tVNS (atVNS) enhanced novel object-recognition (NOR) memory persistence in naïve CD-1 mice (Vázquez Oliver et al., 2020). The brain mechanisms recruited by atVNS to enhance memory performance have not been described, so we further explored the cellular outcome of atVNS under conditions that potentiate object-recognition memory persistence. First, we assessed whether a critical time window for atVNS efficacy may limit its effect on favoring object-recognition memory. Second, we analyzed the expression of an immediate early gene c-Fos as an approach to study brain activity in discrete brain regions (Guzowski et al., 2005). In line with the current view proposing that specific cognitive functions are supported by a network of functionally connected brain regions, rather than isolated areas, together with the region-specific analysis of c-Fos, we also evaluated the c-Fos-based functional network (Vetere et al., 2017). We focused our analysis on areas of the brainstem, hippocampus, amygdala, thalamus, and frontal/dorsal cortex, some of which are components of the DMN. Starting from these c-Fos data, we analyzed and estimated the functional connectivity network based on c-Fos density for both stimulated and non-stimulated brains to find significant changes in network connectivity patterns in atVNS condition.

Invasive and non-invasive vagus nerve stimulations (VNSs) have been shown to modulate memory functions, both in animal (Clark et al., 1995, 1998; Vázquez Oliver et al., 2020) and human studies (Clark et al., 1999; Ghacibeh et al., 2006; Jacobs et al., 2015). However, the presence of an effective time window for atVNS effectiveness has not been explored before. In addition, the brain activity outcome of atVNS under memory retention facilitated conditions is still under study.

Therefore, first we aimed to determine whether there was a critical time window for effective enhancement of object-recognition memory persistence through an acute session of atVNS. We found that atVNS in the concha of the left external ear of naïve CD-1 male mice, improved memory persistence, when the electrostimulation protocol was delivered immediately after the familiarization phase of the NORT. Conversely, atVNS delivered 3 h after the training of the NORT did not show any sign of effect on recognition memory at 48 h. This evidence points out the presence of an effective time window for atVNS efficacy in modulating memory persistence. This result is in agreement with a previous clinical report describing that the stimulation, in this case through an invasive approach, was effective when given around the learning phase of a Hopkins Verbal Learning Test, and not during the recall phase (Ghacibeh et al., 2006). Therefore, both invasive and non-invasive forms of VNS, would principally enhance memory consolidation, leading to a better retention power when the electrical stimulus is delivered immediately after the familiarization or learning process.

Second, we investigated atVNS effects on neuronal and network activity, taking into consideration brainstem regions associated with vagal afferences, brain areas important for memory processes, and brain regions implicated in the DMN. It is well established that memory is not stored in a single brain area, but in a network composed of multiple regions (Tanimizu et al., 2017). In the case of recognition memory, the hippocampal formation comprises the main brain region involved (Brown and Aggleton, 2001), although cortical and subcortical areas are also engaged (Frankland and Bontempi, 2005). It has been postulated that memories are initially retained in the hippocampus, and then the information is transferred into the neocortex where it can be consolidated and stored for longer periods (Frankland and Bontempi, 2005; Insel and Takehara-Nischiuchi, 2013). Furthermore, the amygdala complex plays an important role in memory processes, especially under emotionally arousing experiences (Richter-Levin, 2004). These brain regions are contacted by brainstem regions which are relevant for setting the stage concerning the responsiveness associated with the arousal state (Roozendaal and McCaugh, 2011). In this regard, the LC can convey the information from vagal afferents arriving into the NTS (Grimonprez et al., 2015) and affecting the memory-related regions. Hence, we wondered whether the consolidation of new object-recognition memories, facilitated by atVNS, could be mediated by a change in the neuronal activation or a re-distribution of the activity relation between the brain areas.

Neuronal activation and functional connectivity were analyzed by computing the c-Fos density and interregional correlations across animals that received or not the electrostimulation procedure immediately after the familiarization phase of the NORT. We first bilaterally calculated the c-Fos density in a set of cortical, hippocampal, amygdalar, thalamic, and brainstem regions for No stimulation and atVNS conditions. Notably, atVNS did not promote significant changes in c-Fos density in any of the thirty (fifteen per hemisphere) brain regions studied. Other previous reports found changes in c-Fos density after the electrostimulation, particularly an increase in the number of c-Fos+ cells, in areas of the brainstem (Huffman et al., 2019; Jiang et al., 2019; Katagiri et al., 2019). However, all these studies used percutaneous or invasive VNS approaches, or applied the stimulation longer times. Our atVNS protocol instead is completely non-invasive, involves the auricular branch of the vagus nerve and the stimulation period is limited to 30 min, producing a significant effect in memory performance. Thus, the fact that no significant differences could be observed in the number of c-Fos+ cells in brain areas where other studies have found VNS-associated modulations, are probably due to the non-invasiveness and short stimulation protocol of atVNS procedure. Furthermore, several studies have been conducted using functional MRI (fMRI) and atVNS, especially in humans, to shed some light on the possible mechanisms and brain networks involved during atVNS. In general, left atVNS produced a significant activation in the ipsilateral NTS, LC and prefrontal and cingulate cortices, while bilateral deactivation was found in the hippocampus and hypothalamus, and controversial results were described in the amygdala (Kraus et al., 2007; Dietrich et al., 2008; Frangos et al., 2015; Butt et al., 2020). In contrast to fMRI procedure, we used c-Fos as a proxy for cellular activity; the main advantage of this approach is the cellular resolution mapping, although the poor temporal resolution is the principal limitation (McReynolds et al., 2018). Indeed, higher temporal resolution techniques would likely show distinct inter and intra-hemisphere differences that should be addressed in future studies. Additionally, all the animals were actively handled before sampling collections. This process could influence some of the results presented in the analysis, although extreme care was taken to make sure all animals were similarly handled, in order for the groups to only differ in the stimulation procedure. Notably, when we investigated c-Fos functional network, we found a significant reorganization, due to the electrostimulation procedure. atVNS resulted in an enhanced number of significant inter-hemisphere correlations compared to those observed in the No stimulation condition, especially in frontal and amygdalar areas. Moreover, significant correlations were found for sub-networks in the hippocampus (DG, CA1, and CA3) and frontal areas (PrL, IL, and Cg), with a specific increase in the correlation between the left LC and the dentate gyrus after atVNS. The distribution of the network differed between stimulation conditions. Under atVNS prefrontal cortices were segregated and linked to the remaining structures through DMX and left NTS, compared to No stimulation conditions, suggesting the ability of atVNS to dynamically reconfigure large-scale organization. In addition, a reorganization of the amygdala nuclei was observed pointing to a key role of this structure in the atVNS-mediated memory enhancement.

We also explored the effect of atVNS on selected DMN regions. This network has been also described in the mouse brain and involves the pRSP, RSP, Cg, PrL, and IL areas as components of the network (Stafford et al., 2014). By studying these areas in both hemispheres separately, we found that atVNS does not favor the connectivity of the DMN areas, as there is an absence of communication of the DMN anterior–posterior axis between frontal areas and retrosplenial cortex. This is in agreement with the idea that the DMN is mostly disengaged during task performance, and atVNS does not facilitate its engagement, and it may favor its disengagement.

Previous studies, using a combination of c-Fos expression and network analysis, found that long-term contextual fear memories are stored in a brain network composed by thalamic, hippocampal, and cortical regions (Wheeler et al., 2013), while the brain network composed by hippocampus, medial prefrontal cortex, anterior cingulate cortex, and amygdala was found required for the consolidation of social recognition memory (Tanimizu et al., 2017). These studies suggest that distinct types of memory are supported by exclusive functional memory networks that can be revealed by c-Fos analysis. Our findings in this study suggest that enhanced object-recognition memory consolidation is not prompted by an increase in neuronal activation by acute atVNS. Instead, we found a redistribution of the activity, and identified the correlation between brainstem nuclei and hippocampus and frontal areas as the privileged communication that may support the enhancement in memory persistence. Therefore, future studies should also focus on the real-time assessment of changes in neuronal activity induced by atVNS, to better understand its potential in modulating memory processes.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The procedures involving experimentation on animals were reviewed and approved by the Barcelona Biomedical Research Park Experimentation Animal Ethical Committee and the local competent authorities.

CB-P participated in experimental design, conducted and analyzed experiments, and wrote the manuscript. EM-M produced network analysis and revised the manuscript. IG-A conducted and analyzed experiments. RM participated in the supervision and experimental design, funded the project, and revised the manuscript. AI participated in the supervision and stimulator design and generation, funded the project, and revised the manuscript. GS participated in the supervision and analysis of network data and revised the manuscript. AO conceptualized, participated in experimental design, supervised, funded the project, and wrote the manuscript. All authors reviewed and approved the final version of the manuscript.

The project that gave rise to these results received the support of a fellowship from la Caixa Foundation (ID 100010434). The fellowship code is LCF/BQ/IN18/11660012 (CB-P). This project has received funding from the European Union's Horizon 2020 research and innovation program under the Marie Slodowska-Curie grant agreement No. 713673. This work was supported by the Ministerio de Economía, Innovación y Competitividad (MINECO) (#RTI2018-099282-B-I00 to AO and PID2020-120029GB-I00/MICIN/AEI/10.13039/501100011033 to RM); the Instituto de Salud Carlos III (#RD16/0017/0020 to RM and PI18/00893 cofunded by ERDF A way to make Europe to GS); the Generalitat de Catalunya (2017SGR-669 to RM); the ICREA (Institució Catalana de Recerca i Estudis Avançats) Academia to AO, AI, and RM; Grant Unidad de Excelencia María de Maeztu, funded by the MINECO (#MDM-2014-0370); and PLAN E (Plan Español para el Estímulo de la Economía y el Empleo). FEDER funding is also acknowledged.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.