Umbilical cord occlusions in near-term ovine fetus induce increased beat-to-beat heart rate variability correlating to decreases in neuroinflammation: a case for the afferent cholinergic anti-inflammatory pathway?

Supported by grants from the Canadian Institute of Health Research (CIHR) and Lawson Health Research Institute (LHRI) Internal Research Fund (MGF and BSR); CIHR, Fonds de la recherche en santé du Québec (FRSQ) (MGF). BSR is the recipient of the Canada Research Chair in Fetal and Neonatal Health and Development. Neuroinflammation in utero may contribute to brain injury resulting in life long neurological disabilities. The pivotal role of the efferent cholinergic anti-inflammatory pathway (CAP) in controlling inflammation has been described in adults, but its importance in the fetus is unknown. Moreover, it is unknown whether CAP may also exert anti-inflammatory effects on the brain via CAP’s afferent component of the vagus nerve. Based on multiple clinical studies in adults and our own work in fetal autonomic nervous system, we gauged the degree of CAP activity in vivo using heart rate variability measures reflecting fluctuations in vagus nerve activity. Measuring microglial activation in the ovine fetal brain near-term, we show in vivo that afferent fetal CAP may translate increased vagal cholinergic signaling into suppression of cerebral inflammation in response to near-term hypoxic acidemia as might occur during labour. Our findings suggest a new control mechanism of fetal neuroinflammation via the vagus nerve, providing novel possibilities for its non-invasive monitoring in utero and for targeted treatment.

Neuroinflammation in utero may contribute to brain injury resulting in life long neurological disabilities. 33 The pivotal role of the efferent cholinergic anti-inflammatory pathway (CAP) in controlling 34 inflammation has been described in adults, but its importance in the fetus is unknown. Moreover, it is 35 unknown whether CAP may also exert anti-inflammatory effects on the brain via CAP's afferent Induced animal sepsis and clinical-pathologic studies in adults indicate that loss of the cholinergic anti-46 inflammatory pathway's (CAP) inhibitory influence unleashes innate immunity, producing higher levels 47 of pro-inflammatory mediators that exacerbate tissue damage. This decrease in CAP activity also 48 decreases short-term heart rate variability (HRV), e.g., as measured by the beat-to-beat HRV measures, 49 such as root mean square of successive differences in R-R intervals of ECG (RMSSD), a measure of 50 vagal modulation of HRV. [1,2] Thus, short-term HRV measures reflect CAP activity in adults.
[3] Of 51 note, RMSSD also reflects vagal activity in fetal sheep. [4] 52 Increased CAP vagal activity inhibits the release of pro-inflammatory cytokines such as interleukin (IL)-53 1β. [1] This systemic CAP effect is mediated via the α7 nicotinic acetylcholine receptor (α7nAChR) 54 expressed on macrophages. [5] However, recent studies have shown a similar α7nAChR-dependent 55 effect in brain microglia in vitro. [6,7],[8] 56 In adult species, high-mobility group box protein 1 (HMGB1), a non-histone DNA-binding protein, acts 57 as an important pro-inflammatory cytokine linking necrosis with ensuing inflammation by translocating 58 from the neuronal nucleus to the cytosol and then to the extracellular space, leading to microglial 59 activation.
[9] Much attention has been paid to the effects of α7nAChR stimulation on HMGB1 60 secretion because of its therapeutic potential to treat sepsis; HMGB1 represents a crucial link between 61 neuronal necrosis and the cerebral inflammatory response mediated by microglia, thus impacting the 62 long term outcome of neurological injury. [10,9] HMGB1 also acts as a potent pro-inflammatory 63 cytokine when secreted by microglia in response to inflammatory stimuli. [ First, we hypothesized that the fetal inflammatory response induced by hypoxic-acidemia will result in 73 an increase of systemic CAP activity as a compensatory mechanism and an inhibitory effect of CAP on 74 the cerebral inflammatory response. The systemic inflammatory response will be reflected by an 75 increased vagal activity and hence a correlation of RMSSD and IL-1β. 76 Second, we sought to determine the effect of fetal hypoxic-acidemia insult on brain regional activation 77 of the microglia expressing α7nAChR, and the relation of systemic and cerebral CAP activation to the 78 intracellular HMGB1 localization in these cells. Thus, we hypothesized that the cerebral inflammatory 79 response will result in microglial HMGB1 translocation from the nucleus to the cytosol due to increased 80 microglial activation via α7nAChR and this HMGB1 translocation will correlate with the degree of 81 CAP's vagal activation measured by RMSSD.
[16] 85 RMSSD and IL-1β increased ~2 fold from baseline versus the time of nadir pH (p < 0.05), and fell by 1 86 h of recovery ( Fig. 2A, 2B; for IL-1β cf. [17]). Of note, at 1 h of recovery the values of RMSSD and IL-87 1β were still clearly, but not statistically significantly elevated. Baseline, nadir pH and 1 h of recovery 88 RMSSD correlated to corresponding IL-1β levels at R = 0.57 (p=0.02, n=17, Fig. 2C). 89 Fetal gender did not contribute to this correlation. As reported, MG cell counts were increased in the 90 white matter of the treated animals versus the control group.
[17] RMSSD measurements at baseline and 91 1 h of recovery correlated inversely to white matter MG cell counts determined at 24 h of recovery at R 92 = -0.71 (p=0.05) and R = -0.89 (p=0.03), respectively (Fig. 2D). 93 For both HMGB1 translocation index and α7nAChR expression, the main effect of belonging to the 94 UCO or the control group was not significant (p=0.25 and p=0.54, respectively, Fig. 3). However, the 95 effect of group on HMGB1 translocation differed according to brain region (mostly in hippocampus and 96 GM46) and microglia status (p<0.001 for interaction terms) (Table 1). Interestingly, statistically 97 significant interaction effects on HMGB1 translocation were observed in cortical and hippocampal 98 regions, but not in the subcortical (thalamus, white matter) brain regions. In grey matter, these effects 99 applied to quiescent MG (qMG), but not to activated MG (aMG); meanwhile, in hippocampus this was 100 mostly apparent in aMG. 101 Similarly, a model that accounted for interactions of group and microglia status and HMGB1 was able to 102 predict α7nAChR expression (p<0.001). Notably, the effect of UCO group was in the same direction but 103 of much greater magnitude in aMG compared to qMG (Table 2). 104 In parallel, RMSSD at 1 hour of recovery correlated highly with cytosolic HMGB intensity per area in 105 aMG of thalamus (R = -0.94, p=0.005, Fig. 4A) and RMSSD at pH nadir correlated with α7nAChR 106 intensity per area in aMG of WM (R = 0.83, p=0.04, Fig. 4B). Similar to the relationship shown in Fig.   107 4A, within both qMG and aMG in WM, HMGB1 translocation index correlated to RMSSD at 1 hour of 108 recovery (R = -0.83, p=0.04 and R = -0.89, p=0.02, respectively). This finding was again replicated for 109 qMG of GM13 and aMG of GM46: HMGB1 translocation index correlated there to RMSSD at pH nadir 110 (R = -0.99, p<0.001 and R = -0.83, p=0.04, respectively). That is, higher RMSSD values correlated with 111 lower HMGB1 translocation and higher α7nAChR intensity per area in brain region-specific and 112 microglia status-specific manner.

114
We demonstrate that UCO-induced insult simulating human labour results in increasing fHRV 115 properties known to reflect fluctuations of vagal activity which are correlated with systemic and brain 116 inflammatory responses and with shifts in nucleus-cytosol HMGB1 distribution in α7AChRs-positive 117 microglia. We propose that this behaviour is due to an increase in afferent CAP activity exerting anti-118 neuroinflammatory effects via microglial α7AChRs by limiting microglial activation. We observe that 119 such neuroprotective effects occur in a brain region-dependent manner.

122
Our findings support the first hypothesis that near-term asphyxia with worsening acidemia leads to vagal 123 activation correlated to the degree of systemic inflammatory response, which suggests CAP activation. 124 [4],[16],[17] Fetal acidemia, but not intermittent chronic hypoxia alone, seems to be required for an 125 aseptic induction of fetal inflammatory response and activation of CAP.
[17] Consequently, the 126 inhibition of brain regional cellular innate immune response was found in fetuses that were subject to 127 acute worsening acidemia, but not to chronic intermittent hypoxia. Of note, duration of UCO insults, but reaching most severe acidosis. The α7nAChR expression is not static within a given brain region, but 163 rather appears to be dynamically regulated by stimuli such as hypoxic-ischemic injury which down-164 regulates α7nAChR mRNA, unless the receptor is agonistically stimulated. [7] We propose that such 165 stimulation may not only be exogenous, pharmacologically driven, but also endogenous and occurring 166 via afferent CAP. Indeed, fetal cerebral inflammatory response and ensuing CAP activation resulted in a 167 pronounced microglial α7nAChR intensity per area increase in the periventricular WM and reduced 168 cytoplasmic HMGB1 intensity per area in the thalamus, which suggests that CAP homeostatically 169 modulates microglia activity towards a neuroprotective phenotype via α7nAChR, similar teleologically 170 to what has been proposed by Tracey et al. for systemic peripheral efferent effects of CAP. [5] 171 We identified microglia status-dependent differences in HMGB1 translocation between cortical and 172 subcortical brain regions. These differences seemed amplified by the UCO insult. UCOs are known to 173 induce a redistribution of the regional cerebral blood flow (rCBF) toward subcortical structures. minutes, and this series was continued until the targeted fetal arterial pH was attained. Following the 274 mild as well as the moderate UCO series 10 minute periods with no UCO were undertaken, during 275 which fetal arterial blood was sampled and arterial blood pressure, ECOG, and ECG data were recorded 276 in the absence of fetal heart rate decelerations. After attaining the targeted fetal arterial pH of <7.00 and 277 stopping the repetitive UCO, animals were allowed to recover for ~24 hours. 278 Fetal arterial blood samples were obtained during the baseline period (3 mL), at the end of the 1 st UCO 279 of each UCO series (1 mL), and ~5 minutes after each UCO series (3 mL). In addition, fetal arterial 280 blood samples were obtained between UCO at ~20 and 40 minutes of the moderate and severe UCO 281 series (1 mL), and at 1, 2 and 24 hours of recovery (3 mL). At ~4 p.m. on day 2, the animals were killed 282 as described above. Probes/Invitrogen, Carslbad CA). All tissue sections for analysis were processed simultaneously, using 349 pooled reagents and antibodies for consistency. Images were captured on a Zeiss AxioImager Z1 350 microscope, equipped with an Apotome grid-a structured illumination device, which isolates optical 351 slices much like a confocal microscope (Carl Zeiss Canada Ltd, Toronto, ON). For each animal, eight 352 random fields of view in each brain region were collected for analysis. All images within each brain 353 region were captured at the same instrument settings to ensure consistent illumination and detection 354 parameters among samples. To avoid crosstalk between channels that might create non-specific intensity 355 signals, bandpass dichroic filters for each dye were carefully selected based on the spectral profiles of 356 the fluorescent tags, and tested against controls that are positive for that desired dye wavelength, but 357 negative for all other dyes used. Images were captured sequentially using one filter set at a time for each 358 channel. The image analysis was done in Image Pro Plus 7.0 (Media Cybernetics, Bethesda, MD, USA). 359 We had co-labeled with the HMGB1, α7 nAChR, Iba1 and DAPI, yielding 4 separate channels that we 360 used as follows. First, we created a binary mask of the DAPI channel (white nuclei/ black background), then subtracted that from the Iba1 channel image. Since pure binary white has an intensity of 255, the 362 subtraction left us with an Iba1 channel image with black holes of 0 intensity where the nuclei had been 363 located, i.e., an image with intensity only in the cytosolic area. Second, we created a threshold on the 364 resulting Iba1 image to isolate the outlines of the cytosolic region only of the microglial cell bodies. 365 Third, using the "Load Outlines" function in Image Pro, we were able to apply the outlines of the 366 cytosolic area from the Iba1 image of the matching images, and measure data only within those outlines. 367 Fourth, we created similar outlines of the nuclei using the DAPI channel, and applied those onto the 368 HMGB1 and α7 nAChR channels respectively, to obtain the intensity and area data solely from the while qMG did not exhibit either of these features (Fig. 1). We recognize that such distinction, albeit 379 based on general consensus regarding distinct morphological features of active (hypertrophy and 380 increased processes) versus quiescent microglia, is also somewhat subjective, when it comes to the 381 notion of co-localization to or engulfing of neurons. We return to this in the Discussion section. 382 For analyses of HMGB1 and α7 nAChR intensity per area between the brain regions within each group 383 we had to take into account that their absolute values could not be compared due to region-specific 384 optimized acquisition as described above. For HMGB1, we worked around this limitation by deriving 385 relative measures of HMGB1 intensity per area as cytosol/nucleus ratio for each brain region, as 386 HMGB1 translocation indices (the higher the ratio, the more translocation occurs), thus making them 387 comparable.

389
Fetal heart rate variability analysis to gauge activity of the cholinergic anti-inflammatory pathway 390 The fHRV methodology was described elsewhere. [2, 42] Briefly, R peaks were triggered by steepest 391 ascent criterion to derive fHRV, and RMSSD was calculated from five minutes of fHRV using Matlab     group*brain region*microglia status interaction (p<0.001). See Table 1 for details. B. Effect of UCO 543 and microglia status on α7 nAChR immunofluorescence measured as intensity per area: significant 544 group*microglia status*HMGB1 translocation interaction (p<0.001). See Table 2 for details. 545 Note that, for α7 nAChR signal, between-brain regions comparisons were not possible, because gain 546 settings were optimized for each brain region and kept constant between cell compartment and groups 547 (but not from region to region). HMGB1 signal is expressed as ratio of cytosolic to nuclear signal, i.e., 548 the higher the ratio, the more HMGB1 translocation is observed; this normalization permits between-549 brain regions comparisons.