high mobility group box-1 protein is an inflammatory mediator in necrotizing enterocolitis (nec):...

High-mobility group box 1 protein is an inflammatory mediator in necrotizingenterocolitis: protective effect of the macrophage deactivator semapimod

Ruben Zamora,1 Anatoli Grishin,1 Catarina Wong,1 Patricia Boyle,1 Jin Wang,1

David Hackam,1 Jeffrey S. Upperman,1 Kevin J. Tracey,2 and Henri R. Ford1

1Department of Pediatric Surgery, Children’s Hospital of Pittsburgh, and Department ofSurgery, University of Pittsburgh, Pittsburgh, Pennsylvania; and 2Laboratory of BiomedicalScience, North Shore-Long Island Jewish Research Institute, Manhasset, New York

Submitted 15 February 2005; accepted in final form 6 June 2005

Zamora, Ruben, Anatoli Grishin, Catarina Wong, PatriciaBoyle, Jin Wang, David Hackam, Jeffrey S. Upperman, Kevin J.Tracey, and Henri R. Ford. High-mobility group box-1 protein is aninflammatory mediator in necrotizing enterocolitis: protective effectof the macrophage deactivator semapimod. Am J Physiol GastrointestLiver Physiol 289: G643–G652, 2005. First published June 9, 2005;doi:10.1152/ajpgi.00067.2005.—High-mobility group box 1 (HMGB1)is a late mediator of endotoxemia known to stimulate the productionof proinflammatory cytokines that are putative mediators of intestinalinflammation associated with necrotizing enterocolitis (NEC). Wehypothesized that HMGB1 is also involved in the pathogenesis ofNEC. We examined the expression of HMGB1 and the effect of thenovel drug semapimod on intestinal inflammation in an experimentalmodel of NEC in neonatal rats. Newborn rats were subjected tohypoxia and fed a conventional formula by gavage (FFH) or werebreast fed (BF). Rats were killed on day 4, and the distal ileum washarvested for morphological studies and Western blot analysis. FFHnewborn rats but not BF controls developed intestinal inflammationsimilar to the histological changes observed in human NEC. We foundthat the expression of HMGB1 and its receptor for advanced glycationend products (RAGE) as well as that of other apoptosis/inflammation-related proteins (Bad, Bax, inducible nitric oxide synthase, and cy-clooxygenase 2) was upregulated in the ileal mucosa of FFH newbornrats compared with BF animals. Administration of the drug semapi-mod inhibited the upregulation of those proteins and partially pro-tected the animals against the FFH-induced intestinal injury. Elevatedlevels of HMGB1 were also found in ileal samples from infantsundergoing intestinal resection for acute NEC. Our results implicateHMGB1 and RAGE as important mediators of enterocyte cell deathand hypoxia-induced injury in NEC and support the hypothesis thatinhibitors such as semapimod might play a therapeutic role in chronicintestinal inflammation characterized by this animal model.

gut inflammation; hypoxia; CNI-1493; newborn rats

NECROTIZING ENTEROCOLITIS (NEC) is the most frequent and mostlethal disease that affects the gastrointestinal tract of thepremature infant (16). The overall mortality rate for patientswith NEC ranges from 10% to 70% (22) and approaches 100%for patients with the most severe form of the disease, which ischaracterized by involvement of the entire bowel (pan-necro-sis) (37). Survivors may develop short-bowel syndrome, recur-rent bouts of central line-associated sepsis, malabsorption andmalnutrition, and liver failure as a result of longstandingadministration of total parenteral nutrition and eventually mayrequire liver-small bowel transplantation (27, 37). Thus NEC

represents one of the most important diseases seen in theneonatal population.

Although numerous risk factors including prematurity (47),hypoxia (26), formula feeding (18), bacterial infection (40),and intestinal ischemia (1) have been implicated in the patho-genesis of NEC, the exact etiology of the disease remainsundefined. To understand the pathogenesis of NEC, previousauthors have relied on an intravenous administration of proin-flammatory cytokines or invasive surgical procedures, such asocclusion of the mesenteric vessels, to reproduce the morpho-logical and clinical changes seen in infants with NEC (16).However, these artificial insults do not parallel the humandisease. We recently revisited an animal model originallydescribed by Barlow et al. that uses formula feeding after abrief period of hypoxia (FFH) to induce experimental NEC (5).We were able to modify this model and use it to study theexpression of various cytokines in the intestine (32). Becauseimmunological and cellular events can be easily studied in therat, this model is also ideally suited for further investigationsinto the pathogenesis and treatment of NEC.

Numerous inflammatory mediators have been implicated inthe cascade leading to the hemodynamic instability associatedwith sepsis. Elevated levels of IL-6 and TNF-� have beenmeasured in infants with NEC, bacterial sepsis, or meningitis(21). More recently, we have shown that nitric oxide (NO)modulates intestinal changes in septic shock (17, 43). Further-more, the end products of NO metabolism have been shown tobe elevated in newborn infants as well as in adult patients withclinical sepsis (34, 41). Several studies suggest these and othermediators of sepsis are also involved in the pathogenesis ofNEC. In this context, we (17) previously reported the upregu-lation of inducible NO synthase (iNOS) mRNA in the intestineof infants with acute NEC and its downregulation by the timeof intestinal stoma closure, after the infants recovered from theacute inflammatory process (17). A similar pattern was seen forIFN-� but not for IL-1, IL-6, transforming growth factor(TGF)-�, or TNF-�. The pattern of cytokine mRNA expressionin our model of experimental NEC parallels that seen inpatients with NEC (17).

More recently, Wang et al. identified and characterizedhigh-mobility group (HMG) box 1 (HMGB1) protein (previ-ously known as HMG-1 or amphoterin) as a potential latemediator of lethality in a mouse model (44) and a regulator ofhuman monocyte proinflammatory cytokine synthesis (3). This

Address for reprint requests and other correspondence: R. Zamora, Dept. ofSurgery, Univ. of Pittsburgh, W1540 Biomedical Science Tower, 200 LothropSt., Pittsburgh, PA 15213 (e-mail: [email protected]).

The costs of publication of this article were defrayed in part by the paymentof page charges. The article must therefore be hereby marked “advertisement”in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Am J Physiol Gastrointest Liver Physiol 289: G643–G652, 2005.First published June 9, 2005; doi:10.1152/ajpgi.00067.2005.

0193-1857/05 $8.00 Copyright © 2005 the American Physiological Societyhttp://www.ajpgi.org G643

protein belongs to the HMGB group, one of the three super-families of HMG nuclear proteins subdivided according totheir characteristic functional sequence motifs. The functionalmotif of this family is called the HMG box, and, according tothe revised nomenclature of HMG nuclear proteins, HMG-1has been renamed to HMGB1 (9). The expression of HMGB1and its role in intestinal inflammation and, in particular, inNEC are not known at present. The objective of this study wastwofold. First, we examined whether the cell death-relatedprotein HMGB1 is differentially expressed in the injured smallintestine of newborn rats in experimental NEC and in theinflamed intestine of human infants with acute NEC. We thenevaluated the effects of the macrophage deactivator semapi-mod (formerly known as CNI-1493) on the expression ofproapoptotic and proinflammatory proteins and on the severityof intestinal inflammation in our experimental model of NECas well as its effects on rat intestinal epithelial cells challengedwith bacterial LPS. The results indicate that HMGB1 is animportant mediator of hypoxia-induced gut injury and thatsuppression of proinflammatory cytokines with inhibitors suchas semapimod partially protects against intestinal epithelial celldeath both in vitro and in vivo.

MATERIALS AND METHODS

Reagents. LPS from Escherichia coli 0127:B8 was from Sigma (St.Louis, MO). Peroxynitrite was from Alexis Biochemicals (San Diego,CA). The drug semapimod was kindly provided by Cytokine Pharma-Sciences (King of Prussia, PA). The different antibodies used forWestern blot analysis were purchased from the following sources:rabbit polyclonal anti-Bad (sc-943) and anti-Bax (P-19) antibodieswere from Santa Cruz Biotechnology (Santa Cruz, CA); cyclooxy-genase (COX)-2 (murine) polyclonal antibody was from CaymanChemical (Ann Arbor, MI); anti-HMGB1 was a polyclonal rabbitantisera against native calf thymus HMG1/HMGB1 from UpstateBiotechnology (Lake Placid, NY); rabbit polyclonal anti-iNOS anti-body was from BD Transduction Laboratories (Lexington, KY);rabbit polyclonal anti-receptor for advance glycation end products(RAGE) antibody was from Affinity BioReagents (Golden, CO); andanti-p38 and anti-phospho-p38 antibodies were from Cell SignalingTechnology (Beverly, MA).

Animal model of NEC. All experiments were carried out accordingto an animal protocol approved by the Animal Research and CareCommittee of the Children’s Hospital of Pittsburgh. Pregnant time-dated Sprague-Dawley rats (Charles River Laboratories; Wilmington,MA) were induced at term using a subcutaneous injection (1–2U/animal) of Pitocin (Monarch Pharmaceuticals; Bristol, TN). Imme-diately after birth, the neonates were weighed and randomized intoone of the different treatment groups. Group 1 consisted of neonatalrats left with their mother and thus were breast fed (BF). Group 2(FFH group) consisted of neonates separated from their mothers,housed in a temperature- and humidity-controlled incubator (OhioMedical Products; Madison, WI), and gavaged with a special rodentformula (0.2 ml, see below) 2 times/day and subjected to 10 min ofhypoxia (5% O2-95% N2; Prax Air; Pittsburgh, PA) thrice daily in amodular Chamber (Billups-Rothenberg; Del Mar, CA) as follows:pups were fed in the morning posthypoxia, exposed to a secondhypoxic insult after 4 h, and then subjected to the final hypoxic insultfollowed by the final feed. Rats in group 3 were treated in a similarfashion to those in group 2; however, the animals received theexperimental drug semapimod (0.1–1 mg/kg ip once daily, vehicle:5% dextrose) right before hypoxia. The formula composition con-sisted of 15 g Similac 60/40 (Ross Pediatrics; Columbus, OH) in 75ml of Esbilac canine milk replacer (Pet-Ag; Hampshire, IL) asdescribed by Barlow et al. (5) and was designed to approximate the

protein and caloric content of rat breast milk. The rats were killed ondifferent days as indicated, and intestinal samples (segments of theterminal ileum for the present study) were harvested for morpholog-ical studies and Western blot analysis as described below.

Morphological evaluation of intestinal samples. Rats were killedon different days as indicated, and the distal ileum was harvested formorphological studies. For light microscopy, hematoxylin and eosin-stained slides were prepared per the standard protocol (15). A pathol-ogist from the Children’s Hospital of Pittsburgh (Dr. R. Jaffe) blindedto the experimental groups graded the morphological changes in theintestinal epithelium. The criteria for each histological grade (0–3)were as follows: 0, normal (no pathological change, different epithe-lial patterns are noted: clear, vacuolar, and inclusion type); 1, mild(occasional neutrophils, separation of villus core, and mild damage toenterocytes); 2, moderate (submucosal edema, epithelial sloughing,and marked presence of neutrophils); and 3, severe (denudation ofepithelium with loss of villi and transmural necrosis or perforation) (32).

Western blot analysis. Newborn rats were killed as indicated, andsegments of the terminal ileum were isolated. The mucosa was gentlyscraped from each segment and immediately placed in cold lysisbuffer containing 62.5 mM Tris (pH 6.6), 10% glycerol, 1% SDS, andprotease inhibitors (10 �g/ml leupeptin, 5 �g/ml pepstatin, 2 �g/mlaprotinin, and 0.5 mM PMSF, all from Sigma). The samples were thenhomogenized and boiled for 1 min followed by centrifugation at10,000 g for 30 min to remove the cell debris. The protein concen-tration in the supernatant was determined using the bicinchoninic acidprotein assay kit from Sigma with BSA as the standard. Proteinsamples (equivalent to 50 �g) were resolved on 12% SDS-polyacryl-amide gels using a Bio-Rad minigel system (Hercules, CA) and thenelectroblotted onto polyvinylidene difluoride membranes (Millipore;Beford, MA). After being blocked for 1 h with milk (5% in PBS with0.1% Tween 20) at room temperature, the membranes were probed for1 h at room temperature with the primary antibodies dissolved in 1%milk-PBS-Tween 20 at the following dilutions: HMGB1 (1:1,000),Bad (1:1,000), Bax (1:500), iNOS (1:750), COX-2 (1:500), andRAGE (1:1,000). The membranes were then thoroughly washed andincubated with secondary antibodies (horseradish peroxidase-conju-gated goat anti-rabbit or goat anti-mouse IgG, Pierce; Rockford, IL) at1:15,000 dilution (in PBS-Tween with 1% milk) for 1 h beforedetection. Protein bands were visualized using a SuperSignal chemi-luminescence substrate (Pierce) according to the manufacturer’s in-structions.

Cell culture and Western blot analysis. The rat intestinal crypt cellline IEC-6 was purchased from the American Type Culture Collection(Manassas, VA). Cells were grown in tissue culture medium consist-ing of DMEM with 4.5 gm/l glucose (Bio-Whittaker; Walkersville,MD) supplemented with 5% fetal bovine serum (Bio-Whittaker), 0.02mM glutamine (GIBCO-BRL; Grand Island, NY), 0.1 U/ml insulin,100 U/ml penicillin, and 100 �g/ml streptomycin (GIBCO-BRL) at37°C and 10% CO2. Cells were pretreated with semapimod (0.1–10�M solutions prepared in 5% dextrose) followed by a short exposureto LPS (5 �g/ml), peroxynitrite (ONOO�; 50 �M) or cytomix [acombination of TNF-� (10 ng/ml), IL-� (1 ng/ml), and IFN-� (1,000U/ml)] for 10 min. Preparation of the ONOO� solution and celltreatment was as previously described (36). The lysates were collectedfor Western blot analysis as follows: proteins were extracted for 10min at 4°C with lysis buffer [20 mM Tris (pH 8.0), 150 mM NaCl, 1%NP-40, 0.5% sodium deoxycholate, and 0.1% SDS) supplementedwith protease and phosphatase inhibitors (1 mM PMSF, 1 mMbenzamidine, 50 �g/ml aprotinin, 0.5 mM Na3VO4, 20 mM NaF, and0.5 mM phenylarsine oxide). Aliquots containing 50 �g total proteinwere resolved by SDS-PAGE. Gels were electroblotted onto nitrocel-lulose membranes. After a 1-h incubation in blocking solution (PBS,0.1% Tween 20, and 2% fish gelatin; pH 7.5), membranes wereincubated with primary (phospho-p38 and p38) and secondary horse-radish peroxidase-conjugated antibodies as recommended by the man-

G644 HMGB1 EXPRESSION AND PROTECTIVE ROLE OF SEMAPIMOD IN NEC

AJP-Gastrointest Liver Physiol • VOL 289 • OCTOBER 2005 • www.ajpgi.org

ufacturers. Membranes were then impregnated with luminol reagentand exposed to X-ray film.

Statistical analysis. Results are expressed as means � SE or SD asindicated. Differences among groups were analyzed by Student’st-test, �2-test, or one-way ANOVA followed by Tukey test or Fisher’sleast-significance difference test where appropriate (SigmaStat 2.03,SPSS; Chicago, IL).

RESULTS

Overall mortality in the animal model of NEC. Newborn ratswere fed either breast milk (BF) or a commercial formula up to4 days with or without a 10-min hypoxic insult thrice asindicated in MATERIALS AND METHODS. As we previously re-ported, whereas BF animals gained weight in a normal fashion,formula-fed animals gained significantly less weight than theirBF counterparts (32). Moreover, we observed a higher mortal-ity rate (days 0–3) in formula-fed animals compared with theBF group: 34.33% for FFH rats (484/1,410) vs. 1.13% for BFcontrols (8/707) (P � 0.001, analyzed by �2-test). The totalnumbers of animals are cumulative and reflect the work overthe past 3 1/2 yr.

Morphological changes of intestinal samples in experimen-tal NEC. We have previously shown that morphological anal-ysis of ilea from 4-day-old formula-fed rats revealed variousdegrees of intestinal inflammation compared with ileal seg-ments from BF controls. These inflammatory changes closelyresembled the histological alterations observed in human NECsamples (32). To determine the kinetics of the development ofintestinal inflammation in this model, we killed both BF andformula-fed animals on each consecutive day after birth andharvested the last 2 cm of their distal ilea for morphologicalanalysis. The specimens for day 0 were from animals killedwithin 3 h after birth and showed minimal morphologicalchanges (18.8% had a histological score of �1, n 48animals). The overall incidence of morphological and patho-logical changes characteristic of NEC is shown in Table 1.Intestines of BF pups were normal and rarely showed abnormalmorphology. In contrast, in the FFH group, morphologicalevidence of NEC was detected as early as day 1 (mainly villouscore separation and presence of occasional neutrophils) withmoderate to severe damage at day 4 (histological scores �1).Analysis of ileal specimens from FFH newborn rats killed atday 5 or day 7 showed severe intestinal damage but notsignificantly different from FFH animals on day 4 (not shown).On the basis of this observation, we performed our subsequent

studies utilizing the intestinal samples harvested from BF andFFH animals exposed to hypoxia killed on day 4 after birth.

Expression of HMGB1 protein and its receptor RAGE in theintestine. To determine whether HMGB1 is expressed in theileum of newborn rats and whether its expression is related tointestinal damage, we analyzed the expression of this protein inileal mucosa from both BF and FFH animals killed on differentdays as described in MATERIALS AND METHODS. Although themature HMGB1 protein generally migrates as a 30-kDa bandon SDS-PAGE (48), it can also appear as a doublet or two veryclose bands around 30 kDa, where the lower band is usuallymore intense (K. Tracey, unpublished observations). We foundthat intestines of BF pups showed a basal expression ofHMGB1; however, larger protein amounts were present in theileal mucosa of FFH animals at day 4 (Fig. 1). Interestingly, insome cases, those animals that showed histological evidence ofmoderate to severe intestinal damage also showed increasedexpression of a lower molecular mass protein (below the14-kDa molecular mass marker and referred herein as 7- to10-kDa bands) detected with the same anti-HMGB1 polyclonalantibody (Fig. 1).

Semapimod protects against formula-feeding/hypoxia-medi-ated intestinal injury and reduces ileal HMGB1 and RAGEprotein expression. We hypothesized that the macrophagedeactivator semapimod will exert a protective effect againstformula-feeding/hypoxia-mediated intestinal injury and that itwill affect the expression of HMGB1 protein. We found that anintraperitoneal administration of semapimod (0.1–1 mg/kgdaily) prevented intestinal damage at doses higher than 0.5mg/kg, as represented in Fig. 2 as a reduction in the incidenceof NEC. We also examined the effect of semapimod onHGMB1 protein expression. Whereas we found a basal expres-sion of HMGB1 protein in the ileal mucosa of BF animals, asignificant increase was observed in the ileal specimens of FFHanimals, which was reduced after the administration of semapi-mod (Fig. 3A). Administration of vehicle alone had no effectand was not different from the FFH group (not shown). Thisinhibition of protein expression correlated with a decrease inthe severity of injury as assessed by histology (Fig. 3B).Receptor signal transduction of HMGB1 occurs in part throughRAGE in different cell types (2). We also examined theexpression of RAGE in ileal mucosal scrapings in our NECmodel. The anti-RAGE antibody used detects two bands in the45-kDa range representing RAGE protein pre- and postglyco-sylation in mouse lung extract. In addition, this antibodydetects a 25-kDa protein believed to be a proteolytic degrada-tion product. We found that mucosal scrapings of the FFH ratshad a higher expression of RAGE (both the 45- and 25-kDabands) than the BF controls. Similar to the effect on HMGB1in the FFH semapimod group, expression of RAGE wasreduced after the administration of semapimod (Fig. 3C).

Ileal expression of inflammatory proteins in experimentalNEC: effect of semapimod. We examined the expression ofthree members of the Bcl-2 family known to either promote orinhibit apoptosis, namely, Bax, Bad, and Bcl-2, in the ilealmucosa of newborn rats. We found that FFH significantlyupregulate the expression of the apoptotic proteins Bad andBax in mucosal scrapings of FFH animals compared with BFcontrols (Fig. 4). Expression of both proteins in FFH pups wassignificantly decreased when the animals were administeredthe drug semapimod (0.75 mg/kg ip daily). Interestingly, there

Table 1. Time course effect of FFH on incidenceof necrotizing enterocolitis

BF FFH

Day 1 6.6 (91) 22 (109)Day 2 0 (55) 33.3 (84)Day 3 2.1 (94) 40 (127)Day 4 2.3 (390) 48.3 (487)

Data are the percentages of rats in each group with the incidence of intestinaldamage graded between 1 and 3; numbers in parentheses are the total numbersof animals per group. Newborn rats were randomized into two groups:breast-fed (BF) animals that were left with their mothers and formula-fed pupsthat were exposed to hypoxia (FFH) as described in MATERIALS AND METHODS.The terminal ileum of each rat was harvested on the day indicated, fixed, andstained for morphological analysis as described.

G645HMGB1 EXPRESSION AND PROTECTIVE ROLE OF SEMAPIMOD IN NEC


was no significant difference in the expression of Bcl-2 in BFand FFH samples (not shown).

Other proteins whose altered expression have been associ-ated with inflammation in a number of experimental andclinical conditions are iNOS and COX-2. In our NEC model,intestinal samples from BF animals had very little or no iNOSprotein, as shown by Western blot analysis (Fig. 5A). Expres-sion of COX-2, however, was present in those samples (Fig.5B). As expected, elevated levels of iNOS and COX-2 werefound in the terminal ileum from FFH newborn rats comparedwith BF controls. Expression of both proteins was significantlydecreased when FFH animals were administered the drugsemapimod (0.75 mg/kg ip daily; Fig. 5).

Semapimod inhibits activation of p38 MAPK by bacterialLPS in rat intestinal epithelial cells. We have previouslyshown that LPS causes rapid but transient activation of p38MAPK, as determined by increased phosphorylation of p38. In

IEC-6 cells, activating phosphorylation of p38 increases nearly10-fold within 5–10 min of LPS treatment and returns to basallevels 30 min after the addition of LPS (19). To examine theeffect of semapimod on activation of p38 by LPS, IEC-6 cellswere pretreated with different doses of the drug (0.1–10 �M)for 1 h before stimulation with 5 �g/ml LPS for 10 min. Theprotein lysates were then collected, and Western blot analysisfor phospho-p38 and p38 (loading control) was performed asdescribed in MATERIALS AND METHODS. We found that semapi-mod dose dependently inhibited the phosphorylation of p38with a maximal effect at concentrations higher than 1 �M (Fig.6A). We next sought to evaluate the effect of semapimod onp38 activation by inflammatory mediators other than LPS. Asshown in Fig. 6B, a similar exposure of IEC-6 cells to ONOO�

resulted in a rapid phosphorylation of p38, which, in contrast tothe LPS-induced effect, was not inhibited by 1-h pretreatmentwith 5 �M semapimod. Pretreatment with semapimod alsofailed to inhibit the stimulatory effect on p38 activation by amixture of TNF�, IL-1�, and IFN-� (not shown).

HMGB1 is increased in human intestinal specimens frompatients with acute NEC. Intestinal specimens from seveninfants with acute NEC and control intestinal specimens fromseven patients who underwent intestinal resection for inflam-matory conditions other than NEC were analyzed by Westernblot analysis for the presence of HMGB1 protein in twoindependent experiments. The frozen samples (segments fromthe whole distal ileum) were processed as described above, andthe results are shown in Fig. 7. In contrast to the intestinalspecimens from the control patients, ileal specimens from NECpatients demonstrated a twofold increase in the levels ofHMGB1 protein. This finding correlates with upregulation ofiNOS protein (53) and higher levels of nitroso species (51) inresected ileal segments from infants with acute NEC.Fig. 1. High-mobility group box 1 (HMGB1) is elevated in formula-fed

animals exposed to hypoxia (FFH). Newborn rats were randomized into twogroups: breast-fed (BF) animals (which were left with their mothers) and FFHanimals, as described in MATERIALS AND METHODS. The terminal ileum of eachrat was harvested on day 4, and mucosal scrapings were processed for proteinisolation followed by Western blot analysis. Top: representative blot with 3animals/group. Bottom: densitometric analysis for HMGB1 protein (30-kDaband) and low-molecular-mass products (7- to 10-kDa bands) performed asdescribed. Results in graph are means � SE; n 13 animals/group. *P � 0.05vs. BF control [analyzed by one-way ANOVA followed by Fisher’s least-significant difference (LSD) test].

Fig. 2. Administration of semapimod in vivo is protective in an experimentalnecrotizing enterocolitis (NEC) model. Newborn rats separated from theirmothers were fed a conventional formula and exposed to hypoxia after theadministration of the drug semapimod (0.1–1 mg/kg ip) (FFH semapimodgroup) as described in MATERIALS AND METHODS. A segment of the terminalileum of each rat was harvested on day 4 for morphological analysis asdescribed. The incidence of NEC was calculated as the percentage of animalsdisplaying pathology scores higher than 0 as assessed by histology. Results ingraph are means � SE; n 15–45 animals/group. *P � 0.05 vs. FFH withoutsemapimod (analyzed by one-way ANOVA followed by Fisher’s LSD test).



DISCUSSION

NEC is still the most frequent and lethal gastrointestinaldisease of premature infants. Several risk factors for NEC havebeen identified, including low birth weight, formula feeding,bacterial colonization of the gut, and hypoxia (1–3). Severalexperimental animal models have been reported (4, 10, 12, 14,20) that have intestinal hypoxia as the common denominator.To understand the pathogenesis of this disease, we (32) devel-oped a consistent and reproducible rat model that relies onformula feeding and hypoxia, two major risk factors for NEC.In the present study, we presented novel findings regarding themolecular mechanisms associated with the development ofNEC that may shed some light on the pathogenesis of thisdisease.

First, we modified our experimental model (32) so thatnewborn rats were fed with formula twice a day and exposed

to hypoxia for a longer period of time (10 min instead of 3min). Examining a larger number of animals, we observed ahigher mortality rate in the FFH animals compared with the BFgroup. Although malnutrition could be a contributing factor(the volume of formula consumed by the neonatal rats was lessthan the volume of milk consumed by BF pups), there was noapparent cause of death for most of the animals in the FFHgroup. The significant difference in mortality could only beattributed to aspiration during gavage. The longer exposure tohypoxia, however, exacerbated the inflammatory changes seenin the intestine of the FFH group and rendered the neonatal ratsmore susceptible to gut barrier failure as shown by the in-creased incidence of NEC starting at day 3 after birth.

In previous studies, we (32, 33) showed that NEC is char-acterized by the presence of a number of proinflammatorycytokines both in experimental and human NEC. Here, we

Fig. 3. Administration of semapimod in vivodecreases the expression of HMGB1 and itsreceptor for activated glycation end products(RAGE) in ileal samples from FFH animals.Newborn rats were randomized into threegroups: BF animals, FFH animals, andFFH semapimod animals, as described inMATERIALS AND METHODS. The terminal il-eum of each rat was harvested on day 4, andmucosal scrapings were processed for pro-tein isolation followed by Western blot anal-ysis. A: representative blot with 3 animals/group (top) and densitometric analysis forHMGB1 protein (30-kDa band; bottom). Re-sults in graph are means � SE; n 6–13animals/group. *P � 0.001 vs. BF control;#P � 0.05 vs. FFH group (analyzed byone-way ANOVA followed by Fisher’s LSDtest). B: morphological analysis of rat intes-tinal samples in experimental NEC at day 4.B,A: ileal segment from a BF animal show-ing normal histology (the villi are tall andhealthy); B,B: ileal segment from a FFHanimal showing fewer Goblet cells, villuscore separation, and presence of inflamma-tory cells (black arrows); B,C: ileal segmentfrom a FFH animal that received semapimod(0.75 mg/kg ip) and displays similar mor-phology as the BF control. C: representativeblot with 3 animals/group (top) and densito-metric analysis for RAGE protein (45- and25-kDa bands; bottom) performed as de-scribed. Results in graph are means � SE;n 9–24 animals/group. *P 0.001 vs. BFcontrol; #P 0.009 vs. FFH group; **P 0.016 vs. BF control (analyzed by one-wayANOVA followed by Tukey test).



examined the expression of the protein called HMGB1 in thesmall intestine of newborn rats with NEC. HMGB1 is 30-kDanuclear and cytosolic protein member of the HMG proteinsuperfamily that has been widely studied as a transcription aswell as growth factor. HMGB1 is also defined as a cytokine

Fig. 4. Administration of semapimod in vivo decreases the expression of Bcl-2family members Bad and Bax in an experimental NEC model in rats. Newbornrats were randomized into three groups: BF animals, FFH animals, and FFH semapimod animals, as described in MATERIALS AND METHODS. The terminalileum of each rat was harvested on day 4, and mucosal scrapings wereprocessed for protein isolation followed by Western blot analysis. A: repre-sentative blot with 3 animals/group (top) and densitometric analysis for Badprotein (bottom). Results in graph are means � SE; n 17–22 animals/group.*P � 0.001 vs. BF control; #P � 0.001 vs. FFH group (analyzed by one-wayANOVA followed by Tukey test). B: representative blot with 3 animals/group(top) and densitometric analysis for Bax protein (bottom). Results in graph aremeans � SE; n 4 animals/group. *P � 0.005 vs. BF control; #P � 0.005 vs.FFH group (analyzed by one-way ANOVA followed by Fisher’s LSD test).

Fig. 5. Administration of semapimod in vivo decreases the expression ofinducible nitric oxide synthase (iNOS) and cyclooxygenase (COX)-2 in anexperimental NEC model in rats. Newborn rats were randomized into threegroups: BF animals, FFH animals, and FFH semapimod animals, asdescribed in MATERIALS AND METHODS. The terminal ileum of each rat washarvested on day 4, and mucosal scrapings were processed for protein isolationfollowed by Western blot analysis. A: representative blot with 3 animals/group(top) and densitometric analysis for iNOS protein (bottom). Results in graphare means � SE; n 29 animals/group. *P � 0.001 vs. BF control; #P 0.006 vs. FFH group (analyzed by one-way ANOVA followed by Tukey test).B: representative blot with 3 animals/group (top) and densitometric analysis forCOX-2 protein (bottom). Results in graph are means � SE; n 16–27animals/group. *P � 0.001 vs. BF control; #P � 0.05 vs. FFH group (analyzedby one-way ANOVA followed by Tukey test).



because it stimulates proinflammatory responses in monocytes/macrophages, is produced during systemic and local inflam-matory responses, has been identified as a late mediator oflethal systemic inflammation (e.g., endotoxemia and sepsis),and is required for the full expression of inflammation inanimal models of endotoxemia, sepsis, and arthritis (2, 45).However, the expression of HMGB1 in the injured intestineand its role in intestinal inflammation and, in particular, inNEC are not known at present. HMGB1 is produced by nearlyall cell types, but cellular levels vary with development and age(45). As expected, we found a high expression of HMGB1 inthe ileum of BF newborn rats. However, maximal expressionof HMGB1 in the ileal mucosal scrapings of FFH newborn ratsat day 4 coincided with the marked appearance of morpholog-

ical characteristics of NEC. Interestingly, those animals thatshowed moderate to severe intestinal damage as assessed byhistology and also showed increased expression of a lowermolecular mass protein detected with the same anti-HMGB1polyclonal antibody as a 7- to 10-kDa band. There is a largebody of literature on low-molecular-mass (below 30 kDa)HMG proteins, but the role of these lower bands as cytokinesis not known at present. Of course, we cannot rule out thepossibility that they could also be degradation products ofHMGB1. Although beyond the scope of the present work,clearly the identification and study of those bands would beuseful to the understanding of the role of HMG proteins indisease.

It has been shown that HMGB1�/� necrotic cells have agreatly reduced ability to promote inflammation, which provesthat the release of HMGB1 can signal the demise of a cell to itsneighbors (39). On the contrary, apoptotic cells do not releaseHMGB1 even after undergoing secondary necrosis and partialautolysis and thus fail to promote inflammation even if notcleared promptly by phagocytic cells (39). HMGB1 is eitheractively secreted by monocytes/macrophages or passively re-leased from necrotic cells from any tissue (45). Because inexperimental NEC enhanced epithelial apoptosis is an initialevent underlying the gross tissue necrosis of the intestinal wall(11, 25), the elevated levels of HGMB1 found in the ilealmucosal scrapings of FFH newborn rats suggest that there ismore synthesis of HMGB1 protein and/or more translocationfrom the nucleus to cytoplasm of the protein ready to getreleased into the extracellular milieu, where it mediates down-stream inflammatory responses. Such HMGB1-related re-sponses have been reported in endotoxemia, arthritis, and

Fig. 7. HMGB1 protein expression in human NEC. Representative ilealsegments from four newborn patients undergoing bowel resection for NEC andileal specimens from four neonates undergoing intestinal resection for inflam-matory conditions other than NEC (control) were analyzed for the presence ofHMGB1. The frozen intestinal samples were processed, and Western blotanalysis (top) was performed as described in MATERIALS AND METHODS. Resultsin graph (bottom) are means � SD; n 7 patients/group in 2 independentexperiments. *P � 0.05 vs. control (analyzed by Student’s t-test).

Fig. 6. Semapimod decreases the activation of p38 MAPK by LPS in vitro. A:rat intestinal epithelial cells (IEC-6) were pretreated with different doses of thedrug (0.1–10 �M) for 1 h before stimulation with 5 �g/ml LPS for 10 min (A)or were pretreated with semapimod (5 �M) for 1 h before stimulation with 5�g/ml LPS or ONOO� (50 �M) for 10 min (B). The protein lysates were thencollected, and Western blot analysis for phospho-p38 (pp38) and p38 (loadingcontrol) was performed as described in MATERIALS AND METHODS. A: represen-tative blot for pp38/p38 (top) and densitometric analysis showing the ratio ofpp38 to p38 band intensity (bottom). Ctrl, nonstimulated cells. Results in graphare means � SE; n 3 independent experiments. *P � 0.05 vs. LPS-treatedcells without pretreatment with semapimod (analyzed by one-way ANOVAfollowed by Fisher’s LSD test).



sepsis (2). Consistent with our observations in rat NEC, wefound that intestinal segments of infants with acute NECshowed greater expression of HMGB1 than intestinal segmentsfrom non-NEC patients. This finding suggests that an overex-pression of HMGB1may be related to the intestinal damageinduced by FFH in experimental NEC as well as human NEC.

Little is known about how HMGB1 mRNA is transcription-ally regulated. Interestingly, we previously found that amongthe several DNA repair genes that were affected by iNOS inmurine hepatocytes, there was an increased expression of thegene coding for HMG-2 protein as determined by microarrayDNA analysis (49). However, the exact relationship andwhether iNOS-derived NO alters the expression or activity ofHMG proteins and specifically of HMGB1 remain to be inves-tigated. Experiments to address these questions are currentlybeing performed in our laboratory. Furthermore, althoughHMGB1 has not been associated with gene transcription invivo, it can stimulate transcription in vitro (45). Thus wecannot exclude the possibility that elevated levels of HMGB1affect the expression of other cell-death related genes in thisexperimental model of NEC.

We tested the effect of semapimod (formerly known asCNI-1493), a tetravalent guanylhydrazone known to inhibitmouse macrophage arginine transport and NO production andto suppress the release of proinflammatory cytokines (7), in ourexperimental NEC model. In this study, we show that thismacrophage-deactivating drug inhibits the expression and ac-cumulation of HMGB1 in the ileal mucosal scrapings of FFHanimals, thereby limiting the degree of intestinal injury causedby formula feeding and exposure to hypoxia. The fact thatsemapimod did not completely prevent the occurrence ofNEC-like morphological changes in our experimental modelconfirms the complexity of this disease and implies that othermechanism(s) are also involved in its pathogenesis.

Outside the cell, HMGB1 binds with high affinity to RAGE,a transmembrane receptor of the immunoglobulin superfamilythat is in part responsible for the receptor signal transduction ofHMGB1 (24, 44). RAGE is expressed on monocytes/macro-phages, endothelial cells, neurons, and smooth muscle cells (2).The accumulation of RAGE ligands leads to inflammatorydisorders, and the biology of RAGE is driven by the settings inwhich these ligands accumulate, such as diabetes, inflamma-tion, neurodegenerative disorders, and tumors (8). Althoughone study (52) has shown that a human colon adenocarcinomacell line expresses RAGE, the expression of this protein innontransformed intestinal epithelial cells or normal intestinaltissue has not been reported yet. In our rat model, we couldhardly detect any RAGE in ileal mucosal scrapings of BFnewborn rats. However, there was a higher expression ofRAGE in intestinal samples from FFH animals analyzed usingan anti-RAGE antibody that detects two bands in the 45-kDarange (RAGE protein pre- and postglycosylation in mouse lungextract) and a 25-kDa protein believed to be a proteolyticdegradation product. Interestingly, the intensity of the 25-kDaband was significantly higher than that of the 45-kDa band inFFH animals. At present, we do not have an explanation forthis finding, but it is unlikely to be due to total proteindegradation as shown by normal �-actin levels. Interestingly,not only the expression of HMGB1 was significantly reducedby semapimod but also that of its receptor RAGE. This findingsuggests that the upregulation of RAGE is involved in the

intestinal injury caused by FFH. In this respect, a recent study(29) using novel animal models with defective or tissue-specific RAGE expression showed that deletion of RAGEprovides protection from the lethal effects of septic shockcaused by cecal ligation and puncture. It should be noted,however, that, although RAGE has been shown to interact withHMGB1, other putative HMGB1 receptors are known to existbut have not been characterized yet. A recent study (35)showed that, whereas RAGE played only a minor role inmacrophage activation by HMGB1, the interactions ofHMGB1 with Toll-like receptor (TLR)2 and TLR4 couldexplain the ability of HMGB1 to generate inflammatory re-sponses that are similar to those initiated by LPS. Although therole of RAGE and other receptors in mediating inflammatoryresponses to HMGB1 still requires further investigation, ourresults suggest that RAGE accumulation and degradation ac-companies the upregulation of its ligand HMGB1 associatedwith FFH-induced intestinal injury in experimental NEC.

As in many other cell types and tissues, intestinal apoptosisand cell death are the result of different pathways that involvea number of proinflammatory cytokines and mediators otherthan HMGB1. We found that expression of Bax and Bad, twoproapoptotic members of the Bcl-2 family of proteins, weresignificantly upregulated by FFH in mucosal scrapings of FFHanimals compared with BF controls. Expression of both pro-teins in FFH pups was also significantly decreased when theanimals were administered the drug semapimod. Similarly,other proteins like iNOS and COX-2, whose altered expressionhas been associated with inflammation in a number of exper-imental and clinical conditions (for reviews, see Refs. 28, 46,and 50), were elevated in the terminal ileum of FFH newbornrats compared with BF controls and downregulated when theanimals were administered the drug semapimod. Although wedid not assess the integrity of the mucosal barrier in our model,it has previously been shown that HMGB1 is capable ofcausing derangements in intestinal barrier function in culturedCaco-2 human enterocytic monolayers and that this effectdepends on the formation of NO and ONOO� (38). Thus theprotection conferred by semapimod may also be related to itsability to inhibit the FFH-induced increase in intestinal perme-ability by decreasing both HMGB1 and iNOS protein expres-sion.

Because bacterial colonization is an important factor in thepathogenesis of NEC, it is necessary to understand the re-sponse of intestinal epithelial cells to LPS. In our model, thisbecomes even more important because the animals are not keptin a pathogen-free environment. Our results show that LPSactivates p38 MAPK both in IEC-6 rat intestinal epithelial cellsand in ileal mucosal scrapings cultured ex vivo (19). Thisincrease in p38 activation could significantly be inhibited uponpreincubation of IEC-6 cells with semapimod in a concentra-tion-dependent fashion. More importantly, the inhibitory effectof semapimod rather than a nonspecific inhibitory effect seemsto be a selective effect on the LPS signaling pathway. That isillustrated by the fact that semapimod inhibited p38 activationonly in LPS-treated cells but not in cells exposed to ONOO�,the toxic NO-related species that also activates p38. A previousreport (31) has suggested diverse effects of semapimod that aretissue specific and that confer protection against the hemody-namic and inflammatory responses to LPS. Also, semapimodhas previously been shown to inhibit the expression of proin-



flammatory cytokines through a pathway involving MAPKs,specifically to inhibit the phosphorylation and activation of p38MAPK in both human monocytes and the murine macrophagecell line RAW 264.7 (13, 30). Although inflammatory MAPKs,in particular p38 and JNK, are critically involved in thepathogenesis of Crohn’s disease (23), their involvement in theregulation of inflammatory responses in the gut needs to beconclusively demonstrated.

A recently described neural pathway, termed the “cholin-ergic anti-inflammatory pathway”, involves vagus nerve stim-ulation that inhibits the release of TNF, HMGB1, and othercytokines and protects against endotoxemia and ischemia-reperfusion injury (6). By examining the effects of pharmaco-logical and electrical stimulation of the intact vagus nerve inadult male Lewis rats subjected to endotoxin-induced shock, itwas found that intact vagus nerve signaling was required forthe anti-inflammatory action of semapimod (6). In that study,the intracerebroventricular administration of semapimod was100,000-fold more effective in suppressing endotoxin-inducedTNF release and shock compared with the intravenous route. Inour model, intracerebroventricular administration of the drug isextremely difficult due to the small size and sensitivity ofnewborn pups, and thus we could not explore this route ofdosing. The question of whether semapimod can be protectivethrough the cholinergic anti-inflammatory pathway in ourmodel and whether this novel pathway plays a role in thedevelopment of both experimental and human NEC remain tobe investigated.

In summary, our study demonstrates that the proinflamma-tory protein HMGB1 and its receptor RAGE are elevated in theileal mucosa of formula-fed newborn rats exposed to hypoxia.We show that the synthetic guanylhydrazone semapimod, acytokine inhibitor and MAPK blocker, inhibits the upregula-tion of both HMGB1 and RAGE as well as of other proinflam-matory mediators, thereby limiting the intestinal injury causedby FFH in experimental NEC. In addition, this drug selectivelyaffects the activation of p38 MAPK by LPS in IEC-6 intestinalepithelial cells, supporting the involvement of MAPKs in theprotective effect of semapimod against bacterial LPS-inducedintestinal injury. Our results implicate HMGB1 and RAGE asimportant mediators of enterocyte apoptosis/cell death andhypoxia-induced gut barrier failure associated with NEC.Semapimod is currently being developed as a potential treat-ment for Crohn’s disease and other inflammatory pathologies(42). Strategies to suppress the release of proinflammatorycytokines and to inhibit the expression and activity of cytotoxicmolecules such as HMGB1 with inhibitors like semapimodmay offer a novel therapeutic modality to combat inflammatoryconditions such as NEC.

ACKNOWLEDGMENTS

We thank Dr. Vidal de la Cruz (Cytokine Pharmasciences; King of Prussia,PA) for helpful discussion.

GRANTS

This study was supported by National Institute of Allergy and InfectiousDiseases Grant RO1-AI-14032 and by the Benjamin R. Fisher Chair inPediatric Surgery (both to H. R. Ford).

REFERENCES

1. Albanese CT and Rowe MI. Necrotizing enterocolitis. Semin PediatrSurg 4: 200–206, 1995.

2. Andersson U and Tracey KJ. HMGB1 in sepsis. Scand J Infect Dis 35:577–584, 2003.

3. Andersson U, Wang H, Palmblad K, Aveberger AC, Bloom O, Er-landsson-Harris H, Janson A, Kokkola R, Zhang M, Yang H, andTracey KJ. High mobility group 1 protein (HMG-1) stimulates proin-flammatory cytokine synthesis in human monocytes. J Exp Med 192:565–570, 2000.

4. Barlow B and Santulli TV. Importance of multiple episodes of hypoxiaor cold stress on the development of enterocolitis in an animal model.Surgery 77: 687–690, 1975.

5. Barlow B, Santulli TV, Heird WC, Pitt J, Blanc WA, and SchullingerJN. An experimental study of acute neonatal enterocolitis–the importanceof breast milk. J Pediatr Surg 9: 587–595, 1974.

6. Bernik TR, Friedman SG, Ochani M, DiRaimo R, Ulloa L, Yang H,Sudan S, Czura CJ, Ivanova SM, and Tracey KJ. Pharmacologicalstimulation of the cholinergic antiinflammatory pathway. J Exp Med 195:781–788, 2002.

7. Bianchi M, Bloom O, Raabe T, Cohen PS, Chesney J, Sherry B,Schmidtmayerova H, Calandra T, Zhang X, Bukrinsky M, Ulrich P,Cerami A, and Tracey KJ. Suppression of proinflammatory cytokines inmonocytes by a tetravalent guanylhydrazone. J Exp Med 183: 927–936,1996.

8. Bucciarelli LG, Wendt T, Rong L, Lalla E, Hofmann MA, Goova MT,Taguchi A, Yan SF, Yan SD, Stern DM, and Schmidt AM. RAGE is amultiligand receptor of the immunoglobulin superfamily: implications forhomeostasis and chronic disease. Cell Mol Life Sci 59: 1117–1128, 2002.

9. Bustin M. Revised nomenclature for high mobility group (HMG) chro-mosomal proteins. Trends Biochem Sci 26: 152–153, 2001.

10. Caplan MS, Lickerman M, Adler L, Dietsch GN, and Yu A. The roleof recombinant platelet-activating factor acetylhydrolase in a neonatal ratmodel of necrotizing enterocolitis. Pediatr Res 42: 779–783, 1997.

11. Clark JA, Lane RH, Maclennan NK, Holubec H, Dvorakova K,Halpern MD, Williams CS, Payne CM, and Dvorak B. Epidermalgrowth factor reduces intestinal apoptosis in an experimental model ofnecrotizing enterocolitis. Am J Physiol Gastrointest Liver Physiol 288:G755–G762, 2005.

12. Cohen IT, Nelson SD, Moxley RA, Hirsh MP, Counihan TC, andMartin RF. Necrotizing enterocolitis in a neonatal piglet model. J PediatrSurg 26: 598–601, 1991.

13. Cohen PS, Nakshatri H, Dennis J, Caragine T, Bianchi M, Cerami A,and Tracey KJ. CNI-1493 inhibits monocyte/macrophage tumor necrosisfactor by suppression of translation efficiency. Proc Natl Acad Sci USA93: 3967–3971, 1996.

14. Crissinger KD, Burney DL, Velasquez OR, and Gonzalez E. An animalmodel of necrotizing enterocolitis induced by infant formula and ischemiain developing piglets. Gastroenterology 106: 1215–1222, 1994.

15. Dickinson EC, Gorga JC, Garrett M, Tuncer R, Boyle P, Watkins SC,Alber SM, Parizhskaya M, Trucco M, Rowe MI, and Ford HR.Immunoglobulin A supplementation abrogates bacterial translocation andpreserves the architecture of the intestinal epithelium. Surgery 124: 284–290, 1998.

16. Ford HR, Sorrells DL, and Knisely AS. Inflammatory cytokines, nitricoxide, and necrotizing enterocolitis. Semin Pediatr Surg 5: 155–159, 1996.

17. Ford HR, Watkins S, Reblock K, and Rowe M. The role of inflamma-tory cytokines and nitric oxide in the pathogenesis of necrotizing entero-colitis. J Pediatr Surg 32: 275–282, 1997.

18. Go LL, Albanese CT, Watkins SC, Simmons RL, and Rowe MI. Breastmilk protects the neonate from bacterial translocation. J Pediatr Surg 29:1059–1063, 1994.

19. Grishin A, Wang J, Hackam D, Qureshi F, Upperman J, Zamora R,and Ford HR. p38 MAP kinase mediates endotoxin-induced expressionof cyclooxygenase-2 in enterocytes. Surgery 136: 329–335, 2004.

20. Halpern MD, Holubec H, Dominguez JA, Williams CS, Meza YG,McWilliam DL, Payne CM, McCuskey RS, Besselsen DG, and DvorakB. Up-regulation of IL-18 and IL-12 in the ileum of neonatal rats withnecrotizing enterocolitis. Pediatr Res 51: 733–739, 2002.

21. Harris MC, Costarino AT Jr, Sullivan JS, Dulkerian S, McCawley L,Corcoran L, Butler S, and Kilpatrick L. Cytokine elevations in criti-cally ill infants with sepsis and necrotizing enterocolitis. J Pediatr 124:105–111, 1994.

22. Holman RC, Stehr-Green JK, and Zelasky MT. Necrotizing enteroco-litis mortality in the United States, 1979–85. Am J Public Health 79:987–989, 1989.



23. Hommes D, van den BB, Plasse T, Bartelsman J, Xu C, MacphersonB, Tytgat G, Peppelenbosch M, and Van Deventer S. Inhibition ofstress-activated MAP kinases induces clinical improvement in moderate tosevere Crohn’s disease. Gastroenterology 122: 7–14, 2002.

24. Hori O, Brett J, Slattery T, Cao R, Zhang J, Chen JX, Nagashima M,Lundh ER, Vijay S, Nitecki and D. The receptor for advanced glycationend products (RAGE) is a cellular binding site for amphoterin. Mediationof neurite outgrowth and co-expression of RAGE and amphoterin in thedeveloping nervous system. J Biol Chem 270: 25752–25761, 1995.

25. Jilling T, Lu J, Jackson M, and Caplan MS. Intestinal epithelialapoptosis initiates gross bowel necrosis in an experimental rat model ofneonatal necrotizing enterocolitis. Pediatr Res 55: 622–629, 2004.

26. Kliegman RM. Models of the pathogenesis of necrotizing enterocolitis.J Pediatr 117: S2–S5, 1990.

27. Kliegman RM and Fanaroff AA. Necrotizing enterocolitis. N EnglJ Med 310: 1093–1103, 1984.

28. Kubes P and McCafferty DM. Nitric oxide and intestinal inflammation.Am J Med 109: 150–158, 2000.

29. Liliensiek B, Weigand MA, Bierhaus A, Nicklas W, Kasper M, HoferS, Plachky J, Grone HJ, Kurschus FC, Schmidt AM, Yan SD, MartinE, Schleicher E, Stern DM, Hammerling GG, Nawroth PP, andArnold B. Receptor for advanced glycation end products (RAGE) regu-lates sepsis but not the adaptive immune response. J Clin Invest 113:1641–1650, 2004.

30. Milligan ED, O’Connor KA, Armstrong CB, Hansen MK, Martin D,Tracey KJ, Maier SF, and Watkins LR. Systemic administration ofCNI-1493, a p38 mitogen-activated protein kinase inhibitor, blocks intra-thecal human immunodeficiency virus-1 gp120-induced enhanced painstates in rats. J Pain 2: 326–333, 2001.

31. Molina PE, Qian L, Schuhlein D, Naukam R, Wang H, Tracey KJ,and Abumrad NN. CNI-1493 attenuates hemodynamic and pro-inflam-matory responses to LPS. Shock 10: 329–334, 1998.

32. Nadler EP, Dickinson E, Knisely A, Zhang XR, Boyle P, Beer-Stolz D,Watkins SC, and Ford HR. Expression of inducible nitric oxide synthaseand interleukin-12 in experimental necrotizing enterocolitis. J Surg Res92: 71–77, 2000.

33. Nadler EP, Stanford A, Zhang XR, Schall LC, Alber SM, Watkins SC,and Ford HR. Intestinal cytokine gene expression in infants with acutenecrotizing enterocolitis: interleukin-11 mRNA expression inversely cor-relates with extent of disease. J Pediatr Surg 36: 1122–1129, 2001.

34. Ochoa JB, Udekwu AO, Billiar TR, Curran RD, Cerra FB, SimmonsRL, and Peitzman AB. Nitrogen oxide levels in patients after trauma andduring sepsis. Ann Surg 214: 621–626, 1991.

35. Park JS, Svetkauskaite D, He Q, Kim JY, Strassheim D, Ishizaka A,and Abraham E. Involvement of toll-like receptors 2 and 4 in cellularactivation by high mobility group box 1 protein. J Biol Chem 279:7370–7377, 2004.

36. Potoka DA, Upperman JS, Zhang XR, Kaplan JR, Corey SJ, GrishinA, Zamora R, and Ford HR. Peroxynitrite inhibits enterocyte prolifer-ation and modulates Src kinase activity in vitro. Am J Physiol GastrointestLiver Physiol 285: G861–G869, 2003.

37. Rowe MI, Reblock KK, Kurkchubasche AG, and Healey PJ. Necro-tizing enterocolitis in the extremely low birth weight infant. J Pediatr Surg29: 987–990, 1994.

38. Sappington PL, Yang R, Yang H, Tracey KJ, Delude RL, and FinkMP. HMGB1 B box increases the permeability of Caco-2 enterocyticmonolayers and impairs intestinal barrier function in mice. Gastroenter-ology 123: 790–802, 2002.

39. Scaffidi P, Misteli T, and Bianchi ME. Release of chromatin proteinHMGB1 by necrotic cells triggers inflammation. Nature 418: 191–195,2002.

40. Scheifele DW. Role of bacterial toxins in neonatal necrotizing enteroco-litis. J Pediatr 117: S44–S46, 1990.

41. Shi Y, Li HQ, Shen CK, Wang JH, Qin SW, Liu R, and Pan J. Plasmanitric oxide levels in newborn infants with sepsis. J Pediatr 123: 435–438,1993.

42. Sitaraman SV, Hoteit M, and Gewirtz AT. Semapimod. Cytokine. CurrOpin Investig Drugs 4: 1363–1368, 2003.

43. Sorrells DL, Friend C, Koltuksuz U, Courcoulas A, Boyle P, GarrettM, Watkins S, Rowe MI, and Ford HR. Inhibition of nitric oxide withaminoguanidine reduces bacterial translocation after endotoxin challengein vivo. Arch Surg 131: 1155–1163, 1996.

44. Wang H, Bloom O, Zhang M, Vishnubhakat JM, Ombrellino M, CheJ, Frazier A, Yang H, Ivanova S, Borovikova L, Manogue KR, FaistE, Abraham E, Andersson J, Andersson U, Molina PE, Abumrad NN,Sama A, and Tracey KJ. HMG-1 as a late mediator of endotoxin lethalityin mice. Science 285: 248–251, 1999.

45. Wang H, Yang H, and Tracey KJ. Extracellular role of HMGB1 ininflammation and sepsis. J Intern Med 255: 320–331, 2004.

46. Warner TD and Mitchell JA. Cyclooxygenases: new forms, new inhib-itors, and lessons from the clinic. FASEB J 18: 790–804, 2004.

47. Wilson R, Kanto WP Jr, McCarthy BJ, and Feldman RA. Age at onsetof necrotizing enterocolitis. Risk factors in small infants. Am J Dis Child136: 814–816, 1982.

48. Yang H, Wang H, and Tracey KJ. HMG-1 rediscovered as a cytokine.Shock 15: 247–253, 2001.

49. Zamora R, Vodovotz Y, Aulak KS, Kim PK, Kane JM, III, Alarcon L,Stuehr DJ, and Billiar TR. A DNA microarray study of nitric oxide-induced genes in mouse hepatocytes: implications for hepatic hemeoxygenase-1 expression in ischemia/reperfusion. Nitric Oxide 7: 165–186,2002.

50. Zamora R, Vodovotz Y, and Billiar TR. Inducible nitric oxide synthaseand inflammatory diseases. Mol Med 6: 347–373, 2000.

51. Zamora R, Bryan NS, Boyle P, Wong C, Milsom AB, Jaffe R, FeelischM, and Ford HR. Nitrosative stress in an animal model of necrotizingenterocolitis. Free Rad Biol Med. In press.

52. Zill H, Gunther R, Erbersdobler HF, Folsch UR, and Faist V. RAGEexpression and AGE-induced MAP kinase activation in Caco-2 cells.Biochem Biophys Res Commun 288: 1108–1111, 2001.

53. Zuckerbraun BS, Otterbein LE, Boyle P, Jaffe R, Upperman J,Zamora R, and Ford HR. Carbon monoxide protects against the develop-ment of experimental necrotizing enterocolitis. Am J Physiol GastrointestLiver Physiol. First published May 12, 2005; doi:10.1152/ajpgi.00055.2005.



high mobility group box-1 protein is an inflammatory mediator in necrotizing enterocolitis (nec):...

Documents