Main

Vagus nerve stimulation (VNS) has wide-ranging clinical applications, including Food and Drug Administration-approved treatments for epilepsy and depression in adults. Over 75 000 patients have been implanted with VNS devices (1). VNS is currently being investigated as a clinical treatment for sepsis and rheumatoid arthritis, because it suppresses peripheral inflammation and may be important in modulating neuroinflammation (2, 3). The effects of VNS have been correlated with serum tumor necrosis factor (TNF) concentrations as a way to assess the efficacy of VNS to treat endotoxemia (4). The anti-inflammatory effect also seems to depend upon reduction of inflammation via descending cholinergic efferent output (4, 5). Although VNS has therapeutic value in adult patients, its impact on pregnancy, birth, and on fetal development and well-being remain largely unknown. Owing to the inevitability of pregnancy in women undergoing VNS treatment, the impact of this therapy on pregnancy and fetal well-being is an important consideration—particularly with the increasing number of vagus nerve stimulators implanted over the past decade with expanding Food and Drug Administration approval (1). Case reports of VNS therapy in pregnant women found no adverse effects on pregnancy or the postpartum neonate (6, 7), though these studies did not include physiological end points or extend to the use of animal models.

The vagus nerve interconnects the medulla, the heart, lungs, stomach, and other viscera, including the colon and female reproductive tract (8), thus any change in vagal activity would be expected to have an impact on these systems. Approximately 80% of vagus fibers carry afferent information to the central nervous system while 20% of the vagus is efferent information to the periphery (9). Though the uterus is virtually denervated during pregnancy, cervix remodeling involves parasympathetic innervation from the vagus and pelvic nerves to the cervix (10), in part via brain nitric oxide synthase and calcitonin gene-related peptide fibers (11). These sensory neuropeptides mediate inflammatory responses in other tissues (11). We have previously shown that vagus nerve transection reduces the presence of macrophages in the cervix but also resulted in distension of the bladder and stomach in our previous work—suggesting that descending vagal input is key to normal cervical function (12) as well as autonomic tone to the viscera. As vagus nerve transection in pregnant rats has such a significant impact, investigating the effect of VNS on the progression of pregnancy and inflammatory processes in the fetus is warranted and understanding the role of VNS on pregnancy and pup outcome are the major motivators for this work.

VNS may be an effective treatment during pregnancy for infections that result from a compromised cervix immune barrier, as may occur with chorioamnionitis or premature rupture of fetal membranes (13, 14, 15, 16), which exposes the developing fetus to pro-inflammatory cytokines and is associated with neonatal morbidity and brain injury. These cytokines are part of the developmental signals that regulate prenatal neural connectivity (17) in specific brain regions, and there is limited evidence to indicate a reduced size of brainstem nuclei following systemic treatment with endotoxin, a prenatal inflammatory agent (18). Inflammation in these regions causes impaired breathing responses to hypoxia in neonatal rat pups (19, 20). Thus in this study, we tested the hypothesis that maternal VNS would interfere with parturition and alter the level of pro-inflammatory cytokines in cardio-respiratory regions of the brainstem. We tested this hypothesis by using immunohistochemistry to label for pro-inflammatory cytokines and the transcription factor high mobility group box 1 (HMGB1). Our objective was to determine whether chronic VNS impacts morbidity and mortality in infants born to mothers with implanted vagal stimulators.

Methods

Adult Long Evans rats (Harlan, Indianapolis, IN) were individually housed in a temperature- and humidity-controlled vivarium with a 12-h light/dark cycle with food and water ad libitum. Animal care and usage followed the National Institutes of Health guidelines and all experimental procedures were approved by the LLU Institutional Animal Care and Use Committee (IACUC).

Dam Surgery and VNS

Two days after acclimatization to the vivarium, we performed surgery on pregnant dams at day 16 postbreeding, under 2–4% isoflurane–oxygen gas anesthesia (Figure 1, overview of study design and timeline). A midline incision was made below the rib cage to expose the subdiaphragmatic vagus nerve, which was then separated from the connective sheath and vasculature to obtain an accessible length of approximately 1 cm. The vagus nerve was isolated at the subdiaphgramatic portion of the esophagus following previously described procedures (12). After gentle, blunt dissection, a glass hook and fine forceps were used to separate the posterior vagal trunk and isolate a length of nerve sufficient to allow electrodes to encircle the nerve. VNS stimulators were fabricated by Harald Stauss Scientific and activated via a Hall Effect switch. We used the RNS model stimulator (http://haraldstauss.com/HaraldStaussScientific/default.html) to maximize current density over the course of the stimulation period. The electrode-wrapped vagus nerve was covered with a thin film of a silicone-based sealant (Kwik-sil, WPI, Sarasota, FL) to isolate and insulate the stimulation area. The electrode wires were connected to a battery-powered, implantable stimulator unit that was inserted into the abdominal cavity. Following surgery, the abdomen wall was sutured and the skin secured with clips. Pregnant dams were randomly assigned to groups in which the implanted VNS device was set to deliver 5 or 1,000 Hz at 1 mA amplitude with 500 μs pulse width or no current (Sham). Our stimulation parameters were based on parameters commonly used in clinical practice and preclinical studies assessing the effect of vagal nerve stimulation on selective fiber blockade (9, 21).

Figure 1
figure 1

Vagus nerve stimulation protocol. Vagus nerve stimulation protocol and timeline for experiments. In each treatment group: 1,000 Hz, 5 Hz, and Sham, we implanted stimulators with helical electrode contacts encircling the subdiaphragmatic, esophageal branch of the vagus nerve at embryonic day 16 (E16) in pregnant dams. Pregnancies proceeded to birth and the pups were killed and the brainstems removed within 12 h after birth.

Full size image

Stimulation duration was 30 min (on-time) at the given frequency for that animal and was followed by an off-time of 5.5 h repeated 4 × per day. The programmed frequencies were confirmed each day using a digital AM/FM radio (Sony Corp of America, New York, NY, USA). Weight and food consumption were monitored daily to assess well-being of the dams and verify the stimulation regimen. Thus the pups assessed for this project were exposed to either 5 or 1,000 Hz over 5–6 days post-VNS stimulator implantation as dams were allowed to continue in the study until parturition (typically gestational day 21–22 in rats).

Processing of Brainstems, Immunohistochemistry, and Analyses

Within 12 hours of delivery, dams were asphyxiated with CO2. Pups were deeply anesthetized with isofluorane and their brainstems were removed, rinsed with chilled saline, and stored in 4% paraformaldehyde for 24 h before transfer to 30% sucrose for 48 h to provide cryoprotection. Brainstems were frozen in Tissue-Tek OCT and sectioned at 20 μm on a Leica CM 3050S (Buffalo Grove, IL, USA) cryostat. Sections of brainstem were stained using immunohistochemistry with selective antibodies for interleukin (IL)-6 (1:100 sc-1265-r, Santa Cruz, Dallas, TX, USA), TNFα (1:100 ab6671, Abcam, Cambridge, MA, USA), IL-1β (1:100 ab9722, Abcam), and HMGB-1 (1:1,000 ab18256, Abcam) as previously described (22). The number of cells expressing IL-6, TNFα, IL-1β, and HMGB1 protein were counted using a Zeiss Axio Imager A1 (Buffalo Grove, NY, USA) and unbiased stereology (Stereologer2000, Stereology Resource Center). All cell counts were normalized to volume assessed to account for variability in tissue morphology across sections. Photographs were acquired within Stereologer and postprocessing was carried out using GIMP (http://gimp.org) and Adobe Illustrator CS6 (San Jose, CA, USA). We analyzed all data using one-way analysis of variance with the Duncan post hoc test (SPSS, v22, IBM, Armonk, NY, USA) with a P<0.05 considered significant.

Results

VNS Effects on Pregnancy, Parturition, and Pup Viability

Neither surgery nor VNS interfered with either the progression of pregnancy or the parturition process. Pregnancy proceeded to term and litters had delivered in Sham (n=4) and VNS (5 Hz n=10, 1,000 Hz n=11) groups by day 21–22 postbreeding (see Figure 1 for timing of VNS surgery and exposure). Pups were viable—breathing, with vigorous movement and vocalization when held. The pups were pink and had milk present in their stomachs when harvest of tissue began, consistent with the metrics used to assess neonatal well-being using an Apgar scale (23). This is consistent with data seen from vagus nerve transection pups from dams with transected vagus nerve pups from previous work (12). Table 1 summarizes the gestational age at delivery, number of live pups per litter, number of pups resorbed in utero, and postpartum pup mortality for each treatment group (5 Hz, 1,000 Hz, and Sham).

Table 1 VNS effects on pregnancy outcome and demographic features
Full size table

Brainstem Cell Counts

We used unbiased stereology to count the number of specific stained cells in pups from each group in the nucleus tractus solitarii from −12.0 to −14.7 mm (relative to Bregma), which encompasses the critical cardio-respiratory control regions in the brainstem. In all treatment groups, we counted >500 cells across the nucleus tractus solitarii. Compared with Sham controls, the number of IL-1β (Figure 2, panels a1–c1), IL-6 (Figure 2, panels a2–c2), TNFα (Figure 2, panels a3–c3), and HMGB1 (Figure 2, panels a4–c4) labeled were not significantly different in pups from dams that received the 5 or 1,000 Hz VNS. Table 2 shows the summary counts (mean±SD) and P-value for each marker. Figure 3 shows summary histograms with mean cell counts for Il-1β (panel a), IL-6 (panel b), TNFα (panel c), and HMGB1 (panel d) (±SD) for all treatments and stains.

Figure 2
figure 2

Immunohistochemistry staining for cytokines and high mobility group box 1 (HMGB1). From top to bottom, we show interleukin (IL)-1β (a1–c1), IL-6 (a2–c2), tumor necrosis factor-α (TNFα) (a3–c3), and HMGB1 (a4–c4) immuno-staining for each treatment (Sham, 5 Hz, 1,000 Hz). Each panel shows an inset photomicrograph (4 ×) with the distance from Bregma and higher resolution photomicrograph at 40 ×. Arrows indicate representative stained cells. Scale bar=1 mm (4 ×) and 100 μm (40 ×).

Full size image
Table 2 Cell counts per volume across stimulation groups
Full size table
Figure 3
figure 3

Summary histograms of unbiased stereology results for immunolabeling. All results are reported as the number of stained cells vs. stimulation parameter. Markers shown are interleukin (IL)-1β (a), IL-6 (b), tumor necrosis factor-α (TNFα) (c), and high mobility group box 1 (HMGB1) (d). Cell count numbers were divided by volume of area of interest to normalize for tissue size and then averaged (reported as average±SD). N=3 for all groups.

Full size image

In addition to the canonical, early-onset pro-inflammatory cytokines, we stained for HMGB1, due to its putative anti-inflammatory role. We saw a non-significant increase in HMGB1 staining in the nucleus tractus solitarii in the 1,000 Hz stimulation group when compared with 5 Hz and Sham treatments, which is opposite of the trend seen with IL-6 and TNFα.

Discussion

In this study, we found that maternal VNS does not significantly increase the expression of pro-inflammatory cytokines in the cardio-respiratory control regions of the medulla, does not interfere with the parturition process, does not cause preterm birth, nor negatively impacts viability of the pups born to VNS-treated dams. Because we saw no ill-effects of maternal VNS, our data suggest that there is no impairment of the physiological processes essential for normal birth and viability of neonates. These findings support and extend observations in case reports describing the effectiveness of VNS therapy during pregnancies in which VNS has been used to treat depression and refractory epilepsy. The authors’ of these reports found no impact on the timing of delivery of the mothers’ infants (6, 24). An additional study of eight pregnancies in which the mothers received VNS for refractory epilepsy showed no adverse effects on either pregnancy or neonatal viability (7). Collectively, these findings support the suggestion that VNS may be included in therapeutic approaches to diseases during pregnancy.

The importance of these findings is underscored by the choice of physiologically relevant VNS characteristics. We used low frequency and current stimulation parameters that mimicked those seen in clinical use (25). Additionally, we used a higher stimulus frequency (1,000 Hz) that blocks a significant proportion of the afferent traffic carried via the vagus (21, 26). Neither VNS stimulus regimes resulted in significant changes in IL-1β, IL-6, TNFα, or HMGB1 across our treatment groups. We focused on the early-onset cytokines and HMGB1 as they represent the “first responders” of inflammatory regulation. HMGB1 is a chromatin-associated protein that has a key role in transcription and regulation of gene expression. It has also been implicated in the anti-inflammatory response observed in other studies of VNS (27, 28). We also constrained our focus to the brainstem regions critical for cardio-respiratory control and implicated in neonatal morbidity and mortality (29, 30). In the brain, IL-6 is typically produced in response to the early upregulation of proinflammatory cytokines such as IL-1β via the nuclear factor-κB pathway (31, 32). In addition to IL-6, TNFα, and IL-1β are produced and released as part of the early inflammatory response (33). IL-1β is a key pro-inflammatory cytokine that can positively feedback to the nuclear factor-κB (31) signaling cascade and exacerbate the production of pro-inflammatory cytokines to amplify the inflammatory response (34). TNFα triggers local inflammatory responses and serves as a co-factor for modulation of presynaptic release in the central nervous system (35) and the succeeding cascade of both pro- and anti-inflammatory cytokines and chemokines. The trophic role of TNFα in modulation of brainstem autonomic circuits is not yet known but it has been shown to be necessary for appropriate synaptic formation (35). Work has previously been carried out in murine models looking at obesity-induced inflammatory changes in the expression of IL-6 in the hypothalamus (17), and this is an area of research that warrants further research.

VNS has been implicated in the modulation of inflammatory tone and been suggested as a therapy to alleviate the impact of chronic inflammation. VNS upregulates the expression of TNFα (5) systemically in non-pregnant adult murine models, and this has been shown to activate other cytokines, such as IL-1β, and HMGB1 (4) which is implicated in both pro- and anti-inflammatory responses to VNS as part of hypothalamic–pituitary–adrenal axis activation (36, 37). IL-1β is the proto-typical early-onset pro-inflammatory cytokine and lipopolysccharide injection induces local expression of IL-1β in respiratory control centers of the brainstem (19, 20). Because these cytokines are upregulated very quickly and early in the inflammatory response, their expression in response to inflammation may provide an early indication of the efficacy of VNS in either promoting or blocking their expression and subsequent changes in inflammatory tone.

The work we report here suggests that VNS lacks significant negative impact on the production and release of cytokines and does not impair the timing of parturition or the viability of the newborn. An obvious limitation of our study is that we constrained our stimulus parameters to two sets (relatively low and high) and we did not employ animal models of inflammation or seizure to determine the efficacy of VNS as an intervention. Further studies will need to be performed to determine whether there are stimulus parameters for VNS that are inherently pro- and anti-inflammatory.

Limitations of the Study

Our work is innovative in assessing the role of VNS in a pregnant rat model but there are inherent limitations to our study. Because we used implantable stimulators with limited battery, we were limited in the current density available to block using kilohertz frequency alternating current. Other investigators have used anesthetized animal models to perform kilohertz frequency stimulation and show complete neural block of peripheral nerves at higher frequencies than we used in our experiments (21, 26, 38, 39, 40). We used a maximum of 1,000 Hz for our stimuli after consulting with the vendor of our implantable stimulators to assure continuous stimulation over the entire experimental time period. Additionally, we did not assess anti-inflammatory markers; however, HMGB1 is implicated in both anti- and pro-inflammatory roles (37, 41). Future work will incorporate a more extensive panel of pro- and anti-inflammatory markers and a wider range of stimulus parameters to assess the impact that VNS has on immune state in both pregnant mothers and their offspring.

Further neurodevelopmental evaluation of neonates exposed to vagal nerve stimulation would provide a broader understanding of longer-term consequences of maternal VNS therapy. Assessing systemic C-reactive protein and pro-calcitonin as well as pro- and anti-inflammatory cytokines in babies born to these mothers would provide further information about the immune response evoked by VNS. Ultimately, the goal is to understand the effects of VNS on neuroinflammatory processes that contribute to the physiology of pregnancy, parturition, and fetal neurodevelopment. It may be possible to tune VNS to arrest or prophylactically prevent chorioamnionitis or other sources of inflammation in the mother and improve maternal and fetal outcome.