Abstract Oil of mustard (OM), administered intracolonically, produces severe colitis in mice that is maximized within 3 days. The purpose of this study was to characterize the cytokine response, and to establish expression patterns of enteric neuronal mediators and neuronal receptors affected during active colitis. We measured the changes in the mRNA levels for neuronal receptors and mediators by real-time PCR, and cytokine and chemokine protein levels in the affected tissue. Significant increases in neuronal receptors, such as transient receptor potential A1 (TRPA1), cannabinoid type 1 receptor, neurokinin 1 receptor (NK1R) and delta-opioid receptor; prokineticin-1 receptor; and soluble mediators, such as prodynorphin, proenkephalin1, NK1, prokineticin-1 and secretory leukocyte protease inhibitor, occurred. Significant increases in cytokines, such as interleukin (IL)-1β, IL-6 and granulocyte macrophage colony stimulating factor (GM-CSF), and in chemokines, such as macrophage chemotactic protein 1 (MCP-1), macrophage inflammatory protein 1 (MIP-1α) and Kupffer cell derived chemokine (KC), were detected, with no changes in T-cell-derived cytokines. Furthermore, immunodeficient C57Bl/6 RAG2−/− mice exhibited OM colitis of equal severity as seen in wt C57Bl/6 and CD-1 mice. The results demonstrate rapidly increased levels of mRNA for neuronal receptors and soluble mediators associated with pain and inflammation, and increases in cytokines associated with macrophage and neutrophil activation and recruitment. Collectively, the data support a neurogenic component in OM colitis coupled with a myeloid cell-related, T- and B-cell-independent inflammatory component.
A considerable amount of energy has been devoted to detailing the cytokines and neuronal mediators elicited during inflammatory bowel disease (IBD), with a consensus forming around ulcerative colitis as a T-helper 2 (Th2) predominant disease, while Crohn's disease is T-helper 1 (Th1) predominant. Although these definitions are only able to approximate some aspect of disrupted cytokine patterns, there are some experimental models of colitis that reflect these facets of clinical disease. For example, trinitrobenzenesulfonic acid (TNBS) -induced colitis is recognized as a Th1-dependent rodent model of Crohn's disease, while oxazolone-induced colitis is a Th2-dependent model of ulcerative colitis.1–3 In contrast, dextran sodium sulfate (DSS) colitis which also is considered to be an example of ulcerative colitis,4 has been shown to elicit colitis in T- and B-cell-deficient RAG2−/− mice.5
In addition to altered cytokine expression patterns in colitis, experimental IBD models show that there is considerable neuronal signaling in the intrinsic enteric nervous system (ENS), and also from the ENS to extrinsic central pathways, which then return to the intestinal tract.6 Examples of the influence of neuronal components to colitis models are the inhibition of TNBS colitis by the anaesthetic lidocaine,7 the inhibition of DSS colitis by vanilloid receptor antagonists,8,9 and the stimulation of small intestinal AH neurons during TNBS-induced colonic inflammation.10
Oil of mustard (OM), allyl isothiocyanate, is a direct stimulant of small nerve fibres,11 and a potent, acute inflammatory irritant.12 OM has been used experimentally to evoke allodynia and visceral hyperalgesia following intracolonic administration to mice,13,14 and to elucidate the sensory neural pathways activated in response to its application.13–18 We have reported that intracolonic application of a 0.5% solution of OM produced a severe, transient colitis19 with the greatest damage occurring within the first 3 days.
Because OM is a neuronal stimulant, we hypothesized that OM-induced colitis may be a model for neurogenic colonic inflammation. We therefore reasoned that it would be informative to measure changes in the mediators of neuronal signalling as a consequence of OM treatment. A second aim of this work was to measure cytokines in the inflamed colon tissue in order to help determine the relative importance of myeloid and lymphocyte-derived proteins to the promulgation of inflammation in this model, and to place those findings in the context of other colitis models and human disease. Therefore, we examined distal colon tissues taken from OM-treated mice with active colitis for changes in mRNA for neuropeptides and neuronal receptors, and for cytokine and chemokine protein changes during the most severe phase of colonic inflammation. We found significant increases in colonic mRNA for a number of neuronal receptors and neuropeptides, and increased the protein expression for proinflammatory cytokines and chemokines. Furthermore, we report that immunodeficient RAG2−/− mice are susceptible to colitis induction by OM.
Materials and methods
Male CD-1 mice (Charles River Laboratories, Kingston, NC, USA), 9–10 weeks old, were used throughout these studies. Male C57Bl/6J mice were obtained from Jackson Laboratories (Bar Harbor, ME, USA), and RAG2−/− mice were purchased from Taconic Farms (Germantown, NY, USA). All treatments were carried out in accordance with the Federal Animal Welfare Act and with methods approved by the Institutional Animal Care and Use Committee of Johnson and Johnson Pharmaceutical Research and Development, LLC.
Induction of colitis
Freshly opened OM (98% pure allyl isothiocyanate; Sigma-Aldrich, St. Louis, MO, USA), was used in each experiment. Male CD-1 mice were briefly anaesthetized with ketamine/xylazine (Sigma, St. Louis, MO, USA), and held in a vertical position so that 50 μL of a solution of 0.5% OM in 30% ethanol could be administered intracolonically. The OM administration occurred to a depth of 4 cm via a syringe equipped with a ball-tipped 22 G feeding needle. C57Bl/6 mice and RAG2−/− mice received 0.5% OM in 35% ethanol, as these strains were empirically determined to be less sensitive to colitis induction when 30% ethanol was used. In all cases, the ethanol vehicle had no intrinsic colitogenic activity, causing no shrinkage, oedema, or diarrhoea, but only mild erythema in some cases, and no histological damage.19,20
At the termination of the experiment, the colons were resected, examined for signs of inflammation, weighed after removing faecal contents (which were examined for signs of diarrhoea) and the length from the aboral end of the caecum to the anus was determined. These data and observations were assigned a score as previously reported.19. The individual macroscopic indices (colon weight, colon length, diarrohea score and colon damage score) were summed to generate a macroscopic score for each colon, where 0 = normal and 15 = maximally affected. For histology, distal colon segments between the first and fourth cm from the anus of each animal were removed, rinsed in saline, fixed in 10% neutral buffered formalin, embedded longitudinally in paraffin, sectioned and stained with haematoxylin and eosin (H&E).
Distal colon homogenization and cytokine determinations
Distal colons were weighed, placed in a 2-mL Safe-Lock tube (Brinkmann Instruments, Westbury, NY, USA) and stored at −80 °C until ready for protein isolation. For homogenization, a 4.0-mm conical, stainless steel tissue grinding bead (Montreal Biotech, Montreal, Quebec, Canada) was added to each tube, followed by isotonic NaCl containing protease inhibitor cocktail (Pierce Chemicals, Rockford, IL, USA) at a ratio of 2-μL buffer for every μg of the tissue. The tissues were agitated on a Retsch MM300 Benchtop Shaker (Qiagen, Valencia, CA, USA) for 2 min at 30 cycles per second. Insoluble material in the homogenates was removed by centrifugation for 15 min at 13 000 g. The protein levels in the clarified supernatant, as measured by a bicinchoninic acid (BCA) method (Pierce Chemicals), were consistently between 20 to 30 mg mL−1 for all treatments. Cytokines and chemokines in the homogenates were measured using a Lincoplex mouse cytokine and chemokine kit (Linco Research, St. Charles, MO, USA) according to the manufacturer's instructions. The measured levels exceeded the lower limits of sensitivity, which ranged from 0.3 pg mL−1 for interleukin (IL)-4 to 10.3 pg mL−1 for IL-10, and 6 pg mL−1 for IL-12.
After dissection, the colons were rinsed in sterile saline to remove faecal matter. Fifty to one-hundred milligrams of tissue from the distal colon were each immersed in 1.5 mL RNAlater (Qiagen, Valencia, CA, USA) in a 2-mL Safe-Lock tube (Brinkmann Instruments, Westbury, NY, USA) and stored at 4 °C. For RNA isolation, RNAlater was removed and replaced with 1-mL Qiazol (Qiagen, Valencia, CA, USA). A 4.0-mm conical, stainless steel tissue grinding bead (Montreal Biotech, Montreal, Quebec, Canada) was added to each vial and the samples were then agitated on a Retsch MM300 Benchtop Shaker (Qiagen, Valencia, CA, USA) for 2 min at 30 cps. Following chloroform addition and centrifugation, the aqueous phase was collected and mixed with 70% ethanol and added to the wells of an Rneasy 96 Plate (Qiagen, Valencia, CA, USA). Following the manufacturer's instructions, the samples were washed and eluted in RNase-free water. The RNA preparation was then treated with Dnase using a DNA-free Dnase Treatment Kit (Ambion, Austin, TX, USA), according to the manufacturer's instructions. RNA quantity was assessed by ultraviolet spectrophotometry. RNA integrity was assessed on the 2100 Bioanalyzer using the RNA 6000 Nano LabChip kit (Agilent Technologies, Palo Alto, CA, USA) according to the manufacturer's instructions.
A 1 : 1 (v/v) addition of the reagent mixture from the High Capacity cDNA Archive Kit (Applied Biosystems, Foster City, CA, USA) was added to equal amounts of each RNA sample, which had first been diluted to 250 ng μL−1. This mixture was incubated at 37 °C for 2 h. Triplicate samples were diluted by a factor of 15.625 with water, and 40 ng of cDNA in 5 μL were loaded into 384-well clear, optical reaction plates (Applied Biosystems, Foster City, CA, USA). The following 6-carboxy fluorescein/minor groove binding (FAM/MGB) primer/probes were purchased from Applied Biosystems (Foster City, CA, USA). TRPA1, Mm00625268_m1; carboxypeptidase E (CPE), Mm00516341_m1; cannabinoid type 1 receptor (CB1R), Mm00432621_s1; delta-opioid receptor (dOPR), Mm00443063_m1; NK1R, Mm00436892_m1; Prokineticin 1 receptor (PK1R), Mm00517546_m1; NK1, Mm00436880_m1; prodynorphin (pDYN), Mm00457572_m1; Proenkephalin 1 (pENK1), Mm01212875_m1; transient receptor potential V1 (TRPV1), Mm01246302_m1; PK1, Mm01204733_m1. Primers and probes for secretory leukocyte protease inhibitor (SLPI) were as follows: forward primer-GCTGTGAGGGTATATGTGGGAAA; reverse primer-CGCCAATGTCAGGGATCAG; hybridization probe-VIC-TCTGCCTGCCCCCGATGTGAG-TAMRA. Primer/probes were diluted with 2X RT-PCR Master Mix (Applied Biosystems, Foster City, CA, USA), and 7 μL of this solution was added per well. The plates were analysed in a 7900HT Fast Real-Time (RT) PCR System (Applied Biosystems, Foster City, CA, USA). Real-time PCR can reliably detect mRNAs at 10 copies per cell.
The experimental groups were analysed for significance of differences between the means of treatment groups and control groups by anova with Bonferroni's or Dunnett's post-test using GraphPad Prism (version 3.03; GraphPad Software, San Diego, CA, USA). Prism utilizes the Kolmogorov–Smirnov test for normality in all analyses.
Time course of colitis development
OM promptly induces colon inflammation, with damage detected within 2 h of OM application, at which time there are signs of inflammatory damage, diarrhoea, and increased colon weight as a measure of oedema. Inflammatory damage was increased further at 24 h and was then maintained out to 72 h. Oedema, seen as increased colon weight, also exhibited a stepwise increase at 48 and 72 h. There was no colon shrinkage observed until 24 h, and this shrinkage was maintained for the ensuing 48 h. Macroscopic damage scores, as a comprehensive measure of colitis severity, showed strong colitis development at 2 and 6 h, with increased severity at 24 to 72 h (Fig. 1).
Neuronal receptors in colonic tissues
Because OM is a known neuronal activator, we first looked at the changes in mRNA for neuronal receptors associated with pain and inflammation, or in some cases known for the effects on intestinal motility. The data in Fig. 2 show early and sustained increases in TRPA1, the receptor for OM,21 while mRNA for TRPV1, the capsaicin-sensitive vanilloid receptor1,22,23 was not significantly increased (P > 0.05) and after 24 h diminished to below starting levels (P < 0.05). This is consistent with the published data showing OM signalling and inflammation is independent of TRPV1.11 Sustained increases in NK1R mRNA were observed, while dOPR mRNA increased in the first 6 h, but then decreased to below basal levels. In contrast, mu-opioid receptor (mOPR) mRNA was not detected before or after colitis induction. Cannabinoid type 1 receptor mRNA expression increased in the first 6 h, and then decreased to below baseline levels at 24 h. We found a sharp increase in PK1R at 2 h, which then returned to baseline. Previous studies demonstrate that PK1R is expressed on gastrointestinal smooth muscle24 and on macrophages.25
Neuropeptides and soluble mediators in colonic tissues
Neuropeptide mRNA levels that are related to the aforementioned receptors were examined. Fig. 3 demonstrates significant increases in NK1 mRNA over the 72-h period, and increases in prodynorphin and proenkephalin 1 mRNA, even though there had been no detectable mOPR mRNA. Message RNA for carboxypeptidase E, an enzyme that processes a number of peptide precursors,26 such as enkephalins, was transiently increased in the first 6 h, after which it decreased to below baseline levels. Message RNA for SLPI,27 a marker of epithelial cell activation, increased throughout the 72-h period examined. The profile for SLPI expression essentially matched that for the macroscopic scores. PK1 mRNA was dramatically increased from essentially negligible levels, peaking at 2 h and then declining as rapidly.
Cytokines and chemokines stimulated during colitis
The data in Fig. 4 show cytokine and chemokine expression data from tissues taken from mice spanning the period of maximum inflammatory damage and colitis. There were very high levels of the macrophage-derived cytokines IL-1β and IL-6, reaching 706 ± 168 and 2428 ± 1375 pg mL−1, respectively. The tumor necrosis factor (TNF) levels were not statistically significantly elevated, although there appeared to be a trend towards elevated levels. Macrophage and neutrophil activating/recruiting cytokines and chemokines that were likewise elevated were GM-CSF, MIP-1α, MCP-1, and KC. There were no statistically or biologically significant increases in the T-cell derived chemokine regulated, upon activation, normal T-cell expressed and secreted (RANTES). In addition, there were no significant increases observed for T-helper cell-derived cytokines interferon (IFN)-γ, IL-4, IL-5, IL-9, IL-10 or IL-13 (data not shown). The baseline levels ranged between 8 and 10 pg mL−1, with no changes occurring after OM administration. IL-12 levels were increased from a baseline of 8 pg mL−1 up to a peak of 26 pg mL−1 at 24 h, which then returned to baseline. Although it was statistically significant, given the small magnitude of this increase when compared with the higher levels seen for proinflammatory chemokines and cytokines in other IBD models, this may not be as biologically relevant.
OM colitis in immunodeficient RAG-2−/− mice
The absence of a robust T-cell-derived cytokine response in OM colitis called into question the necessity for T-cell-mediated pathology. Accordingly, we induced colitis in T-cell and B-cell-deficient RAG2−/− mice. Fig. 5 shows photographs of colons from RAG2−/− mice. The vehicle-treated colon had a normal appearance, while the colon from an OM-treated mouse had a shortened, thickened, distended proximal and distal colon, with inflammation in the distal and proximal colon, and diarrhoea throughout. Internally, the proximal colons displayed profound erythema, and the distal colons exhibited multiple petechiae, overt bloody lesions and transmural ulcers that in instances completely penetrated the bowel wall. An inflamed lesion is prominent in the distal colon (red arrow, Fig. 5). These colons had exactly the same appearance as those previously reported for CD-1 mice with OM colitis.19
The data in Fig. 6 compare macroscopic scores for RAG2−/− mice on a C57Bl/6 background and compare them with wt C57Bl/6 mice. Both strains exhibited similar macroscopic scores, with a score of 7.0 ± 1.1 for RAG2−/− mice and 6.8 ± 1.4 for C57Bl/6 wt mice. These values indicate that the inflammation that occurs following OM administration is not lymphocyte dependent, and is consistent with the cytokine data described earlier.
The digital images shown in Fig. 7 reveal comparable histological damage in colons from OM-treated RAG2−/− and wt CD-1 mice. Both show heavy infiltrates in mucosa, submucosa and muscularis, where none exist in the normal or ethanol-treated tissue. Smooth muscle and epithelial architecture showed extensive destruction in OM-treated tissue from both strains, and submucosa were greatly expanded. No differences were observed between normal and ethanol-treated tissues in the RAG2−/− mice. We have never observed histological differences between tissues from normal and ethanol-treated CD-1 mice.19
The objectives of this study were to characterize the effects of intracolonic OM on intestinal cytokines, and molecules associated with neuronal signalling. A strong body of evidence supports a relationship between intestinal inflammation and changes in neuronal signalling and receptor expression.10–12,28–32 What may be less clear in the etiology of human IBD is whether the primary events in colitis induction are neurogenic or if the inflammation and tissue damage precede neurochemical and neurophysiological changes. Because OM is a known neuronal stimulant, the OM colitis model we developed19 may be a useful means to study the colitis of neurogenic origin. In this report, changes in the expression of a number of neuronal mediators and their receptors were determined following induction of OM colitis. Indeed, neuropeptides and neuronal receptors implicated in pain sensation and transmission were upregulated. This is consistent with the known actions of OM, which include increased signalling via NK1, CB1R and endogenous opioids, and nitric oxide, but not via TRPV111,12,15–17,29 Similarly, proinflammatory cytokines and chemokines, known to activate macrophages and neutrophils were affected, but T-helper cell-derived cytokines were not. Finally, OM induced colitis in wild-type and immunodeficient C57Bl/6 strains of mice in addition to the originally described CD-1 mice.
Neuronal receptor and neuropeptide mRNA expression were upregulated within the first 2 h. These include TRPA1, the receptor for OM which also responds to noxious cold.21 TRPV1, associated with pain transmission, responses to low pH, capsaicin and noxious heat9,22,23,33 was not significantly increased, and was, in fact, decreased after 24 h. NK1R, associated with pain transmission and sensation,16 was stimulated for the entire 72-h period. Delta-opioid receptor was transiently upregulated, while mOPR was not detected. Similar results for mOPR were observed in other rodent colitis models as well (S. Prouty, personal communication). CB1R was transiently increased, consistent with other reports related to pathways associated with pain sensation stimulated by OM, and is also consistent with data demonstrating increased CB1R expression in human IBD34 and during OM colitis,20 where it is found not only on myenteric neurons, but also significantly elevated on endothelial cells. It is therefore important to realize that although considered to be of neuronal origin, many of these mediators and their receptors are also expressed on and by non-neuronal tissues.
We found that mRNA for a number of peptides increased in colon tissues from mice undergoing OM colitis. NK1 (substance P) was upregulated, as were proenkephalin-1 and prodynorphin, which are precursors to endogenous mu- and delta-opioids, respectively. Secretory leukocyte protease inhibitor, a surrogate marker for epithelial regrowth, also increased in expression throughout the 72-h period, indicating that during this period of intestinal inflammation and epithelial damage, there is a drive to restore the epithelial barrier. Message for PK1, which was originally associated with increased intestinal motility,24 and more recently has been found to have proinflammatory properties,25 was sharply increased at 6 h. Thus, OM colitis, as regards its neurogenic component, has some similarities to other colitis models that have been demonstrated to evoke neuronal signalling.7,35–37
Cytokine protein determinations in the colon tissues revealed increases in proinflammatory cytokines and chemokines that are macrophage derived or which contribute to macrophage and neutrophil activation and recruitment; IL-1β, IL-6, GM-CSF, MIP-1α, MCP-1 and KC. As TNF is an early response gene whose action on other cells is a critical component of their subsequent activation, it is not surprising that there were low levels after 24 h in this acute model. It is important to note that many of those proinflammatory cytokines and chemokines can be expressed by, and targeted also to non-hematopoietic cell populations. For example, MCP-1 and IL-6 are expressed by endothelium, and KC by fibroblasts occurring at sites of tissue injury. We had previously reported data showing that the cellular infiltrate in OM colitis consists of macrophages, neutrophils and lymphocytes.19 It is therefore significant that the absence of major changes in T-helper cytokines correlates with the ability to induce colitis in T- and B-cell-deficient RAG2−/− mice. Collectively, these data support a role for myeloid lineage cells in the development or progression of OM colitis, and less for lymphocytes. In addition, the ability to induce OM colitis in mouse strains other than CD-1 mice, the strain originally used,19 extends the utility of the model to genetically engineered mice. In the case of mice with a C57Bl/6 background, it was necessary to use 0.5% OM dissolved in 35% ethanol instead of 30% ethanol.
The data presented in this study highlight a number of important similarities and possibly more important differences between some aspects of OM colitis and other models. OM colitis clearly has a vital neuronal component, as evidenced by the elevated expression of NK1, and mu- and delta-opioids, and the receptors NK1R, dOPR, CB1R and TRPA1. This is the first example of TRPA1 and dOPR increases in a colitis model, albeit there are other reports related to the receptors for cannabinoid, tachykinin and other TRP receptors in experimental colitis. Tachykinins are well known to be involved in various aspects of gut function (reviewed by Holzer and Holzer-Petsch),38 but there are conflicting data in colitis models regarding NK1R antagonism. TNBS colitis was reported to show decreased expression of NK1 receptor and not to be improved by NK1R antagonism,39–41 whereas an NK1R antagonist decreased inflammation in the DSS colitis model.37 In addition, blocking NK1R also displayed a reduction in colitis in acetic acid induced colitis,42 and Clostridium difficile ileitis.43 These data implicate the importance of increased NK1R in these models, and it may play a similar role in OM colitis. Cannabinoid receptors were recently shown to be increased in human IBD tissues34 and CB1R mRNA and protein20– the latter occurring on endothelium is increased in OM colitis. DSS colitis is sensitive to TRPV1 antagonism,8,9 with indications that TRPV1 is increased in the experimental colitis. In contrast, we found in the present study that there was no significant increase in TRPV1 mRNA in OM colitis.
The rapid and sustained upregulation of NK1 and NK1R mRNA are consistent with the mode of action of OM,16 and as a matter of speculation, suggest a starting point for testing a mechanism of action for OM colitis. Neuronally derived proinflammatory NK1R agonists can stimulate macrophages, mast cells and endothelium to release cytokines and mediators which can contribute to and sustain an inflammatory infiltrate, thereby engendering the tissue damage and barrier loss observed. Simultaneous stimulation of other neuronal receptors and mediators contribute to altered gut function. However, the temporal appearance of neuronal mediators, receptors and cytokines, and their selective inhibition, need to be studied more carefully in order to relate nerve stimulation to cytokine production, tissue damage and altered gut function before a mechanism could properly be proposed.
The cytokine profile seen for OM colitis is different from those exhibited by other models, and has a unique macrophage predominance, lacking significant changes in T-cell cytokines. In contrast, TNBS and SAMP-1/Yit IBD models exhibit elevated Th1 cytokine production, oxazolone colitis exhibits over-production of Th2 cytokines, and DSS colitis first exhibits a preference for Th1 cytokines that shifts under chronic conditions to a combined Th1 and Th2 response.2 The absence of strong increases for T-cell cytokines in OM colitis may explain its transient nature, while the strong acute response may be explained by the robust macrophage cytokine and chemokine response. This may serve to highlight its utility for examining mechanisms responsible for the induction and early phase of colitis.
We have previously reported that cannabinoid receptor (CBR) agonists were able to ameliorate OM colitis and DSS colitis,20 thereby demonstrating the potential utility of the OM model for testing experimental therapeutics and its mechanism being consistent with other published data on the influence of cannabinoids in colitis.34,36 The OM model is also noteworthy in that this is the first report of increases in PK1 and PK1R in colitis, and for SLPI, whose expression may be part of a protective or tissue restorative mechanism. After a 3- to 4-week delay, a postinflammatory IBS-like syndrome occurs,19 indicative of altered neuronal signalling as a result of the colonic inflammation. These data provide strong support for examining the early changes in, and influences of, neuronal receptors and secreted mediators in OM models.
In summary, OM colitis differs from other models, having a rapid onset, an abbreviated temporal profile, and severe tissue damage appearing within hours of induction. Although there is a heavy inflammatory infiltrate comprised of macrophages, neutrophils and lymphocytes, there does not appear to be a T-helper requirement for colitis induction. This model may therefore be useful for studying the possible early neuronal- and myeloid cell-mediated events in colitis development. The rapid onset and acute nature of the model reduces the time required for pharmacological testing. It can also allow the testing for T-cell-independent immunomodulators and neuromodulators associated with inflammation. Oil of mustard colitis clearly has a neurally related component and a myeloid cell-related component. The active production of proinflammatory cytokines and chemokines, and of neuropeptides and their receptors strengthens the concept of neuronally mediated bowel dysfunction and inflammation in the context of a neuro-immune axis of inflammation as a potentially important factor in the early development of colitis and localized intestinal dysfunction.
The authors are grateful to Cynthia Smith and John Mabus (J&J PRD, Spring House, PA, USA) for their assistance in RNA isolation and RT-PCR studies.