Focal, but not global, cerebral ischaemia causes loss of myenteric neurons and upregulation of vasoactive intestinal peptide in mouse ileum

Summary

Reduced blood flow to the brain induces cerebral ischaemia, potentially causing central injury and peripheral complications including gastrointestinal (GI) dysfunction. The pathophysiology behind GI symptoms is suspected to be neuropathy in the enteric nervous system (ENS), which is essential in regulating GI function. This study investigates if enteric neuropathy occurs after cerebral ischaemia, by analysing neuronal survival and relative numbers of vasoactive intestinal peptide (VIP) and neuronal nitric oxide synthase (nNOS) expressing neurons in mouse ileum after three types of cerebral ischaemia. Focal cerebral ischaemia, modelled by permanent middle cerebral artery occlusion (pMCAO) and global cerebral ischaemia, modelled with either transient occlusion of both common carotid arteries followed by reperfusion (GCIR) or chronic cerebral hypoperfusion (CCH) was performed on C56BL/6 mice. Sham‐operated mice for each ischaemia model served as control. Ileum was collected after 1–17 weeks, depending on model, and analysed using morphometry and immunocytochemistry. For each group, intestinal mucosa and muscle layer thicknesses, neuronal numbers and relative proportions of neurons immunoreactive (IR) for nNOS or VIP were estimated. No alterations in mucosa or muscle layer thicknesses were noted in any of the groups. Loss of myenteric neurons and an increased number of VIP‐IR submucous neurons were found in mouse ileum 7 days after pMCAO. None of the global ischaemia models showed any alterations in neuronal survival or relative numbers of VIP‐ and nNOS‐IR neurons. We conclude that focal cerebral ischaemia and global cerebral ischaemia influence enteric neuronal survival differently. This is suggested to reflect differences in peripheral neuro‐immune responses.

Cerebral ischaemia, commonly called stroke, results from a reduction in blood flow to the brain and is manifested by, for example, hypoxia, energy depletion, inflammation, cellular injury or death (Harukuni & Bhardwaj 2006; Offner et al. 2006; Doyle et al. 2008). There are two main types of cerebral ischaemia: focal and global. Focal ischaemia results when blood flow to restricted parts of the brain is insufficient, while global cerebral ischaemia involves widespread areas of the brain. In addition to the ischaemic insults, restoration of blood flow to the brain postischaemia exacerbates the injury through increased levels of oxidative stress, inflammation, and homeostatic imbalance (Offner et al. 2006; Doyle et al. 2008; Khatri et al. 2012). Besides brain injury, peripheral organ dysfunction has been observed after cerebral ischaemia, including gastrointestinal (GI) tract functional disorder (Schaller et al. 2006; Hsu et al. 2009; Ma et al. 2015). Studies on the mechanisms behind central ischaemia‐induced intestinal dysfunction indicate mucosal barrier disruption (Feng et al. 2010; Tascilar et al. 2010), or immune system disorders (Schulte‐Herbrüggen et al. 2009). Based on our previous study (Cheng et al. 2016), cerebral ischaemia also influences the enteric nervous system (ENS), causing neuronal death in both ileum and colon.

The ENS is an extensive intrinsic nervous system within the GI tract. It plays major roles in regulating GI motility, secretion and local blood flow and it also interacts with the immune and endocrine systems (Furness 2012; Veiga‐Fernandes & Pachnis 2017). Changes in neuronal structure, numbers, or transmitter expression, known as enteric neuroplasticity, alter GI functions (Brierley & Linden 2014). Neuroplasticity is essential in intestinal adaption but also occurs in a number of diseases (Giaroni et al. 1999; Lomax et al. 2005; Vasina et al. 2006; Schäfer et al. 2009; Brierley & Linden 2014).

In the ENS, the neurotransmitters nitric oxide (NO) and vasoactive intestinal peptide (VIP) are ascribed important roles in neuronal maintenance and protection (Sandgren et al. 2003a,b; Lin et al. 2004; Voss et al. 2012; Voss & Ekblad 2014) as well as in several pathophysiological conditions (Sandgren et al. 2002; Ekblad & Bauer 2004; Sigalet et al. 2010; Abad et al. 2012; Rivera et al. 2012; Padua et al. 2016).

We have recently described significant enteric neuronal losses in myenteric ganglia from mouse ileum 3 and 7 days after permanent middle cerebral artery occlusion (pMCAO), a model of focal cerebral ischaemia (Cheng et al. 2016). Here, we expand on previous study focusing on neuronal survival and relative numbers of VIP and nNOS expressing neurons in mouse ileum, including not only pMCAO but also global cerebral ischaemia with reperfusion (GCIR), and chronic cerebral hypoperfusion (CCH). The aims of the two latter are to mimic cardiac arrest with successful cardiopulmonary resuscitation and chronic hypotension respectively.

Material and methods

Animals

C57BL/6 mice (male, weight 22–27 g) purchased from Charles River were used in GCIR experiments. C57BL/6 mice (male, weight 25–30 g) used for pMCAO and CCH were bred in‐house. For all models, mice were anesthetized with 5% isofluorane (IsobaVet, Schering‐Plough Animal Heath, GB) in oxygen and maintained on 2% isofluorane. Postoperative pain relief was by subcutaneous administration of Marcain® (1.25 mg/kg). The endpoint of each model was set in regard to the central neuronal injury (Mracskó et al. 2010; Cheng et al. 2016; Svensson et al. 2016; Liu et al. 2017b), and no impact on enteric neurons is to be expected until after the brain injury.

Animal models

Focal cerebral ischaemia

Focal cerebral ischaemia was achieved using the pMCAO model as previously described (Clausen et al. 2014; Cheng et al. 2016). Mice (sham n = 5, pMCAO n = 7) had an incision made between the left lateral part of the orbit and the left ear, and the parotid gland and the temporal muscle were dislocated in distal direction. Craniotomy, above the anterior distal branch of the middle cerebral artery (MCA), was performed with a 0.8‐mm high‐speed microdrill. The MCA was exposed and occluded permanently by electrocoagulation using an electrosurgical unit (ICC 30; Erbe, D). Interruption of blood flow was confirmed by visual inspection. Sham‐operated animals were subjected to the same surgical protocol, except that no occlusion of the MCA was made. Mice were sacrificed after 7 days.

Global cerebral ischaemia with reperfusion (GCIR)

Global cerebral ischaemia with reperfusion was induced by transient occlusion of both common carotid arteries (Olsson et al. 2003; Svensson et al. 2016). Mice (sham n = 5, GCIR n = 5) had a cut made parallel to the trachea, and the common carotid arteries were exposed bilaterally. The arteries were encircled with a thin silk thread and occluded using micro‐aneurysm clips for 13 min. Clips and thread were then removed and restoration of blood flow to each artery confirmed by visual inspection. Sham operation was performed using the same surgical protocol except that arteries were not occluded. Mice were sacrificed after 14 days.

Chronic cerebral hypoperfusion (CCH)

Chronic cerebral hypoperfusion was performed by bilateral reduction in blood flow from the common carotid arteries to the brain (Shibata et al. 2004; Manouchehrian et al. 2015). Mice (sham n = 5 and CCH n = 5) had a small cut made parallel to the trachea. The common carotid arteries were exposed bilaterally and narrowed with metal coils (wire diameter of 0.08 mm; inner diameter (ID): 0.18 mm; pitch: 0.50 mm; total length: 2.5 mm; surface: Au‐coated, Invitrotech Co., LTD. Shimogasa‐cho, Kusatsu, Japan), to reduce blood flow to the brain by 30%. The sham group underwent the same operation but without inserting the coils. Mice were sacrificed after 17 weeks.

Tissue collection

Mice were sacrificed by cardiac puncture under anaesthesia (isoflurane, >2.5%). The abdomen was opened and the visceral organs inspected. Distal small intestine was collected, fixed in 4% paraformaldehyde in 0.1 M phosphate buffer at 4°C overnight and rinsed in Tyrode's solution (containing 10% sucrose) three times. Specimens from ileum were orientated and mounted for longitudinal sectioning in FSC 22^® Clear (Leica Biosystems, Stockholm, Sweden), frozen and sectioned (10 μm). The sections were processed for immunofluorescence and morphometry.

Morphometric analyses

Morphometry was performed on toluidine blue‐stained sections. In brief, sections were washed 10 min in PBS‐T (PBS with 0.25% Triton X‐100) followed by 2 min in toluidine blue (0.01% toluidine blue in 60% ethanol). After washing in 70% ethanol for 5 min, 99.5% ethanol for 5 min and xylene for 1 min, sections were mounted in Pertex®. The heights of mucosa and muscularis propria were expressed as the mean of 5–10 representative measurements from each animal using a computerized image analysing system (NDP view 2, Hamamatsu, Japan).

Immunofluorescence

Sections were subjected to antigen retrieval by microwaving 2 × 8 min at 650 W in citrate acid buffer (0.01 M, pH = 6). After cooling, sections were washed 20 min in running tap water and rinsed in PBS‐T. Sections were incubated with an (single immunolabelling) antiserum raised in rabbit against nNOS (code no. 9223, dilution 1:1200 (Ekblad et al. 1998), Euro‐Diagnostica, Malmö, Sweden) or against VIP (code no. 7852, dilution 1:1200 (Ekblad et al. 1998), Euro‐Diagnostica) or with a mix of two primary antibodies (double immunolabelling) raised in different species. Mouse monoclonal antibodies against the pan‐neuronal marker HuC/D (code no. A21272, dilution 1:400 (Lin et al. 2003), Thermo Fisher Scientific, Stockholm, Sweden), were used in combination with antibodies against either nNOS or VIP raised in rabbit (see above), at 4°C overnight. Sections were then rinsed in PBS‐T for 3 × 10 min. For visualization of single immunolabelling, sections were exposed to fluorescein‐conjugated swine anti rabbit IgG (DAKO, DK) in dilution 1:400 for 1 h. In double immunolabelling, sections were incubated with a mix of secondary antibodies raised in goat against mouse IgG (Alexa Fluor488, 115‐545‐166, 1:1000) and in donkey against rabbit IgG (Alexa Fluor594, 711‐585‐152, 1:1500 Jackson Immuno Research, Stockholm, Sweden) at room temperature for 1 h. Mounting was in PBS: glycerol 1:1, and the sections were analysed using epifluorescence microscopy (Olympus BX43, LRI, Lund, Sweden) with appropriate filter setting.

Analysis

To minimize variability in size of myenteric ganglia, the numbers of HuC/D‐immunoreactive (IR) myenteric and submucous neurons were counted in longitudinally cut sections and expressed as number of neurons/mm section. The subpopulations of enteric neurons containing nNOS or VIP were counted on sections double‐immunostained for HuC/D and nNOS or VIP. The relative numbers of VIP‐ and nNOS‐IR neurons were expressed in percentage of HuC/D‐IR neurons. At least 10 mm from a minimum of three sections cut at different depths (at least 200 μm between each section) was estimated from each mouse. Counting was performed by one blinded observer (XC).

Statistics

Data are presented as means ± SEM. T‐test was used to determine possible differences between sham and ischaemic group. A confidence level of 95% was considered significant.

Ethics approval and consent to participate

All procedures and treatments were approved by the regional Malmo/Lund committee for experimental animal ethics, Swedish board of Agriculture [(pMCAO (M429‐12), GCIR (M426‐12) and CCH (M425‐12)]. Animals were used in accordance with the European Community Council Directive (2010/63/EU) and the Swedish Animal Welfare Act (SFS 1988:534).

Results

General observations

All mice returned to normal activity after surgery. The extent of the ischaemic brain injuries is assessed in separate studies (pMCAO and CCH (Deierborg, unpublished); GCIR (Svensson et al. 2016)). No lesions or abnormalities were noted upon inspection of the visceral organs at necropsy. No differences in mucosal or muscle layer thicknesses between sham and ischaemic group were detected in the morphometric analyses (see Table 1).

Table 1.

Thicknesses of muscularis propria and mucosa from mouse ileum subjected to cerebral ischaemia or sham‐operated

Group	Muscularis propria (μm)	Mucosa (μm)
pMCAO	54.3 ± 8.4	370.2 ± 29.3
pMCAO Sham	43.1 ± 5.5	308.9 ± 6.3
GCIR	40.2 ± 2.8	315.9 ± 11.4
GCIR Sham	38.2 ± 1.4	313.7 ± 8.3
CCH	40.5 ± 2.6	263.2 ± 6.3
CCH Sham	33.3 ± 5.9	247.9 ± 13.7

CCH, chronic hypoperfusion; GCIR, global cerebral ischaemia with reperfusion; pMCAO, permanent middle cerebral artery occlusion.

Thicknesses of muscularis propria and mucosa from mouse ileum subjected to permanent middle cerebral artery occlusion (pMCAO), global cerebral ischaemia with reperfusion (GCIR) and chronic cerebral hypoperfusion (CCH) and their respective sham groups. Results are presented as means ± SEM. No statistically significant differences were noted between sham and ischaemic groups.

HuC/D‐, VIP‐ and nNOS‐IR neurons in mouse ileum

HuC/D immunoreactivity was observed in both the cytoplasm and the nuclei in myenteric as well as in submucous nerve cell bodies (Figure 1c). VIP (Figure 1a) and nNOS immunoreactivity (Figure 1b, d) were observed in both nerve cell bodies and nerve fibres. VIP‐IR nerve cell bodies were numerous in submucous ganglia but few in myenteric ganglia (Figure 1a). Numerous nNOS‐IR nerve cell bodies were found in myenteric ganglia (Figure 1b, d), but they were few in submucous ganglia. VIP‐IR nerve fibres (Figure 1a) were found in all layers within the intestinal wall, including submucous and myenteric ganglia. Numerous nNOS‐IR fibres (Figure 1b, d) were found within the smooth muscle layers and in the submucous and myenteric ganglia but they were sparse in the mucosa/submucosa. Microanatomy, numbers and topographic distribution of VIP‐ and NOS‐IR neurons were similar in the different study groups.

Figure 1.

Cryostat sections of ileum from sham‐operated mouse stained for VIP, nNOS and double‐immunostained for the neuronal marker HuC/D and nNOS. Upper panel illustrates the topographic distribution of nerve fibres immunoreactive for (a) VIP and (b) nNOS. VIP‐IR neurons are found particularly in submucous ganglia (arrows) issuing a rich nerve fibre network terminating in both mucosa/submucosa and muscularis propria. nNOS‐IR neurons are particularly found in myenteric ganglia (arrow heads) with nerve fibres within the muscularis propria. (c–e) A subpopulation of HuC/D‐immunoreactive enteric, particularly myenteric, neurons contain nNOS as revealed by the merged micrograph (e). Muc: mucosa; CM: circular muscle; LM: longitudinal muscle; SG: submucous ganglia; MG: myenteric ganglia; Bar = 20 μm. [Colour figure can be viewed at http://wileyonlinelibrary.com]

Myenteric neuronal loss after pMCAO, but not after GICR or CCH

Compared to sham, pMCAO caused a significant loss of myenteric (9.9 ± 0.7 and 7.4 ± 0.6, P < 0.05), but not submucous (3.6 ± 0.4 and 3.0 ± 0.3) neurons in ileum. In contrast, no change in neuronal survival was found between sham and GCIR, in myenteric (11.0 ± 1.3 and 11.8 ± 1.2) or in submucous (3.3 ± 0.4 and 3.8 ± 0.3) neurons. Also no difference in neuronal numbers between sham and CCH was found in myenteric (10.1 ± 0.8 and 8.2 ± 0.7) and submucous (3.1 ± 0.1 and 2.7 ± 0.3) ganglia. Results are summarized in Figure 2a, d, g.

Figure 2.

Effects of cerebral ischaemia on neuronal number and relative numbers of VIP‐ and nNOS‐IR neurons. a‐c show effects on enteric neurons after sham operation or permanent middle cerebral artery occlusion (pMCAO). (a) A significant neuronal loss is noted post‐pMCAO in myenteric, but not submucous, ganglia. (b) An increased relative number of VIP‐IR neurons is observed in submucous, but not myenteric, ganglia post‐pMCAO. (c) The relative numbers of nNOS‐IR neurons are unchanged post‐pMCAO, in both myenteric and submucous ganglia. d‐f show effects on enteric neurons after sham operation or global cerebral ischaemia with reperfusion (GCIR). No change in neuronal survival (d), or in relative numbers of VIP‐ (e) or nNOS‐IR (f) neurons are observed after GCIR, neither in myenteric nor in submucous ganglia. g‐i show effects on enteric neurons after sham operation or chronic cerebral hypoperfusion (CCH). No change in total neuronal number (g), or in relative frequency of VIP‐ (h) or nNOS‐IR (i) neurons is observed after CCH. Results on total neuronal number (a, d, g) are expressed as number of neurons per mm section, mean ± SEM. The relative numbers of VIP‐IR (b, e, h) or nNOS‐IR (c, f, i) are expressed in percentage of HuC/D‐IR neurons, means ± SEM. pMCAO (sham n = 3–4, PMCAO n = 4–7), GCIR (sham n = 4–5, GCIR n = 5), CCH (sham n = 3–5, CCH n = 5). *P < 0.05, **P < 0.01.

Upregulation of VIP‐IR submucous neurons after pMCAO, but not after GCIR or CCH

Compared to sham, a significant increase in the relative number of VIP‐IR nerve cell bodies in submucous ganglia was found after pMCAO (20.4 ± 1.4% and 30.1 ± 1.8%, P < 0.01) while their numbers in myenteric ganglia were unchanged (2.0 ± 1.1% and 1.6 ± 0.7%). Compared with their respective sham group, no differences in relative numbers of VIP‐IR neurons were found after either GCIR (myenteric neurons: 2.0 ± 0.8% and 2.8 ± 1.1%; submucous neurons 34.8 ± 1.5% and 35.2 ± 4.5%) or CCH (myenteric neurons: 1.1 ± 0.6% and 2.1 ± 1.4%; submucous neurons 16.8 ± 6.9% and 27.7 ± 7.2%). Results are summarized in Figure 2b, e, h.

No change in relative numbers of nNOS‐IR neurons in pMCAO, GCIR or CCH

The relative numbers of nNOS‐IR nerve cell bodies did not differ between the three sham, the pMCAO (myenteric neurons: 21.5 ± 3.4% and 21.7 ± 3.9%; submucous neurons: 2.3 ± 1.3% and 1.6 ± 1.6%)‐, the GCIR (myenteric neurons: 21.7 ± 4.1% and 17.6 ± 2.2%; submucous neurons: 3.8 ± 0.5% and 2.4 ± 0.7%)‐ and the CCH (myenteric neurons: 18.6 ± 1.2% and 20.7 ± 2.5% and submucous ganglia: 0.5 ± 0.5% and 0.8 ± 0.6%)‐subjected groups. Results are summarized in Figure 2c, f, i.

Discussion

Brain ischaemia leads to central cellular injury but it also influences several peripheral organs, via, for example, the immune and/or the nervous systems (Ma et al. 2015). In the GI tract, a disrupted mucosal barrier (Feng et al. 2010), reduced T‐ and B‐cell counts in Peyer's patches (Schulte‐Herbrüggen et al. 2009) and enteric neuronal loss (Cheng et al. 2016) have been observed in experimental stroke models. In this study, we investigated enteric neuronal survival and plasticity in mouse ileum after three different types of cerebral ischaemia.

Loss of myenteric neurons induced by pMCAO, but not GCIR or CCH

In the present study, we found that, in contrast to pMCAO, global cerebral ischaemic models caused no loss of enteric neurons. The pMCAO‐induced enteric neuronal loss has previously been shown to involve the beta‐galactoside binding protein galectin‐3 (gal‐3) (Cheng et al. 2016). Gal‐3 has immune modulatory capabilities acting as an endogenous Toll‐like receptor 4 (TLR4) agonist (Burguillos et al. 2015). Purified gal‐3 and the bacterial endotoxin lipopolysaccharide (LPS) induce loss of myenteric neurons in vitro through a common TLR4‐activated pathway involving transforming growth factor beta‐activated kinase 1 (TAK1) and AMP‐activated protein kinase (AMPK) (Cheng et al. 2016). The absence of enteric neuronal losses in the global cerebral ischaemic models was unexpected. This is because both models, like pMCAO (Cheng et al. 2016), are reported to cause loss of central neurons and induction of a pro‐inflammatory response, including elevated gal‐3 and TLR‐4 levels in brain and retina (Hua et al. 2007; Lee et al. 2015; Manouchehrian et al. 2015; Svensson et al. 2016). The prevailing view is that central neuronal losses in all three stroke models used involve activation of a common gal‐3/TLR4 pathway. However, the models differ in certain aspects, possibly explaining their different effects on the ENS. The pMCAO model induces permanent damage to local cerebral blood flow, while GCIR causes a transient occlusion and CCH a chronic reduction in central cerebral blood flow. Thus, the neurovascular barrier as well as the peripheral immune system activation may be differently affected. Several authors have reported that pMCAO and GCIR cause disruption of the neurovascular barrier (Liu et al. 2017a,b), while it remains intact after CCH (Farkas et al. 2007). Both reversible MCAO and permanent MCAO (Offner et al. 2006; Wang et al. 2006; Cheng et al. 2016) as well as CCH (Farkas et al. 2007) have been studied in relation to peripheral immune system activation. Where the MCAO models show a robust and sustained peripheral immune response, CCH displays a brief (hours) transient activation of circulating leucocytes and neutrophils. CCH in addition shows a sustained elevation of corticosterone after 3 months (Farkas et al. 2007). This early transient inflammatory response and sustained corticosterone‐induced immune suppression in combination with an intact neurovascular barrier may lead to a cerebral containment of the CCH effects, explaining the maintenance of enteric neurons after CCH. Concerning the GCIR model, little is known about evoked peripheral neuroimmune responses. A recent study showed elevated serum levels of the cytokines interleukin (IL)‐5 and IL‐6 14 days after GCIR (Svensson et al. 2016). Differences in the peripheral early response between pMCAO and GCIR may be of importance, as the enteric neuronal losses in pMCAO were evident already after 3 days (Cheng et al. 2016). It is interesting in this regard that pro‐inflammatory cytokines may play dual roles from harmful to protective depending on their activation pathway. Such cytokines include IL‐1β and tumour necrosis factor α (TNF‐α), which are rapidly elevated in the brain after induction of GCIR and pMCAO (Chu et al. 2012; Yin et al. 2013; Doll et al. 2014; Svensson et al. 2016). In addition, they are able to mediate neurotrophic responses in enteric neurons (Gougeon et al. 2013). If such mechanisms are at play in the current investigation, we can only speculate. Furthermore, during cellular stress, compensatory mechanisms are initiated to maintain primary functions and protect against the injurious environment (Lomax et al. 2005; Brierley & Linden 2014). It can be speculated that compensatory mechanisms activated after GCIR lead to a peripheral protection, for example, by way of peripheral ‘signal dilution’.

Intestinal inactivity causing atrophy and tissue remodelling has been shown to affect neuronal size and numbers (Ekelund & Ekblad 1999). This was, however, not the case in the models investigated in this study, as morphometric analysis showed normal intestinal morphology.

VIP upregulation in submucous ganglia post‐pMCAO, but not after GCIR or CCH

In present study, we found pMCAO, but not GCIR or CCH, to induce an increase in the relative proportion of submucous VIP‐IR neurons. The neuroprotective properties of VIP have been studied in brain (Delgado & Ganea 2003) and in ENS. VIP is found to protect enteric neurons in response to injurious factors both in vivo (Gomariz et al. 2005; Sand et al. 2013) and in vitro (Sandgren et al. 2003a; Arciszewski et al. 2008; Arranz et al. 2008; Sand et al. 2009; Voss et al. 2012). It is speculated that one of the mechanisms by which VIP executes its neuroprotective effects is through its ability to modulate immunological pathways such as TLR4 (Gomariz et al. 2005; Arranz et al. 2006, 2008). Augmented VIP secretion may attenuate a LPS‐induced inflammatory response by downregulating TLR4 expression (Gomariz et al. 2007). In cultured myenteric neurons from mouse ileum, LPS has been shown to induce neuronal loss and upregulation of VIP‐IR neurons (Voss & Ekblad 2014). This may reflect a self‐protective mechanism in response to the innate immune cell activation. As previously described, gal‐3, an endogenous TLR4 agonist (Burguillos et al. 2015), is suggested to be a mediator of the neuronal loss noted after pMCAO (Cheng et al. 2016). LPS is, in pMCAO, not suspected to be responsible for the TLR4 activation as endotoxin/LPS levels were found to be low (Cheng et al. 2016). It may therefore be speculated that increased expression of VIP in submucous neurons, as found after pMCAO, reflects self‐protection against gal‐3‐induced TLR4 activation. In addition, VIP may mediate neuroprotection through other mechanisms, including enhancing intestinal barrier function (Chandrasekharan et al. 2013; Wu et al. 2015). GCIR and CCH did not induce any change in VIP‐IR, at the time points investigated (2 and 17 weeks, respectively); however, considering the normal morphometry and lack of neuronal loss, this finding is not surprising.

nNOS expression was unaffected after cerebral ischaemia

The expression of nNOS gradually increases in the central nervous system after an ischaemic insult (Farkas et al. 2007; Chen et al. 2014; Zhang et al. 2016). However, the functional consequences of this are not fully understood. Increased nNOS correlates positively with microvascular damage and reduced cerebral perfusion (Hsu et al. 2014), but is also ascribed a neuroprotective role in the early defence against excitotoxicity (Zhang et al. 2016). nNOS has not previously been studied in the periphery in regard to cerebral ischaemia, but dual roles of NO release in the ENS have been described (Sandgren et al. 2003b; Lin et al. 2004; Venkataramana et al. 2015). Neuroprotective roles of NO can be seen in vitro where addition of a NO donor promotes and addition of a NOS inhibitor attenuates survival of enteric neurons (Sandgren et al. 2003b; Lin et al. 2004). Moreover, culture per se induces an increased relative frequency of nNOS‐IR enteric neurons. This increase is suggested to primarily reflect a resilience of NO‐producing neurons compared to non‐NO‐producing ones (Sandgren et al. 2003b). The main reasons for increased survival of nNOS‐IR neurons after various insults in vitro have been suggested to be partly due to their NO synthesis and partly to their ability to upregulate their VIP expression (Sandgren et al. 2003a,b). In the current study, the enteric neuronal subpopulations containing nNOS were unchanged in mouse ileum after all three experimental cerebral ischaemia models. It should be noted that nNOS expression does not necessarily reflect the actual NO concentration in the microenvironment. NO can be generated in various nitrogenic species, from different cellular sources and from constitutive and inducible NOS (Galkin et al. 2007). The complexity concerning NO generation and its possible role in both brain and gut after cerebral ischaemia need to be further investigated.

In summary, investigating enteric neuronal changes after three types of cerebral ischaemia throws new light on cerebral ischaemia‐induced GI dysfunction. A significant neuronal cell loss and upregulation of VIP expression were found after focal, but not global, cerebral ischaemia.

Funding

This study was supported by the Swedish Research Council, the Swedish Stroke Association, the Påhlssons Foundation, Sparbanken Färs & Frosta Foundation, the Royal Physiographic Society and the Faculty of Medicine, Lund University. X Cheng acknowledges a PhD scholarship from the China Scholarship Council.

Competing interests

The authors declare that they have no competing interests.

Authors’ contributions

TD, MS and YY performed animal surgery. XC performed histology, morphometry, immunofluorescence and data generation. XC, EE and UV designed the study, analysed the results and wrote the manuscript. All authors approved the final manuscript.