Accepted Manuscript Altered intrinsic and synaptic properties of lumbosacral dorsal horn neurons in a mouse model of colitis Kristen E Farrell, Simon Keely, Marjorie M Walker, Alan M Brichta, Brett A Graham, Robert J Callister PII: DOI: Reference:
S0306-4522(17)30595-X http://dx.doi.org/10.1016/j.neuroscience.2017.08.029 NSC 17978
To appear in:
Neuroscience
Received Date: Revised Date: Accepted Date:
17 January 2017 19 July 2017 14 August 2017
Please cite this article as: K.E. Farrell, S. Keely, M.M. Walker, A.M. Brichta, B.A. Graham, R.J. Callister, Altered intrinsic and synaptic properties of lumbosacral dorsal horn neurons in a mouse model of colitis, Neuroscience (2017), doi: http://dx.doi.org/10.1016/j.neuroscience.2017.08.029
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Altered intrinsic and synaptic properties of lumbosacral dorsal horn neurons in a mouse model of colitis
Kristen E Farrell1,2, Simon Keely1,2, Marjorie M Walker2,3, Alan M Brichta1,2, Brett A Graham1,2 and Robert J Callister1,2 1School
of Biomedical Sciences & Pharmacy, University of Newcastle, Callaghan, NSW, 2308, Australia
2Hunter
Medical Research Institute (HMRI), Rankin Park, NSW, 2305, Australia
3School
of Public Health & Medicine, University of Newcastle, Callaghan, NSW, 2308, Australia
Running title: Colitis alters the functional properties of dorsal horn neurons Key words: in vivo, spinal cord, visceral inflammation Corresponding author Professor Robert J Callister School of Biomedical Sciences and Pharmacy The University of Newcastle Callaghan, NSW, 2308, Australia Telephone: +61 2 49217808 Email: [email protected]
1
Abstract Visceral pain in inflammatory and functional gastrointestinal conditions is a major clinical problem. While the exact mechanisms underlying the development of pain, during and after visceral inflammation, are unknown clinical and preclinical evidence suggests plasticity within the spinal cord dorsal horn is a contributing factor. Here we use an in vivo preparation and patch clamp electrophysiology to test whether the synaptic and intrinsic properties of superficial dorsal horn (SDH) neurons are altered 5 days after the induction of mild colitis in adult male mice (i.e. during acute inflammation of the colon). Whole cell recordings were made from lumbosacral (L6-S1) superficial dorsal horn neurons (SDH), in animals under isoflurane anaesthesia. Noxious colorectal distension (CRD) was used to identify SDH neurons with colonic inputs, while stimulation of the hind paw and tail was employed to assess convergent cutaneous input. Following inflammation, a significantly increased proportion of SDH neurons received both colonic and cutaneous inputs, compared to neurons in naïve animals. In addition, the nature and magnitude of responses to CRD and cutaneous stimulation differed in inflamed animals, as was spontaneous excitatory synaptic drive. Conversely, several measures of intrinsic excitability were altered in a manner that would decrease SDH network excitability following colitis. We propose that during inflammation, sensitization of colonic afferents results in increased signaling to the SDH. This is accompanied by plasticity in SDH neurons whereby their intrinsic properties are changed to compensate for altered afferent activity.
2
Abbreviations list AHP, afterhyperpolarization; ANOVA, analysis of variance; AP, action potential; CNS, central nervous system; CRD, colorectal distension; CRD-NR, colorectal distension-nonresponsive; CRD-R, colorectal distension-responsive; DF, delayed firing; EPSP, excitatory post synaptic potential; ES, effect size; HT, high threshold; IB, initial bursting; IBD, inflammatory bowel disease; IBS, irritable bowel syndrome; IPSP, inhibitory post synaptic potential; LT, low threshold; OR, odds ratio; Rin, input resistance; SDH, superficial dorsal horn; SS, single spiking; TF, tonic firing; TNBS, trinitrobenzenesulfonic acid; U, unclassified; WDR, wide dynamic range
3
Introduction Chronic pain and visceral hypersensitivity are common and debilitating symptoms of several disorders of the gastrointestinal tract, including irritable bowel syndrome (IBS) and inflammatory bowel disease (IBD). IBS is a functional gastrointestinal disorder (FGID) characterized by chronic, recurrent pain and discomfort with low-grade inflammation of the colon mucosa (Verne et al., 2001; Verne et al., 2003; Barbara et al., 2004), whereas IBD describes several chronic inflammatory organic diseases of the gastrointestinal tract. While pain most commonly occurs during disease flare-ups in IBD, 30-50% of patients report IBSlike persistent pain in the absence of active disease (Minderhoud et al., 2004; Farrokhyar et al., 2006; Siegel & MacDermott, 2009). In both conditions, generalized “referred pain” or somatic hyperalgesia is also widely reported by patients (Bernstein et al., 1996; Verne et al., 2001; Verne et al., 2003; Minderhoud et al., 2004). This mixing of visceral and somatic sensory symptoms, or viscerosomatic convergence, is thought to be the result of plasticity within the central nervous system (CNS), particularly in the spinal cord dorsal horn where these sensory pathways can overlap (Ruch, 1961; Farrell et al., 2014b). The precise mechanisms underlying the development and maintenance of chronic visceral and somatic hypersensitivity remain largely unknown (Price et al., 2006; Farrell et al., 2014a; Farrell et al., 2014b).
Our previous review of the literature on CNS plasticity in animal models of visceral hypersensitivity pointed to a crucial role for the dorsal horn (Farrell et al., 2014b). For example, experimental rodent models of colonic inflammation can induce visceral and somatic hyperalgesia (Laird et al., 2001b; Lamb et al., 4
2006). Inflammation also increases neural activation (cFos and pERK), and the release of ‘pain’ neuropeptides (Substance P and CGRP) within the spinal cord dorsal horn (Lu & Westlund, 2001; Sun & Luo, 2004; Traub et al., 2008; Harrington et al., 2012). In addition, functional studies using in vivo extracellular recording have observed that following inflammation, dorsal horn neurons exhibit increased spontaneous action potential (AP) discharge (Al-Chaer et al., 2000; Ness & Gebhart, 2001; Qin et al., 2005) and increased AP discharge in response to distension of the colon (Laird et al., 2001a; Wang et al., 2005b). Together, these studies indicate that visceral inflammation alters the activity or output of dorsal horn neurons, however, the changes to the major determinants of neuron output (i.e. their intrinsic and synaptic properties) that occur during visceral inflammation have yet to be characterized during both acute and chronic models of disease.
Dorsal horn interneurons are involved in the complex processing and integration of inputs from peripheral structures like viscera and skin, higher brain centers, and other local circuit interneurons (Graham et al., 2007). It follows that small changes in their output have the capacity to significantly alter sensory signaling to higher centers and ultimately perception. Until recently, it has proved difficult to study the intrinsic and synaptic properties of the small dorsal horn interneurons that receive visceral inputs. We have developed a mouse preparation that allows high resolution in vivo patch clamp recording from dorsal horn neurons in the spinal segments that receive colonic input (L6-S1) (Farrell et al., 2016). Importantly, this preparation enables detailed analysis of the intrinsic and synaptic properties of neurons that respond to mechanical
5
distension of the colorectum. In this study using our in vivo mouse preparation we hypothesized that the intrinsic and synaptic properties of superficial dorsal horn (SDH) neurons are altered during acute colonic inflammation.
Our data demonstrate that during mild colonic inflammation, the synaptic properties and responses of SDH neurons to colonic and cutaneous stimulation render this population more excitable. Conversely, several changes in the intrinsic properties including a hyperpolarization of membrane potential and an increase in the excitability of presumed inhibitory interneurons are consistent with a decrease in overall SDH excitability. We propose that during inflammation, sensitization of colonic afferents results in increased signaling to the SDH. This is accompanied by plasticity in SDH neurons, which have altered their intrinsic properties in an attempt to compensate for this increased afferent activity.
6
Methods All surgical and experimental procedures were approved by the University of Newcastle’s Animal Care and Ethics Committee (protocol # A-2012-223). Male mice (C57BL/6, 6-8 weeks old, 15.7-27.4 g) were used for all experiments. The adult mouse preparation we used for spinal cord in vivo patch clamp recording and the data collection protocols have been described previously (Graham et al., 2004a, b; Jobling et al., 2010; Farrell et al., 2016). They are briefly summarized here.
Animal model of colitis Colonic inflammation was induced using the modified chemically-induced trinitrobenzenesulfonic acid (TNBS) model of colitis (Wirtz et al., 2007; Marks et al., 2015). Mice were presensitized by epicutaneous application of 1% TNBS (Sigma Chemicals, St. Louis, MO, USA) in acetone/olive oil solution (4:1). After 7 days, mice were anaesthetized with isoflurane (2%) and received a single dose of 0.2ml TNBS (2.5% in 50% ethanol) delivered into the colorectum, 4cm from the anus, via an anal catheter. Mice were suspended upside down for one minute to ensure TNBS remained in the colorectum. Animals were observed daily for changes in body weight, physical appearance and behavior. Electrophysiological experiments were conducted 5 days after the induction of colitis.
Surgery for in vivo spinal cord recording Mice were anaesthetized with isoflurane (2-3%, 2L/min O2 induction; 1-2% maintenance) delivered via a nose cone. When deep anaesthesia was confirmed via the absence of hind limb and corneal reflexes, the animals were head-fixed 7
and stabilized using custom-made ear bars in a stereotaxic frame (Narishige Corp., Tokyo, Japan). Body temperature was maintained with a heat mat at 37°C. All spinal surgery was done under a dissecting microscope (Leica Microsystems, Nuslock, Germany). Vertebral clamps on the T13 and L2 vertebral arches stabilized the vertebral column. A L1 laminectomy was performed to expose the L6-S1 spinal cord segments (Harrison et al., 2013). Small incisions were made in the dura and pia so recording pipettes could be lowered into the spinal cord. The surface of the spinal cord was kept moist with warmed (37oC) artificial cerebrospinal fluid (aCSF) during the experiment. The aCSF contained (in mM): 118 NaCl, 25 NaHCO3, 11 glucose, 2.5 KCl, 1 NaH2PO4, 1 MgCl2, and 2.5 CaCl2. The aCSF was bubbled with 95% O2-5% CO2 to maintain pH at 7.3. Animals were overdosed with pentobarbital (200mg/kg i.p.) when experiments were complete.
Electrophysiology Recording pipettes (4-7 MΩ) were pulled from thin-walled filamented borosilicate glass capillaries (1.5 mm o.d., 1.17 mm i.d., Harvard Apparatus, Edenbridge, UK) using a micropipette puller (PC-10, Narishige Corp., Tokyo, Japan). A K+ based internal solution containing (in mM): 135 KGluconate, 6 NaCl, 2 MgCl2, 10 HEPES, 0.1 EGTA, 2 NaATP, and 0.3 NaGTP (adjusted to pH 7.3 with KOH) was used in all experiments. A Multiclamp 700B amplifier (Molecular Devices, Sunnyvale, CA, USA) was used to record all signals. A Luigs & Neumann SM-6 micromanipulator (Ratingen, Germany) was used to guide pipettes into the spinal cord dorsal horn. Recording depth (in microns) was monitored by a digital readout on the micromanipulator. The pipette was positioned at an angle of 65°
8
(to horizontal) and was advanced to depth of 100 µm to ensure its tip was located in the dorsal horn gray matter (Farrell et al., 2016). Continuous application of positive pressure (~50 kPa) kept the pipette tip clear of debris (Graham et al., 2004a) as it moved through the white matter. Positive pressure was reduced, and the pipette was then advanced in 3 µm steps for a further 200 µm to search for SDH neurons. Our ‘searching’ for neurons was undertaken in voltage-clamp by continuously monitoring changes in pipette resistance to a voltage step (−20 mV, 10 ms). When a neuron was encountered, the positive pressure was released to allow formation of a tight seal (1-4 GΩ, holding potential -55 mV). The whole-cell recording configuration was then established by applying gentle suction (series resistance 0.4 mV).
Neurons were classified as responding to cutaneous brush and pinch based if membrane potential changed by more than 0.4 mV, as described above. Similarly, the type of cutaneous response (suprathreshold versus subthreshold, low threshold versus high threshold verses wide dynamic range) was also assessed. AP discharge frequency, during cutaneous brush and pinch application, were calculated by dividing the AP number by the duration of the stimulation period.
Statistical analysis
13
GraphPad Prism 6 (GraphPad Software Inc., La Jolia, CA, USA) and SPSS v21 (SPSS Inc., Chicago, IL, USA) were used for statistical analysis. Two-tailed Student’s t-tests were employed to determine if any differences existed between naïve and inflamed animals in: colon length to weight ratios, histological scores, AP discharge frequency during brush and pinch, the slope (gain) of frequency/current, adaptation ratio/current and attenuation ratio/current relationships. When data failed the Kolmogorov-Smirnov normality test comparisons were made using the non-parametric Mann-Whitney test. Differences in the proportions of neurons showing responses to colonic and cutaneous stimulation after colitis were determined using Pearson’s χ2 tests and binomial logistic regression, respectively. We used two way ANOVAs to compare the intrinsic, AP, and synaptic properties of CRD-R and -NR neurons in naïve and TNBS-treated mice. Finally, multinomial logistic regression was used to compare the proportions of AP discharge patterns and the responses to hyperpolarizing current injection before and after colitis in CRD-R and CRD-NR neurons. Statistical significance was set at p < 0.05. T-test values are reported as mean ± SEM. Two-way ANOVA values are reported as estimated marginal means ± standard error for both variables (naïve versus inflamed; CRD-R versus CRD-NR).
Only a small proportion of SDH neurons responded to CRD. This meant a limited sample size of CRD-R neurons was unavoidable for some comparisons. Accordingly, we used the effect size (ES) statistic to complement the more traditional statistical tests. This approach meant we could ensure that SDH neurons with CRD responses were randomly sampled. In addition, the use of effect sizes meant the number of animals used in our experiments was kept to a
14
minimum. Previous work from our group (Rank et al., 2015a; Rank et al., 2015b; Tadros et al., 2016) has shown that the ES statistic provides a reliable measure of the strength of biological phenomenon. For example, the magnitude of difference between means (Cohen’s d; d), or the odds of ‘success,’ of a particular outcome in a treatment group (i.e. TNBS-colitis) relative to the odds of ‘success’ in a control group (i.e. naïve; odds ratio; OR). ES is also less susceptible to the influence of sample size and allows detection of biologically relevant effects in smaller samples and when undertaking multiple comparisons. In the experiments described here we calculated ES for comparisons between CRD-R and CRD-NR neurons in naïve and inflamed animals. ES values were considered trivial (i.e. not biologically relevant) when d < 0.3; OR < 1.5, small when d = 0.3 – 0.49; OR = 1.5 – 2.49, moderate when d = 0.5 – 0.79; OR = 2.5 – 4.29, and large when d ≥ 0.8; OR ≥ 4.3.
15
Results Stable whole-cell patch-clamp recordings were obtained from 88 neurons in 45 naïve animals, and 67 neurons in 44 TNBS-challenged (hereafter called inflamed) animals. Data for naïve animals has been published previously in a manuscript describing an in vivo mouse preparation for examining the synaptic and intrinsic properties of neurons that receive inputs from the colorectum (Farrell et al., 2016). Data from that paper have been subject to additional analysis here for comparison with inflamed animals. One potential limitation of our study is that we made comparisons between naïve and TNBS treated animals and did not use vehicle-treated animals as controls. Previous work, however, has recently made similar comparisons (Mickle et al., 2010). In addition, others have shown that naïve, saline and vehicle treated animals provide similar outcomes compared to TNBS-treated animals (Birder et al., 2003).
The depth of each recorded neuron, as measured from the border of the SDH and the overlying white matter, were similar in naïve and inflamed animals (p = 0.2), ranging from 1 to 147 μm (48 ± 3.8 μm, n = 88) for naïve neurons, and 1 to 165 μm (56 ± 5.0 μm, n = 67) for inflamed neurons. As previously reported, these depths indicate recorded neurons were located primarily within the mouse SDH (lamina I-II) (Farrell et al., 2016).
Prior to applying any stimulation protocols, baseline spontaneous synaptic activity was assessed (from resting membrane potential) and then each neuron was subsequently classified as either CRD-responsive (CRD-R) or CRDnonresponsive (CRD-NR) based on the presence or absence of a response to 16
noxious distension of the colon. Neurons were then further categorized based on their responses to innocuous (brush) and noxious (pinch) cutaneous stimulation. In 78% of naïve neurons (69/88) and 93% of inflamed neurons (62/67) recordings were maintained long enough to additionally assess a neurons response to hyperpolarizing current injection and AP discharge properties.
Severity of colitis In order to evaluate the severity of colitis we used two approaches: 1) by measuring the colon weight/length ratio, and 2) using semi-quantitative histological scoring of colon sections (Marks et al., 2015). Inflamed animals had larger colon weight/length ratios compared to naïve animals (30.0 ± 1.0 versus 25.2 ± 0.7 mg/cm; n = 44 and n = 16; p = 0.001; large ES, d = 0.97). Since colon shortening and thickening are markers of colonic fibrosis and inflammation (Keely et al., 2014) this indicates that intra-rectal TNBS-challenge model successfully induced colitis in our ‘inflamed’ mice. Similarly, histological analysis of colon tissue samples showed an increased colitis score (inflammation, injury and colitis activity) in inflamed animals (6.5 ± 0.9 versus 1.5 ± 0.7; n = 44 and n = 16; p = < 0.0001; large ES, d = 1.1; Figure 1). In terms of relative level of disease, these histological scores indicate that our TNBS model induced a mild form of colitis. INSERT FIGURE 1 NEAR HERE
SDH neuron responses during noxious colorectal distension Neurons were classified as CRD-R or CRD-NR based on whether they received inputs from stretch-sensitive colonic afferents activated by noxious CRD (Figure
17
2). In inflamed animals, more neurons responded to CRD compared to those in naïve animals (28% 19/67 versus 11% 10/88, p = 0.007; moderate ES, OR = 3.1; Figure 2A; left). In a minority of these CRD neurons, colorectal distension resulted in the generation of APs (naïve 3/10, inflamed 4/19; Figure 2A, right, B). In neurons from inflamed animals, AP discharge occurred in bursts at distension onset and offset. In contrast neurons from naïve animals showed transient responses to CRD at distension onset only (Figure 2B). In addition, half the CRD responsive neurons with AP discharge from inflamed animals showed evidence of facilitation (Figure 2C). That is, they initially exhibited subthreshold responses to distension (i.e. EPSPs; Figure 2C, left), but with subsequent distensions generated APs (Figure 2C, right). This was never observed in neurons from naïve animals.
The majority of neurons in both naïve and inflamed animals exhibited subthreshold responses to CRD (naïve 7/10, inflamed 15/19; Figure 2A, right, D). Most subthreshold responses were depolarizing in nature (naïve 5/7, inflamed 13/15), however some hyperpolarizing responses were also observed (not shown). The proportion of neurons with AP discharge versus subthreshold responses to CRD was not different following inflammation (p = 0.6; Figure 2A, right). Despite this overall similarity, more detailed analysis showed some evoked EPSP properties differed in neurons from the two groups (Figure 2E). Inflammation increased EPSP peak amplitude (4.5 ± 1.6 versus 5.7 ± 0.5 mV; p = 0.22; moderate ES, d = 0.67), duration (4.2 ± 1.6 versus 6.4 ± 0.7 s; p = 0.18; moderate ES, d = 0.67) and area under curve (5.1 ± 3.0 versus 8.8 ± 1.7 mV.s; p = 0.27; moderate ES, d = 0.59). Together, these data suggest that mild colitis
18
increases afferent signaling to neurons in the SDH particularly in the form of subthreshold synaptic responses (i.e. evoked EPSPs versus APs).
INSERT FIGURE 2 NEAR HERE
SDH neuron responses during peripheral cutaneous stimulation We next examined the responses of SDH neurons to peripheral cutaneous stimulation (brush and/or pinch of the hind paw and/or tail). There was a significant main effect for inflammation on the proportion of neurons responding to cutaneous stimuli. In naïve and inflamed animals all, or most, CRD-R neurons had cutaneous receptive fields (CRD-R: naïve 80% [8/10] versus inflamed 100% [19/19]). However, many CRD-NR neurons also had cutaneous receptive fields, and this proportion was increased following inflammation (CRD-NR: naïve 49% [38/78] versus inflamed 73% [35/48]; p = 0.003; moderate ES, OR = 3.2; Figure 3A). In addition, a main effect was also observed for CRD responsiveness, with more CRD-R neurons responding to cutaneous stimulation compared to CRD-NR neurons (p = 0.007; large ES, OR = 7.9). Thus, mild colitis increased the number of SDH neurons that had a cutaneous receptive field regardless of whether they received stretch sensitive inputs from the colorectum.
Response to brush and pinch stimuli in mild colitis We next classified neurons according to their response to innocuous (brush) and noxious (pinch) cutaneous stimuli and examined whether this changed during mild colitis. Neurons were classified as low threshold (LT), high threshold (HT), or wide dynamic range (WDR) as outlined below (Figure 3B, C). Of the neurons
19
with cutaneous inputs (i.e. both CRD-R and CRD-NR), similar proportions responded only to pinch (naïve 41% [19/46] versus inflamed 36% [20/55]). We classed these neurons as high threshold. Few neurons responded to brush stimuli alone (naïve 18% [8/46] versus inflamed 6% [3/55]). We termed these low threshold neurons. Almost half the neurons responded to both brush and pinch (naïve 41% [19/46] versus inflamed 58% [32/55]). Such neurons were classified as wide dynamic range (Figure 3B, C). There was an overall trend towards a change in the proportions of these response types following inflammation (p = 0.09; Figure 3B). Based on effect size, inflammation reduced the proportions of high threshold (small ES, OR = 1.6) and low threshold (large ES, OR = 4.5) neurons when compared to wide dynamic range neurons (Figure 3B, C). Given that the average depth of recorded neurons was similar in all conditions (Table 2), it is unlikely that this difference is an artefact of recording from neurons in deeper spinal laminae, where WDR neurons are more prevalent.
Nature of responses to peripheral cutaneous stimuli during mild colitis There was substantial variation in the nature of the responses of lumbosacral SDH neurons to peripheral cutaneous stimulation. In many neurons brush or pinch elicited AP discharge (CRD-R: naïve 38% [3/8], inflamed 37% [7/19]; CRD-NR: naïve 63% [24/38], inflamed 63% [22/35]). Inflammation resulted in an increased frequency of AP discharge elicited by both brush (3.4 ± 1.4 versus 7.3 ± 1.7 Hz; n = 12 and n = 14; p = 0.027; moderate ES, d = 0.71) and pinch (3.9 ± 0.8 versus 14.5 ± 5.5 Hz; n = 22 and n = 28; p = 0.056; moderate ES, d = 0.51). Cutaneous stimulation resulted in subthreshold depolarization (i.e. EPSPs) or hyperpolarization (i.e. IPSCs) in the remaining neurons (CRD-R: naïve 62%
20
[5/8], inflamed 63% [12/19]; CRD-NR: naïve 37% [14/38], inflamed 37% [13/35]). Unlike subthreshold responses to CRD, we did not observe slow depolarizations in response to brush or pinch. Together, this suggests that inflammation does not change the proportions of neurons that displayed subthreshold versus suprathreshold responses to cutaneous stimulation (p = 1.0; trivial ES, OR = 0.98), but it did increase the magnitude of suprathreshold responses via increasing AP discharge frequency during brush and pinch. Interestingly, most CRD-R neurons also showed predominately subthreshold responses to cutaneous stimulation (62% and 63% of naïve and inflamed, respectively). This represented a main effect for CRD-responsiveness, with CRDR neurons having significantly larger proportion of subthreshold responses compared to CRD-NR (p = 0.02; moderate ES, OR = 2.89). Together, these data suggest that mild colitis had a ‘sensitizing effect’ on neurons in the SDH regarding their response to peripheral cutaneous stimulation: more SDH neurons were sensitive to cutaneous stimulation, the number of WDR neurons increased compared to single modality responders, and the frequency of AP discharge during both brush and pinch increased. Moreover, while inflammation did not affect the intensity of responses to cutaneous stimulation (i.e. subthreshold versus suprathreshold), there were a significantly higher proportion of subthreshold responses in neurons with convergent colonic inputs.
INSERT FIGURE 3 NEAR HERE
Influence of mild colitis on spontaneous excitatory synaptic activity
21
In order to assess the influence of mild colitis on excitatory synaptic activity (sEPSPs), we recorded baseline synaptic activity in SDH neurons for ~60 s before delivering colonic and cutaneous stimulation protocols. sEPSPs could be clearly resolved as transient depolarizations in almost all recorded neurons (CRD-R: naïve 10/10, inflamed 19/19; CRD-NR: naïve 68/78, inflamed 44/48; Figure 4A, B). Mild colitis caused several changes in the properties of sEPSPs (Figure 4C). sEPSPs half width increased during inflammation (9.9 ± 0.5 versus 11.2 ± 0.5 ms; Two-way ANOVA p = 0.03; small ES, d = 0.37). There was also a trending main effect for increased sEPSP frequency during inflammation (10.6 ± 1.1 versus 13.1 ± 1.1 Hz; p = 0.07; small ES, d = 0.31; Figure 4C). In contrast, inflammation did not affect sEPSP rise time (2.4 ± 0.1 versus 2.6 ± 0.1 ms; p = 0.1; trivial ES, d = 0.25) or amplitude (1.5 ± 0.2 versus 1.4 ± 0.2 mV; p = 0.5; trivial ES, d = 0.11). Together this analysis suggests inflammation increases the level of excitatory synaptic drive to neurons within the SDH.
Two-way ANOVAs also revealed significant main effects for CRD responsiveness, with a longer half width observed for CRD-NR neurons (9.7 ± 0.7 versus 11.3 ± 0.3 ms; p = 0.04; small ES, d = 0.35; Figure 4C). sEPSP rise time (p = 0.09; trivial ES, d = 0.29), amplitude (p = 0.12; trivial ES, d = 0.26) and frequency (p = 0.53; trivial ES, d = 0.11; Figure 4C) were similar in CRD-R and CRD-NR neurons. Since these comparisons represent the estimated marginal means ± standard error, a summary of sEPSP raw means is provided in Table 1.
INSERT FIGURE 4 NEAR HERE INSERT TABLE 1 NEAR HERE
22
Intrinsic properties of SDH neurons Several intrinsic membrane properties of SDH neurons were measured from depolarizing and hyperpolarizing current step responses, and these are summarized in Table 2. Two-way ANOVAs revealed significant main effects for inflammation, with RMPs being more hyperpolarized in SDH neurons during colonic inflammation (−58.5 ± 1.2 versus −63.3 ± 1.1 mV; p = 0.001; moderate ES, d = 0.58). Similar main effects were also observed for CRD responsiveness, as RMP was also more hyperpolarized in CRD-R neurons compared to CRD-NR counterparts (−63.6 ± 1.6 versus −58.2 ± 0.8 mV; p = 0.004; moderate ES, d = 0.5). Additionally, a significant interaction was found between inflammation and CRDR neurons, with the AP threshold of only CRD-R neurons being more hyperpolarized during inflammation (CRD-R: −30.1 ± 2.7 versus −38.5 ± 2.0 mV; p = 0.03; moderate ES, d = 0.63). Taken as a whole, these results indicate that CRD-R neurons are less excitable than CRD-NR neurons, and that inflammation reduces this excitability further by hyperpolarizing RMP.
INSERT TABLE 2 NEAR HERE
SDH neuron responses to current injection Response to hyperpolarizing current injections The responses of lumbosacral SDH neurons to hyperpolarizing current steps (800 ms duration, −20 pA increments) were assessed in a majority of SDH neurons (78% naïve and 93% inflamed). Five main response types were observed at either the onset of, or after release from hyperpolarization (Figure
23
5A). In a majority of neurons, the membrane passively hyperpolarized at current step onset, and returned to RMP after release from hyperpolarization (CRD-R: naïve 67% [6/9], inflamed 35% [6/17]; CRD-NR: naïve 67% [38/57], inflamed 48% [20/42]). In other neurons, the release from hyperpolarization resulted in rebound depolarization with AP discharge (CRD-R: naïve 22% [2/9], inflamed 18% [3/17]; CRD-NR: naïve 19% [11/57], inflamed 7% [3/42]), or rebound depolarization without AP discharge (CRD-R: naïve 11% [1/9], inflamed 18% [3/17]; CRD-NR: naïve 9% [5/57], inflamed 19% [8/42]). In a proportion of SDH neurons, a voltage ‘sag’ was observed at the onset of hyperpolarization with membrane potential passively returning to resting levels (CRD-R: naïve 0% [0/9], inflamed 29% [5/17]; CRD-NR: naïve 5% [3/57], inflamed 26% [11/42]). Finally, a few neurons (CRD-NR: naïve 2/57, inflamed 2/42) exhibited a prolonged hyperpolarizing response after release from hyperpolarization (not shown). These neurons were excluded from statistical analysis.
Following inflammation, the proportion of neurons in the four response categories was altered (p =
S0306-4522(17)30595-X http://dx.doi.org/10.1016/j.neuroscience.2017.08.029 NSC 17978
To appear in:
Neuroscience
Received Date: Revised Date: Accepted Date:
17 January 2017 19 July 2017 14 August 2017
Please cite this article as: K.E. Farrell, S. Keely, M.M. Walker, A.M. Brichta, B.A. Graham, R.J. Callister, Altered intrinsic and synaptic properties of lumbosacral dorsal horn neurons in a mouse model of colitis, Neuroscience (2017), doi: http://dx.doi.org/10.1016/j.neuroscience.2017.08.029
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Altered intrinsic and synaptic properties of lumbosacral dorsal horn neurons in a mouse model of colitis
Kristen E Farrell1,2, Simon Keely1,2, Marjorie M Walker2,3, Alan M Brichta1,2, Brett A Graham1,2 and Robert J Callister1,2 1School
of Biomedical Sciences & Pharmacy, University of Newcastle, Callaghan, NSW, 2308, Australia
2Hunter
Medical Research Institute (HMRI), Rankin Park, NSW, 2305, Australia
3School
of Public Health & Medicine, University of Newcastle, Callaghan, NSW, 2308, Australia
Running title: Colitis alters the functional properties of dorsal horn neurons Key words: in vivo, spinal cord, visceral inflammation Corresponding author Professor Robert J Callister School of Biomedical Sciences and Pharmacy The University of Newcastle Callaghan, NSW, 2308, Australia Telephone: +61 2 49217808 Email: [email protected]
1
Abstract Visceral pain in inflammatory and functional gastrointestinal conditions is a major clinical problem. While the exact mechanisms underlying the development of pain, during and after visceral inflammation, are unknown clinical and preclinical evidence suggests plasticity within the spinal cord dorsal horn is a contributing factor. Here we use an in vivo preparation and patch clamp electrophysiology to test whether the synaptic and intrinsic properties of superficial dorsal horn (SDH) neurons are altered 5 days after the induction of mild colitis in adult male mice (i.e. during acute inflammation of the colon). Whole cell recordings were made from lumbosacral (L6-S1) superficial dorsal horn neurons (SDH), in animals under isoflurane anaesthesia. Noxious colorectal distension (CRD) was used to identify SDH neurons with colonic inputs, while stimulation of the hind paw and tail was employed to assess convergent cutaneous input. Following inflammation, a significantly increased proportion of SDH neurons received both colonic and cutaneous inputs, compared to neurons in naïve animals. In addition, the nature and magnitude of responses to CRD and cutaneous stimulation differed in inflamed animals, as was spontaneous excitatory synaptic drive. Conversely, several measures of intrinsic excitability were altered in a manner that would decrease SDH network excitability following colitis. We propose that during inflammation, sensitization of colonic afferents results in increased signaling to the SDH. This is accompanied by plasticity in SDH neurons whereby their intrinsic properties are changed to compensate for altered afferent activity.
2
Abbreviations list AHP, afterhyperpolarization; ANOVA, analysis of variance; AP, action potential; CNS, central nervous system; CRD, colorectal distension; CRD-NR, colorectal distension-nonresponsive; CRD-R, colorectal distension-responsive; DF, delayed firing; EPSP, excitatory post synaptic potential; ES, effect size; HT, high threshold; IB, initial bursting; IBD, inflammatory bowel disease; IBS, irritable bowel syndrome; IPSP, inhibitory post synaptic potential; LT, low threshold; OR, odds ratio; Rin, input resistance; SDH, superficial dorsal horn; SS, single spiking; TF, tonic firing; TNBS, trinitrobenzenesulfonic acid; U, unclassified; WDR, wide dynamic range
3
Introduction Chronic pain and visceral hypersensitivity are common and debilitating symptoms of several disorders of the gastrointestinal tract, including irritable bowel syndrome (IBS) and inflammatory bowel disease (IBD). IBS is a functional gastrointestinal disorder (FGID) characterized by chronic, recurrent pain and discomfort with low-grade inflammation of the colon mucosa (Verne et al., 2001; Verne et al., 2003; Barbara et al., 2004), whereas IBD describes several chronic inflammatory organic diseases of the gastrointestinal tract. While pain most commonly occurs during disease flare-ups in IBD, 30-50% of patients report IBSlike persistent pain in the absence of active disease (Minderhoud et al., 2004; Farrokhyar et al., 2006; Siegel & MacDermott, 2009). In both conditions, generalized “referred pain” or somatic hyperalgesia is also widely reported by patients (Bernstein et al., 1996; Verne et al., 2001; Verne et al., 2003; Minderhoud et al., 2004). This mixing of visceral and somatic sensory symptoms, or viscerosomatic convergence, is thought to be the result of plasticity within the central nervous system (CNS), particularly in the spinal cord dorsal horn where these sensory pathways can overlap (Ruch, 1961; Farrell et al., 2014b). The precise mechanisms underlying the development and maintenance of chronic visceral and somatic hypersensitivity remain largely unknown (Price et al., 2006; Farrell et al., 2014a; Farrell et al., 2014b).
Our previous review of the literature on CNS plasticity in animal models of visceral hypersensitivity pointed to a crucial role for the dorsal horn (Farrell et al., 2014b). For example, experimental rodent models of colonic inflammation can induce visceral and somatic hyperalgesia (Laird et al., 2001b; Lamb et al., 4
2006). Inflammation also increases neural activation (cFos and pERK), and the release of ‘pain’ neuropeptides (Substance P and CGRP) within the spinal cord dorsal horn (Lu & Westlund, 2001; Sun & Luo, 2004; Traub et al., 2008; Harrington et al., 2012). In addition, functional studies using in vivo extracellular recording have observed that following inflammation, dorsal horn neurons exhibit increased spontaneous action potential (AP) discharge (Al-Chaer et al., 2000; Ness & Gebhart, 2001; Qin et al., 2005) and increased AP discharge in response to distension of the colon (Laird et al., 2001a; Wang et al., 2005b). Together, these studies indicate that visceral inflammation alters the activity or output of dorsal horn neurons, however, the changes to the major determinants of neuron output (i.e. their intrinsic and synaptic properties) that occur during visceral inflammation have yet to be characterized during both acute and chronic models of disease.
Dorsal horn interneurons are involved in the complex processing and integration of inputs from peripheral structures like viscera and skin, higher brain centers, and other local circuit interneurons (Graham et al., 2007). It follows that small changes in their output have the capacity to significantly alter sensory signaling to higher centers and ultimately perception. Until recently, it has proved difficult to study the intrinsic and synaptic properties of the small dorsal horn interneurons that receive visceral inputs. We have developed a mouse preparation that allows high resolution in vivo patch clamp recording from dorsal horn neurons in the spinal segments that receive colonic input (L6-S1) (Farrell et al., 2016). Importantly, this preparation enables detailed analysis of the intrinsic and synaptic properties of neurons that respond to mechanical
5
distension of the colorectum. In this study using our in vivo mouse preparation we hypothesized that the intrinsic and synaptic properties of superficial dorsal horn (SDH) neurons are altered during acute colonic inflammation.
Our data demonstrate that during mild colonic inflammation, the synaptic properties and responses of SDH neurons to colonic and cutaneous stimulation render this population more excitable. Conversely, several changes in the intrinsic properties including a hyperpolarization of membrane potential and an increase in the excitability of presumed inhibitory interneurons are consistent with a decrease in overall SDH excitability. We propose that during inflammation, sensitization of colonic afferents results in increased signaling to the SDH. This is accompanied by plasticity in SDH neurons, which have altered their intrinsic properties in an attempt to compensate for this increased afferent activity.
6
Methods All surgical and experimental procedures were approved by the University of Newcastle’s Animal Care and Ethics Committee (protocol # A-2012-223). Male mice (C57BL/6, 6-8 weeks old, 15.7-27.4 g) were used for all experiments. The adult mouse preparation we used for spinal cord in vivo patch clamp recording and the data collection protocols have been described previously (Graham et al., 2004a, b; Jobling et al., 2010; Farrell et al., 2016). They are briefly summarized here.
Animal model of colitis Colonic inflammation was induced using the modified chemically-induced trinitrobenzenesulfonic acid (TNBS) model of colitis (Wirtz et al., 2007; Marks et al., 2015). Mice were presensitized by epicutaneous application of 1% TNBS (Sigma Chemicals, St. Louis, MO, USA) in acetone/olive oil solution (4:1). After 7 days, mice were anaesthetized with isoflurane (2%) and received a single dose of 0.2ml TNBS (2.5% in 50% ethanol) delivered into the colorectum, 4cm from the anus, via an anal catheter. Mice were suspended upside down for one minute to ensure TNBS remained in the colorectum. Animals were observed daily for changes in body weight, physical appearance and behavior. Electrophysiological experiments were conducted 5 days after the induction of colitis.
Surgery for in vivo spinal cord recording Mice were anaesthetized with isoflurane (2-3%, 2L/min O2 induction; 1-2% maintenance) delivered via a nose cone. When deep anaesthesia was confirmed via the absence of hind limb and corneal reflexes, the animals were head-fixed 7
and stabilized using custom-made ear bars in a stereotaxic frame (Narishige Corp., Tokyo, Japan). Body temperature was maintained with a heat mat at 37°C. All spinal surgery was done under a dissecting microscope (Leica Microsystems, Nuslock, Germany). Vertebral clamps on the T13 and L2 vertebral arches stabilized the vertebral column. A L1 laminectomy was performed to expose the L6-S1 spinal cord segments (Harrison et al., 2013). Small incisions were made in the dura and pia so recording pipettes could be lowered into the spinal cord. The surface of the spinal cord was kept moist with warmed (37oC) artificial cerebrospinal fluid (aCSF) during the experiment. The aCSF contained (in mM): 118 NaCl, 25 NaHCO3, 11 glucose, 2.5 KCl, 1 NaH2PO4, 1 MgCl2, and 2.5 CaCl2. The aCSF was bubbled with 95% O2-5% CO2 to maintain pH at 7.3. Animals were overdosed with pentobarbital (200mg/kg i.p.) when experiments were complete.
Electrophysiology Recording pipettes (4-7 MΩ) were pulled from thin-walled filamented borosilicate glass capillaries (1.5 mm o.d., 1.17 mm i.d., Harvard Apparatus, Edenbridge, UK) using a micropipette puller (PC-10, Narishige Corp., Tokyo, Japan). A K+ based internal solution containing (in mM): 135 KGluconate, 6 NaCl, 2 MgCl2, 10 HEPES, 0.1 EGTA, 2 NaATP, and 0.3 NaGTP (adjusted to pH 7.3 with KOH) was used in all experiments. A Multiclamp 700B amplifier (Molecular Devices, Sunnyvale, CA, USA) was used to record all signals. A Luigs & Neumann SM-6 micromanipulator (Ratingen, Germany) was used to guide pipettes into the spinal cord dorsal horn. Recording depth (in microns) was monitored by a digital readout on the micromanipulator. The pipette was positioned at an angle of 65°
8
(to horizontal) and was advanced to depth of 100 µm to ensure its tip was located in the dorsal horn gray matter (Farrell et al., 2016). Continuous application of positive pressure (~50 kPa) kept the pipette tip clear of debris (Graham et al., 2004a) as it moved through the white matter. Positive pressure was reduced, and the pipette was then advanced in 3 µm steps for a further 200 µm to search for SDH neurons. Our ‘searching’ for neurons was undertaken in voltage-clamp by continuously monitoring changes in pipette resistance to a voltage step (−20 mV, 10 ms). When a neuron was encountered, the positive pressure was released to allow formation of a tight seal (1-4 GΩ, holding potential -55 mV). The whole-cell recording configuration was then established by applying gentle suction (series resistance 0.4 mV).
Neurons were classified as responding to cutaneous brush and pinch based if membrane potential changed by more than 0.4 mV, as described above. Similarly, the type of cutaneous response (suprathreshold versus subthreshold, low threshold versus high threshold verses wide dynamic range) was also assessed. AP discharge frequency, during cutaneous brush and pinch application, were calculated by dividing the AP number by the duration of the stimulation period.
Statistical analysis
13
GraphPad Prism 6 (GraphPad Software Inc., La Jolia, CA, USA) and SPSS v21 (SPSS Inc., Chicago, IL, USA) were used for statistical analysis. Two-tailed Student’s t-tests were employed to determine if any differences existed between naïve and inflamed animals in: colon length to weight ratios, histological scores, AP discharge frequency during brush and pinch, the slope (gain) of frequency/current, adaptation ratio/current and attenuation ratio/current relationships. When data failed the Kolmogorov-Smirnov normality test comparisons were made using the non-parametric Mann-Whitney test. Differences in the proportions of neurons showing responses to colonic and cutaneous stimulation after colitis were determined using Pearson’s χ2 tests and binomial logistic regression, respectively. We used two way ANOVAs to compare the intrinsic, AP, and synaptic properties of CRD-R and -NR neurons in naïve and TNBS-treated mice. Finally, multinomial logistic regression was used to compare the proportions of AP discharge patterns and the responses to hyperpolarizing current injection before and after colitis in CRD-R and CRD-NR neurons. Statistical significance was set at p < 0.05. T-test values are reported as mean ± SEM. Two-way ANOVA values are reported as estimated marginal means ± standard error for both variables (naïve versus inflamed; CRD-R versus CRD-NR).
Only a small proportion of SDH neurons responded to CRD. This meant a limited sample size of CRD-R neurons was unavoidable for some comparisons. Accordingly, we used the effect size (ES) statistic to complement the more traditional statistical tests. This approach meant we could ensure that SDH neurons with CRD responses were randomly sampled. In addition, the use of effect sizes meant the number of animals used in our experiments was kept to a
14
minimum. Previous work from our group (Rank et al., 2015a; Rank et al., 2015b; Tadros et al., 2016) has shown that the ES statistic provides a reliable measure of the strength of biological phenomenon. For example, the magnitude of difference between means (Cohen’s d; d), or the odds of ‘success,’ of a particular outcome in a treatment group (i.e. TNBS-colitis) relative to the odds of ‘success’ in a control group (i.e. naïve; odds ratio; OR). ES is also less susceptible to the influence of sample size and allows detection of biologically relevant effects in smaller samples and when undertaking multiple comparisons. In the experiments described here we calculated ES for comparisons between CRD-R and CRD-NR neurons in naïve and inflamed animals. ES values were considered trivial (i.e. not biologically relevant) when d < 0.3; OR < 1.5, small when d = 0.3 – 0.49; OR = 1.5 – 2.49, moderate when d = 0.5 – 0.79; OR = 2.5 – 4.29, and large when d ≥ 0.8; OR ≥ 4.3.
15
Results Stable whole-cell patch-clamp recordings were obtained from 88 neurons in 45 naïve animals, and 67 neurons in 44 TNBS-challenged (hereafter called inflamed) animals. Data for naïve animals has been published previously in a manuscript describing an in vivo mouse preparation for examining the synaptic and intrinsic properties of neurons that receive inputs from the colorectum (Farrell et al., 2016). Data from that paper have been subject to additional analysis here for comparison with inflamed animals. One potential limitation of our study is that we made comparisons between naïve and TNBS treated animals and did not use vehicle-treated animals as controls. Previous work, however, has recently made similar comparisons (Mickle et al., 2010). In addition, others have shown that naïve, saline and vehicle treated animals provide similar outcomes compared to TNBS-treated animals (Birder et al., 2003).
The depth of each recorded neuron, as measured from the border of the SDH and the overlying white matter, were similar in naïve and inflamed animals (p = 0.2), ranging from 1 to 147 μm (48 ± 3.8 μm, n = 88) for naïve neurons, and 1 to 165 μm (56 ± 5.0 μm, n = 67) for inflamed neurons. As previously reported, these depths indicate recorded neurons were located primarily within the mouse SDH (lamina I-II) (Farrell et al., 2016).
Prior to applying any stimulation protocols, baseline spontaneous synaptic activity was assessed (from resting membrane potential) and then each neuron was subsequently classified as either CRD-responsive (CRD-R) or CRDnonresponsive (CRD-NR) based on the presence or absence of a response to 16
noxious distension of the colon. Neurons were then further categorized based on their responses to innocuous (brush) and noxious (pinch) cutaneous stimulation. In 78% of naïve neurons (69/88) and 93% of inflamed neurons (62/67) recordings were maintained long enough to additionally assess a neurons response to hyperpolarizing current injection and AP discharge properties.
Severity of colitis In order to evaluate the severity of colitis we used two approaches: 1) by measuring the colon weight/length ratio, and 2) using semi-quantitative histological scoring of colon sections (Marks et al., 2015). Inflamed animals had larger colon weight/length ratios compared to naïve animals (30.0 ± 1.0 versus 25.2 ± 0.7 mg/cm; n = 44 and n = 16; p = 0.001; large ES, d = 0.97). Since colon shortening and thickening are markers of colonic fibrosis and inflammation (Keely et al., 2014) this indicates that intra-rectal TNBS-challenge model successfully induced colitis in our ‘inflamed’ mice. Similarly, histological analysis of colon tissue samples showed an increased colitis score (inflammation, injury and colitis activity) in inflamed animals (6.5 ± 0.9 versus 1.5 ± 0.7; n = 44 and n = 16; p = < 0.0001; large ES, d = 1.1; Figure 1). In terms of relative level of disease, these histological scores indicate that our TNBS model induced a mild form of colitis. INSERT FIGURE 1 NEAR HERE
SDH neuron responses during noxious colorectal distension Neurons were classified as CRD-R or CRD-NR based on whether they received inputs from stretch-sensitive colonic afferents activated by noxious CRD (Figure
17
2). In inflamed animals, more neurons responded to CRD compared to those in naïve animals (28% 19/67 versus 11% 10/88, p = 0.007; moderate ES, OR = 3.1; Figure 2A; left). In a minority of these CRD neurons, colorectal distension resulted in the generation of APs (naïve 3/10, inflamed 4/19; Figure 2A, right, B). In neurons from inflamed animals, AP discharge occurred in bursts at distension onset and offset. In contrast neurons from naïve animals showed transient responses to CRD at distension onset only (Figure 2B). In addition, half the CRD responsive neurons with AP discharge from inflamed animals showed evidence of facilitation (Figure 2C). That is, they initially exhibited subthreshold responses to distension (i.e. EPSPs; Figure 2C, left), but with subsequent distensions generated APs (Figure 2C, right). This was never observed in neurons from naïve animals.
The majority of neurons in both naïve and inflamed animals exhibited subthreshold responses to CRD (naïve 7/10, inflamed 15/19; Figure 2A, right, D). Most subthreshold responses were depolarizing in nature (naïve 5/7, inflamed 13/15), however some hyperpolarizing responses were also observed (not shown). The proportion of neurons with AP discharge versus subthreshold responses to CRD was not different following inflammation (p = 0.6; Figure 2A, right). Despite this overall similarity, more detailed analysis showed some evoked EPSP properties differed in neurons from the two groups (Figure 2E). Inflammation increased EPSP peak amplitude (4.5 ± 1.6 versus 5.7 ± 0.5 mV; p = 0.22; moderate ES, d = 0.67), duration (4.2 ± 1.6 versus 6.4 ± 0.7 s; p = 0.18; moderate ES, d = 0.67) and area under curve (5.1 ± 3.0 versus 8.8 ± 1.7 mV.s; p = 0.27; moderate ES, d = 0.59). Together, these data suggest that mild colitis
18
increases afferent signaling to neurons in the SDH particularly in the form of subthreshold synaptic responses (i.e. evoked EPSPs versus APs).
INSERT FIGURE 2 NEAR HERE
SDH neuron responses during peripheral cutaneous stimulation We next examined the responses of SDH neurons to peripheral cutaneous stimulation (brush and/or pinch of the hind paw and/or tail). There was a significant main effect for inflammation on the proportion of neurons responding to cutaneous stimuli. In naïve and inflamed animals all, or most, CRD-R neurons had cutaneous receptive fields (CRD-R: naïve 80% [8/10] versus inflamed 100% [19/19]). However, many CRD-NR neurons also had cutaneous receptive fields, and this proportion was increased following inflammation (CRD-NR: naïve 49% [38/78] versus inflamed 73% [35/48]; p = 0.003; moderate ES, OR = 3.2; Figure 3A). In addition, a main effect was also observed for CRD responsiveness, with more CRD-R neurons responding to cutaneous stimulation compared to CRD-NR neurons (p = 0.007; large ES, OR = 7.9). Thus, mild colitis increased the number of SDH neurons that had a cutaneous receptive field regardless of whether they received stretch sensitive inputs from the colorectum.
Response to brush and pinch stimuli in mild colitis We next classified neurons according to their response to innocuous (brush) and noxious (pinch) cutaneous stimuli and examined whether this changed during mild colitis. Neurons were classified as low threshold (LT), high threshold (HT), or wide dynamic range (WDR) as outlined below (Figure 3B, C). Of the neurons
19
with cutaneous inputs (i.e. both CRD-R and CRD-NR), similar proportions responded only to pinch (naïve 41% [19/46] versus inflamed 36% [20/55]). We classed these neurons as high threshold. Few neurons responded to brush stimuli alone (naïve 18% [8/46] versus inflamed 6% [3/55]). We termed these low threshold neurons. Almost half the neurons responded to both brush and pinch (naïve 41% [19/46] versus inflamed 58% [32/55]). Such neurons were classified as wide dynamic range (Figure 3B, C). There was an overall trend towards a change in the proportions of these response types following inflammation (p = 0.09; Figure 3B). Based on effect size, inflammation reduced the proportions of high threshold (small ES, OR = 1.6) and low threshold (large ES, OR = 4.5) neurons when compared to wide dynamic range neurons (Figure 3B, C). Given that the average depth of recorded neurons was similar in all conditions (Table 2), it is unlikely that this difference is an artefact of recording from neurons in deeper spinal laminae, where WDR neurons are more prevalent.
Nature of responses to peripheral cutaneous stimuli during mild colitis There was substantial variation in the nature of the responses of lumbosacral SDH neurons to peripheral cutaneous stimulation. In many neurons brush or pinch elicited AP discharge (CRD-R: naïve 38% [3/8], inflamed 37% [7/19]; CRD-NR: naïve 63% [24/38], inflamed 63% [22/35]). Inflammation resulted in an increased frequency of AP discharge elicited by both brush (3.4 ± 1.4 versus 7.3 ± 1.7 Hz; n = 12 and n = 14; p = 0.027; moderate ES, d = 0.71) and pinch (3.9 ± 0.8 versus 14.5 ± 5.5 Hz; n = 22 and n = 28; p = 0.056; moderate ES, d = 0.51). Cutaneous stimulation resulted in subthreshold depolarization (i.e. EPSPs) or hyperpolarization (i.e. IPSCs) in the remaining neurons (CRD-R: naïve 62%
20
[5/8], inflamed 63% [12/19]; CRD-NR: naïve 37% [14/38], inflamed 37% [13/35]). Unlike subthreshold responses to CRD, we did not observe slow depolarizations in response to brush or pinch. Together, this suggests that inflammation does not change the proportions of neurons that displayed subthreshold versus suprathreshold responses to cutaneous stimulation (p = 1.0; trivial ES, OR = 0.98), but it did increase the magnitude of suprathreshold responses via increasing AP discharge frequency during brush and pinch. Interestingly, most CRD-R neurons also showed predominately subthreshold responses to cutaneous stimulation (62% and 63% of naïve and inflamed, respectively). This represented a main effect for CRD-responsiveness, with CRDR neurons having significantly larger proportion of subthreshold responses compared to CRD-NR (p = 0.02; moderate ES, OR = 2.89). Together, these data suggest that mild colitis had a ‘sensitizing effect’ on neurons in the SDH regarding their response to peripheral cutaneous stimulation: more SDH neurons were sensitive to cutaneous stimulation, the number of WDR neurons increased compared to single modality responders, and the frequency of AP discharge during both brush and pinch increased. Moreover, while inflammation did not affect the intensity of responses to cutaneous stimulation (i.e. subthreshold versus suprathreshold), there were a significantly higher proportion of subthreshold responses in neurons with convergent colonic inputs.
INSERT FIGURE 3 NEAR HERE
Influence of mild colitis on spontaneous excitatory synaptic activity
21
In order to assess the influence of mild colitis on excitatory synaptic activity (sEPSPs), we recorded baseline synaptic activity in SDH neurons for ~60 s before delivering colonic and cutaneous stimulation protocols. sEPSPs could be clearly resolved as transient depolarizations in almost all recorded neurons (CRD-R: naïve 10/10, inflamed 19/19; CRD-NR: naïve 68/78, inflamed 44/48; Figure 4A, B). Mild colitis caused several changes in the properties of sEPSPs (Figure 4C). sEPSPs half width increased during inflammation (9.9 ± 0.5 versus 11.2 ± 0.5 ms; Two-way ANOVA p = 0.03; small ES, d = 0.37). There was also a trending main effect for increased sEPSP frequency during inflammation (10.6 ± 1.1 versus 13.1 ± 1.1 Hz; p = 0.07; small ES, d = 0.31; Figure 4C). In contrast, inflammation did not affect sEPSP rise time (2.4 ± 0.1 versus 2.6 ± 0.1 ms; p = 0.1; trivial ES, d = 0.25) or amplitude (1.5 ± 0.2 versus 1.4 ± 0.2 mV; p = 0.5; trivial ES, d = 0.11). Together this analysis suggests inflammation increases the level of excitatory synaptic drive to neurons within the SDH.
Two-way ANOVAs also revealed significant main effects for CRD responsiveness, with a longer half width observed for CRD-NR neurons (9.7 ± 0.7 versus 11.3 ± 0.3 ms; p = 0.04; small ES, d = 0.35; Figure 4C). sEPSP rise time (p = 0.09; trivial ES, d = 0.29), amplitude (p = 0.12; trivial ES, d = 0.26) and frequency (p = 0.53; trivial ES, d = 0.11; Figure 4C) were similar in CRD-R and CRD-NR neurons. Since these comparisons represent the estimated marginal means ± standard error, a summary of sEPSP raw means is provided in Table 1.
INSERT FIGURE 4 NEAR HERE INSERT TABLE 1 NEAR HERE
22
Intrinsic properties of SDH neurons Several intrinsic membrane properties of SDH neurons were measured from depolarizing and hyperpolarizing current step responses, and these are summarized in Table 2. Two-way ANOVAs revealed significant main effects for inflammation, with RMPs being more hyperpolarized in SDH neurons during colonic inflammation (−58.5 ± 1.2 versus −63.3 ± 1.1 mV; p = 0.001; moderate ES, d = 0.58). Similar main effects were also observed for CRD responsiveness, as RMP was also more hyperpolarized in CRD-R neurons compared to CRD-NR counterparts (−63.6 ± 1.6 versus −58.2 ± 0.8 mV; p = 0.004; moderate ES, d = 0.5). Additionally, a significant interaction was found between inflammation and CRDR neurons, with the AP threshold of only CRD-R neurons being more hyperpolarized during inflammation (CRD-R: −30.1 ± 2.7 versus −38.5 ± 2.0 mV; p = 0.03; moderate ES, d = 0.63). Taken as a whole, these results indicate that CRD-R neurons are less excitable than CRD-NR neurons, and that inflammation reduces this excitability further by hyperpolarizing RMP.
INSERT TABLE 2 NEAR HERE
SDH neuron responses to current injection Response to hyperpolarizing current injections The responses of lumbosacral SDH neurons to hyperpolarizing current steps (800 ms duration, −20 pA increments) were assessed in a majority of SDH neurons (78% naïve and 93% inflamed). Five main response types were observed at either the onset of, or after release from hyperpolarization (Figure
23
5A). In a majority of neurons, the membrane passively hyperpolarized at current step onset, and returned to RMP after release from hyperpolarization (CRD-R: naïve 67% [6/9], inflamed 35% [6/17]; CRD-NR: naïve 67% [38/57], inflamed 48% [20/42]). In other neurons, the release from hyperpolarization resulted in rebound depolarization with AP discharge (CRD-R: naïve 22% [2/9], inflamed 18% [3/17]; CRD-NR: naïve 19% [11/57], inflamed 7% [3/42]), or rebound depolarization without AP discharge (CRD-R: naïve 11% [1/9], inflamed 18% [3/17]; CRD-NR: naïve 9% [5/57], inflamed 19% [8/42]). In a proportion of SDH neurons, a voltage ‘sag’ was observed at the onset of hyperpolarization with membrane potential passively returning to resting levels (CRD-R: naïve 0% [0/9], inflamed 29% [5/17]; CRD-NR: naïve 5% [3/57], inflamed 26% [11/42]). Finally, a few neurons (CRD-NR: naïve 2/57, inflamed 2/42) exhibited a prolonged hyperpolarizing response after release from hyperpolarization (not shown). These neurons were excluded from statistical analysis.
Following inflammation, the proportion of neurons in the four response categories was altered (p =
