Reasons for performing study: The effects of prostaglandins and nonsteroidal anti-inflammatory drugs (NSAIDs) on repair of equine intestinal mucosa are important since most horses with gastrointestinal diseases are routinely treated with NSAIDs, such as flunixin meglumine (FM), and these drugs can be toxic to equine gastrointestinal mucosa.
Hypothesis: Flunixin meglumine would not affect recovery of equine colonic mucosa in vitro, 18 h after a reversible ischaemic injury.
Methods: In 14 anaesthetised horses, a segment of pelvic flexure was subjected to 2 h of ischaemia and the horses were allowed to recover for 18 h. Seven horses received normal saline and 7 received FM, 1.1 mg/kg bwt i.v., at the end of ischaemia and 12 h later. Colonic mucosa was harvested during a second anaesthesia, 18 h after recovery from ischaemia and then horses were subjected to euthanasia. Transepithelial electrical resistance (TER) and transepithelial flux of tritiated mannitol were used to measure mucosal permeability during 4 h of incubation in Ussing chambers, with the following in vitro treatments: 1) no addition, 2) FM 14 µmol/l as powder, 3) FM 14 µmol/l in injectable form and 4) diluent for injectable FM. Histomorphological changes were assessed by light microscopy.
Results: There were no significant differences in any of the measurements between saline and FM treated horses. The mucosal height of the ischaemic FM tissues incubated in diluent was significantly decreased compared to the nonischaemic tissues.
Conclusions: Flunixin meglumine did not adversely affect barrier integrity in ischaemic equine colonic mucosa.
Large colon volvulus is a common cause of surgical colic in horses and has a high mortality (Ellis et al. 2008). Even resection does not remove all ischaemic colon and recovery of the remaining mucosa can determine the severity of post operative endotoxaemia and outcome (Ellis et al. 2008). Because most horses with gastrointestinal diseases are routinely treated with NSAIDs, such as flunixin meglumine (FM) and these drugs can be toxic to normal equine gastrointestinal mucosa (MacAllister et al. 1993), the effects of NSAIDs on repair of equine intestinal mucosa are important (Tomlinson and Blikslager 2004). In studies that combined in vivo and in vitro examination of the effects of NSAIDs on repair of equine jejunum, FM and etodolac exacerbated barrier loss after ischaemia (Tomlinson and Blikslager 2004; Tomlinson et al. 2004; Little et al. 2007; Cook et al. 2009). By contrast, the NSAIDs, phenylbutazone and indomethacin, enhanced recovery of oxidant-injured equine colon in vitro (Rötting et al. 2004) and FM did not impair recovery of ischaemic equine colon in vivo (Matyjaszek et al. 2009).
The objective of this study was to examine functional and morphological effects of FM in equine colonic mucosa that had been subjected to 2 h of ischaemia followed by 18 h of reperfusion. The hypothesis was that FM would not affect barrier integrity in equine colonic mucosa in vitro after a reversible ischaemic injury in horses that had received FM in vivo. The study was designed to prolong tissue exposure to FM in vitro in horses that already received this drug in vivo (Matyjaszek et al. 2009) and thereby sustain tissue levels throughout the in vitro phase of the study. Different preparations of FM and diluent used for the commercial preparation (Banamine)1 were examined also to ensure that the in vitro effects were related solely to the NSAID. Also, the in vitro approach used in the present study in the colon allowed comparison with results of studies on nonselective NSAIDs, such as FM, on barrier function in equine jejunum in vitro (Tomlinson and Blikslager 2004, 2005).
Materials and methods
Animals and tissue preparation
Fourteen horses, donated for euthanasia for conditions that rendered them unsuitable for use, aged 2.5–20.5 years (mean 10.7, median 11) and weighing 390–574 kg (mean 498.1, median 508.5) of mixed breeds and both sexes and free of gastrointestinal disease were included. Horses were fed grass hay (2% bwt/day) with water ad libitum during a one week quarantine period before they were randomly assigned to 2 groups of 7 each: one that would receive saline solution (CON group) and the other that would receive FM (FM group) after induction of ischaemia. The experimental protocol was approved by the University of Florida Institutional Animal Care and Use Committee. The in vivo component of thestudy and the portion of the in vitro study without additions to the Ussing chambers were described previously (Matyjaszek et al. 2009).
Horses were sedated with xylazine (1.1 mg/kg bwt i.v.) and anaesthesia was induced with ketamine (2.2 mg/kg bwt i.v.) and diazepam (0.1 mg/kg bwt i.v.). Horses were intubated, placed in dorsal recumbency, and a surgical plane of anaesthesia was maintained with isoflurane (1–2%) vaporised in 100% oxygen. Isotonic polyionic fluids were infused continuously i.v. at 5–10 ml/kg bwt/h. Mean arterial blood pressure was maintained at or above 60 mmHg. Physiological monitoring during anaesthesia included electrocardiography, blood gas analysis and measurement of end-tidal partial pressure of CO2.
The ventral abdomen was prepared for aseptic surgery and draped, a ventral midline celiotomy was performed, and the pelvic flexure was exteriorised onto a sterile drape. To induce complete, or ‘no-flow’, ischaemia, intestinal clamps were applied at each end of the selected segment of pelvic flexure to compress mural vessels, and colonic arteries and veins to the same segment were ligated with umbilical tape. The colon was then replaced in the abdomen until the end of the ischaemic period and the abdominal incision was closed temporarily with towel clamps.
After 2 h of ischaemia, the clamps and ligatures were removed and the CON group of horses received saline (0.9% NaCl) solution (12 ml i.v.) and the FM group received FM as 1.1 mg/kg bwt Banamine1 i.v. The celiotomy was closed routinely and horses recovered from surgery in a padded recovery stall. Horses in both treatment groups received butorphanol (0.05 mg/kg bwt i.m. q. 4 h) for the first 8 h after surgery to provide analgesia (Tomlinson et al. 2004; Matyjaszek et al. 2009). After recovery from anaesthesia, each horse was moved to a stall and monitored for signs of pain (Pritchett et al. 2003). Physical examination findings, including rectal temperature, heart rate, respiratory rate, mucous membrane colour, capillary refill time, intestinal sounds, digital pulses, urination and defaecation were monitored for each horse every 4 h.
Each horse received one more treatment of FM or saline solution at 12 h after the first dose was administered. At 18 h after blood flow had been re-established to the ischaemic colon and with horses anaesthetised as described for the first surgery, full-thickness colon wall was harvested from ischaemic and adjacent nonischaemic colon for histological evaluations and in vitro experiments in Ussing chambers. After tissues were harvested, horses were subjected to euthanasia with an overdose of sodium pentobarbital (88 mg/kg bwt i.v.) while anaesthetised.
Ussing chamber experiments
Sheets of full-thickness colon removed from the ischaemic segment and an adjacent nonischaemic segment from each horse were transported to the laboratory at 4°C in a Krebs-Ringer-bicarbonate (KRB) solution. The KRB solution contained 112 mmol/l NaCl, 25 mmol/l NaHCO3, 10 mmol/l glucose, 5 mmol/l KCl, 3 mmol/l sodium acetate, 3 mmol/l sodium butyrate, 2.5 mmol/l CaCl2, 1.2 mmol/l MgSO4, 1.2 mmol/l KH2PO4 and 0.01 mmol/l mannitol. Tissues were constantly exposed to a solution of the same composition as used in the Ussing chambers as soon as they were removed and up to the time they were mounted in the chambers. Sharp dissection was used to remove mucosal sheets to mount in Ussing chambers with an aperture of 1.13 cm2 and 10 ml of each solution was applied to each tissue surface. The KRB solution in each chamber was maintained at a pH of 7.4 by constant perfusion with 95% O2 and 5% CO2 and at 37°C by a gas lift through water-jacketed reservoirs. Four separate solutions were prepared for in vitro incubation by adding the following: 1) no addition (KRB), 2) FM 14 µmol/l as powder (FMP), 3) FM 14 µmol/l in injectable form (FM) and 4) diluent for injectable FM (DIL). The diluent was composed of 0.1 mg/ml edetate disodium, 2.5 mg/ml sodium formaldehyde sulphoxylate, 4.0 mg/ml diethanolamine, 207.2 mg/ml propylene glycol and 5.0 mg/ml phenol in water, and was added in the same volume as for the injectable FM.
The short circuit current (µA/cm2) was recorded in each chamber on voltage clamps2 through Ag-AgCl2 electrodes connected to 4% agar bridges in KRB solution. Junction potentials of electrodes and fluid resistance were measured before mounting the tissues to allow continuous correction for any effects that these factors would have had on the low potential difference readings generated by the tissue. When tissues were mounted in chambers, the voltage clamp could then automatically correct for junction potentials of electrodes and fluid resistance. Throughout incubation, the tissues were continuously short-circuited, except at 15 min intervals when the spontaneous potential difference of tissue was measured. The transepithelial electrical resistance (TER) was calculated as Ω·cm2 by use of Ohm's law (potential difference divided by short circuit current) and used as a measure of integrity of the colonic mucosa and permeability of the paracellular pathway to ions (Matyjaszek et al. 2009). The unidirectional flux of 3H-mannitol3 from the mucosal to the serosal solution was measured in % disintegrations/min (dpm) transferred from the mucosal side to the serosal side of the tissue to measure tissue permeability to low-molecular weight markers (Matyjaszek et al. 2009). Interval from tissue collection to mounting tissue in the first chamber was approximately 12 min and to first recording of TER and addition of 3H-mannitol was 45 min. Total incubation time in Ussing chambers was 240 min.
All tissues were fixed in formalin for light microscopy following completion of Ussing chamber experiments, embedded in paraffin, and cut into 5 µm thick sections on silane-coated glass slides. Slides were stained with haematoxylin/eosin in routine manner. For the histomorphometric assessment by light microscopy, image analysis software (Image Pro Express 5.0)4 was used and 3 fields of each tissue were examined as described (Matyjaszek et al. 2009). Mucosal height (µm) was expressed as the mean vertical distance between tracings of the muscularis mucosae and the epithelial surface. Epithelial height (µm) was expressed as the mean vertical perpendicular distance between the basement membrane and the cell apex. Width (µm) of 5 clearly identifiable epithelial cells was measured in 3 sets in each field. The length (%) of mucosal surface denuded of epithelium was measured and expressed as a percentage of the total surface length of the mucosa in the section. Lifted epithelium (%) was defined as a group of at least 5 epithelial cells separated from the basement membrane but still attached to adjacent epithelial cells that held them in place. The length of lifted epithelium was expressed as a percentage of the total surface length of the mucosa in the section. Detached cells (%) were defined as cells morphologically similar to healthy cells but separated from the basement membrane in groups of at least 5 cells and were completely detached from adjacent epithelium. The length of detached cells was measured and expressed as a percentage of the total surface length of the mucosa in the section.
Tissues from each horse yielded one set of observations for each set of experimental conditions. Data were expressed as least squares mean ± s.d. Statistical software (SAS 8e)5 was used for analysis. Data that were not normally distributed were ranked before repeated measures ANOVAs were performed. Whenever a significant F test statistic was obtained for treatment, time or interaction, appropriate Bonferroni-adjusted P values were used for each family of comparisons. For all statistical analyses, P<0.05 was considered significant.
Transepithelial electrical resistance
Although nonischaemic tissues from both groups had significantly greater TER than did ischaemic tissues from both groups initially, there was no significant difference in TER between treatment groups for ischaemic tissues and for nonischaemic tissues towards the end of the incubation periods (Matyjaszek et al. 2009). There were no significant differences in TER between tissues from horses treated with saline and those treated with FM in vivo (Matyjaszek et al. 2009), and between tissues incubated with no additions and those incubated with both forms of FM and the diluent for injectable FM (Figs 1a,b). None of the additions affected TER in nonischaemic tissues (not shown).
Transmucosal mannitol flux in ischaemic and nonischaemic tissues were not significantly different at any time point between horses that received FM and those that received saline in vivo, and measurements at all time points to and including 105 min of incubation were similar (Fig 2). None of the additions affected mannitol fluxes. However, at 240 min of incubation, the transmucosal flux of mannitol was significantly (P<0.05) different from previous times for all treatments.
The epithelial height of ischaemic tissues was significantly decreased compared to the nonischaemic tissues and the percentage of denuded epithelium was significantly greater in the ischaemic tissues compared to nonischaemic tissues. The mucosal height of the ischaemic FM tissues incubated in DIL was significantly decreased compared to the nonischaemic tissues. No other significant differences were seen between ischaemic or nonischaemic, CON or FM tissues (Table 1).
Table 1. Histomorphometric measurements of colonic mucosa with data expressed as least square mean ± s.d.
Horse treatments in vivo
Ussing chamber conditions
Mucosal height (µm)
Epithelial height (µm)
Epithelial width (µm)
Denuded epithelium (%)
Lifted epithelium (%)
Detached epithelium (%)
CON and FM are colonic segments that were not ischaemic from horses that received i.v. saline and i.v. FM, respectively. CON-I and FM-I are colonic segments that were ischaemic from horses that received i.v. saline and i.v. FM, respectively. *Significant difference compared with nonischaemic tissues (P<0.05).
In this study, FM, in injectable or powdered form, did not alter the permeability of equine colonic mucosa subjected to ischaemic injury, as determined by TER and mannitol flux. These findings are in agreement with results of our previous studies evaluating the effects of in vitro phenylbutazone and indomethacin on restitution and tight junction permeability in chemically-injured equine colonic mucosa (Rötting et al. 2004). However, the results are in contrast to those obtained by studies with the same experimental design in equine jejunum, in which FM and etodolac impaired recovery of epithelial permeability (Tomlinson and Blikslager 2004; Tomlinson et al. 2004; Little et al. 2007).
Although our previous colon study (Matyjaszek et al. 2009) allowed direct comparison with results from jejunal studies on effects of NSAIDs during an 18 h recovery period from 2 h ischaemia (Tomlinson et al. 2004; Little et al. 2007; Cook et al. 2009), it did not include examination of the direct effects of FM on ischaemic colon in vitro. This information is important because previous studies on FM in ischaemic jejunum in vitro demonstrated that this NSAID could increase permeability of the damaged mucosa (Tomlinson and Blikslager 2004, 2005; Little et al. 2007; Cook et al. 2009). In vivo administration of FM before euthanasia produces tissue levels that profoundly affect colonic ion transport during the early incubation period, although this effect seems to diminish over time (Freeman et al. 1997). By adding FM to the incubation medium in the current study, we were able to maintain exposure of the injured tissue to the NSAID in vitro, thereby confirming our original findings that this NSAID does not affect permeability of ischaemic colon. The form in which FM was used and the diluent for the injectable preparation did not affect the findings.
The treatment regimen for FM in the current study (1.1 mg/kg bwt q. 12 h) was similar to that used after colic surgery in our clinic, and similar to that used in previous studies in equine jejunum, as was type and duration of ischaemia and post ischaemic recovery; however, we did use an in vitro concentration of FM that was approximately half that used in those studies (Tomlinson et al. 2004; Little et al. 2007; Cook et al. 2009). The purpose of lowering the concentration was to approximate more closely the plasma concentration achieved in vivo while correcting for protein binding (Freeman et al. 1997). Possibly an even lower dose should be used for this correction because these drugs are so avidly bound to plasma proteins, as has been shown in swine (Burr et al. 2006). To determine the actual concentration of FM that reaches the intestinal mucosa more accurately, protein-binding assays and predictive pharmacokinetic models could be used in the horse as have been used for FM in swine (Burr et al. 2006, 2009).
The disparity between improvement in TER (Fig 1) and declining mannitol permeability (Fig 2) in ischaemic tissues can be explained by the existence of several populations of size- and charge-restrictive pores in tight junctions (Watson et al. 2005; Shen et al. 2011). In an ischaemia-recovery model similar to ours, TER recovered in ischaemia-injured jejunum in horses but transmuscosal fluxes of inulin and LPS did not (Little et al. 2007). It is likely that the inflammation occuring following ischaemic injury in these tissues is responsible (Matyjaszek et al. 2009), because inflammation has been shown to selectively increase paracellular permeability to large molecules by acting on a specific population of nonrestrictive pores (Watson et al. 2005; Shen et al. 2011).
Based on our findings, FM did not adversely affect in vitro recovery of colonic mucosa from ischaemic injury in horses treated with or without FM in vivo and would support continued use of FM in horses with naturally occurring large colon diseases that are characterised by mucosal ischaemia.
Conflicts of interest
No conflicts of interest have been declared.
Source of funding
This paper was funded by a grant from The American College of Veterinary Surgeons.
1 Banamine, Schering-Plough Animal Health Corp., Union, New Jersey, USA.
2 World Precision Instruments, Sarasota, Florida, USA.
3 New England Nuclear, Boston, Massachusetts, USA.