Reduced Paneth cell alpha-defensins in ileal Crohn's disease.

Abstract

The pathogenesis of Crohn's disease (CD), an idiopathic inflammatory bowel disease, is attributed, in part, to intestinal bacteria that may initiate and perpetuate mucosal inflammation in genetically susceptible individuals. Paneth cells (PC) are the major source of antimicrobial peptides in the small intestine, including human alpha-defensins HD5 and HD6. We tested the hypothesis that reduced expression of PC alpha-defensins compromises mucosal host defenses and predisposes patients to CD of the ileum. We report that patients with CD of the ileum have reduced antibacterial activity in their intestinal mucosal extracts. These specimens also showed decreased expression of PC alpha-defensins, whereas the expression of eight other PC products either remained unchanged or increased when compared with controls. The specific decrease of alpha-defensins was independent of the degree of inflammation in the specimens and was not observed in either CD of the colon, ulcerative colitis, or pouchitis. The functional consequence of alpha-defensin expression levels was examined by using a transgenic mouse model, where we found changes in HD5 expression levels, comparable to those observed in CD, had a pronounced impact on the luminal microbiota. Thus, the specific deficiency of PC defensins that characterizes ileal CD may compromise innate immune defenses of the ileal mucosa and initiate and/or perpetuate this disease.

Full text

Page 1

Reduced Paneth cell ?-defensins in ileal
Crohn’s disease
Jan Wehkamp*, Nita H. Salzman†, Edith Porter‡§, Sabine Nuding¶?, Michael Weichenthal**, Robert E. Petras††,
Bo Shen‡‡, Elke Schaeffeler?, Matthias Schwab?, Rose Linzmeier§, Ryan W. Feathers*, Hiutung Chu*,
Heriberto Lima, Jr.‡, Klaus Fellermann¶?, Tomas Ganz§, Eduard F. Stange¶?§§, and Charles L. Bevins*§§
*Department of Microbiology and Immunology, School of Medicine, University of California, Davis, CA 95616;†Department of Pediatrics, Division
of Gastroenterology, Medical College of Wisconsin, Milwaukee, WI 53226;‡Department of Biological Sciences, California State University,
Los Angeles, CA 90032;§Department of Medicine, David Geffen School of Medicine, University of California, Los Angeles, CA 90095;¶Department
of Internal Medicine I, Robert Bosch Hospital, 70376 Stuttgart, Germany;?Dr. Margarete Fischer-Bosch Institute of Clinical Pharmacology,
70376 Stuttgart, Germany; **Department of Dermatology, University of Kiel, 24105 Kiel, Germany;††Gastrointestinal Pathology Services, AmeriPath,
Inc., Department of Pathology, Northeastern Ohio Universities College of Medicine, Rootstown, OH 44272; and‡‡Department of Gastroenterology,
Cleveland Clinic Foundation, Cleveland, OH 44195
Edited by Richard A. Flavell, Yale University School of Medicine, New Haven, CT, and approved October 24, 2005 (received for review June 24, 2005)
The pathogenesis of Crohn?s disease (CD), an idiopathic inflamma-
tory bowel disease, is attributed, in part, to intestinal bacteria that
may initiate and perpetuate mucosal inflammation in genetically
susceptible individuals. Paneth cells (PC) are the major source of
antimicrobial peptides in the small intestine, including human
?-defensins HD5 and HD6. We tested the hypothesis that reduced
expression of PC ?-defensins compromises mucosal host defenses
and predisposes patients to CD of the ileum. We report that
patients with CD of the ileum have reduced antibacterial activity in
their intestinal mucosal extracts. These specimens also showed
decreased expression of PC ?-defensins, whereas the expression of
eight other PC products either remained unchanged or increased
when compared with controls. The specific decrease of ?-defensins
was independent of the degree of inflammation in the specimens
and was not observed in either CD of the colon, ulcerative colitis,
or pouchitis. The functional consequence of ?-defensin expression
levelswasexaminedbyusingatransgenicmousemodel,wherewe
found changes in HD5 expression levels, comparable to those
observed in CD, had a pronounced impact on the luminal micro-
biota. Thus, the specific deficiency of PC defensins that character-
izes ileal CD may compromise innate immune defenses of the ileal
mucosa and initiate and?or perpetuate this disease.
innate immunity ? intestine ? bacteria ? inflammatory bowel disease
I
disease (CD) and ulcerative colitis (UC), based on clinical features
and histopathology. While inflammation in UC is typically re-
stricted to the colon, that of CD occurs at many sites, most
commonly in the ileum of the small intestine and in the colon (1).
In all cases, intestinal microbiota are thought to trigger the disease
in genetically susceptible individuals (2). However, variations in
both inherited susceptibility and clinical phenotypes suggest that
neither UC nor CD is a homogeneous disorder (1, 3, 4).
Multiple lines of evidence support that genetic factors signifi-
cantly contribute to the pathogenesis of IBD (2), and numerous
genome-wide screens have identified several susceptibility loci,
including those referred to as IBD1–8. A seminal advance came
with characterization of IBD1, which revealed that approximately
one-third of CD patients have loss-of-function mutations in
CARD15, which encodes the nucleotide-binding oligomerization
domain 2 (NOD2; for review, see ref. 2). NOD2?CARD15 muta-
tions are especially associated with localization of CD in the small
intestine (for reviews, see refs. 3 and 4). NOD2 is an intracellular
receptor for muramyl dipeptide, a component of bacterial pepti-
doglycan (5, 6), consistent with the notion that defective responses
to luminal bacteria are important in CD pathogenesis.
NOD2 expression has been found in monocytes, monocyte-
derivedcells,andepithelialcells(5,7).Inmonocyticcells,muramyl
nflammatory bowel disease (IBD) is a chronic inflammation of
the intestine often grouped into two major entities, Crohn’s
dipeptide activation of NOD2 leads to NF-?B-dependent proin-
flammatorycytokineexpression(5),andthispathwayisattenuated
in blood monocytes from CD patients with NOD2 mutations (6).
In the intestinal mucosa, NOD2 is predominantly expressed in
Paneth cells (PCs) (7), which are secretory epithelial cells of the
small intestinal crypts. PCs are known to synthesize and secrete
several antimicrobial peptides, including lysozyme, secretory phos-
pholipase A2(sPLA2), and human ?-defensins 5 and 6 (HD5 and
HD6) (8–10), but the function of NOD2 in these cells is unclear.
Recent studies of NOD2-deficient mice show both a reduced
expression in PCs of human ?-defensin homologues (called crypt-
dins in mice) and increased susceptibility to ingested Listeria
monocytogenes, a Gram-positive bacterial pathogen (11). It should
be noted that expression of all cryptdins is not equally affected by
the NOD2 deficiency in this study, consistent with the complexity
of PC gene regulation that is the focus of much current study
(12–14).
In addition to genetic factors in IBD, numerous studies have
implicated a key role of the intestinal microbiota in disease patho-
genesis, both in patients with IBD (15–20) and in rodent models of
IBD(forreviews,seerefs.21and22).Althoughaspecificcausative
pathogen seems unlikely, it appears that intestinal microbes nor-
mally present as commensal microbiota may trigger disease devel-
opment in genetically susceptible hosts (23). The contribution of
luminal microbes to the pathogenesis of IBD is highlighted by
reportsthatsurgicaldiversionofthefecalstreameffectivelyresolves
CD inflammation distal to the surgical site (15), and that in some
cases antibiotics ameliorate IBD (20). In addition, the intestinal
mucosaofCDpatientsiscoveredbyadherentstrainsofEscherichia
coli and other bacteria from the lumen, whereas these bacteria are
absent from the normal small bowel mucosa (16, 18, 19). Further-
more, a loss in immunological tolerance toward luminal microbiota
is observed in patients with CD (17). Thus, IBD pathogenesis likely
results from a breach in effective mucosal barrier functions to
constituents of the commensal microbiota (24).
PCs are key contributors to mucosal innate immunity in the
small intestine by secretion of antimicrobial peptides including
?-defensins, cationic peptides with a broad spectrum of antimicro-
bial activity (25–27). PC granules are released into the intestinal
lumenuponstimulationwithbacterialproducts,includingmuramyl
Conflict of interest statement: No conflicts declared.
This paper was submitted directly (Track II) to the PNAS office.
Freely available online through the PNAS open access option.
Abbreviations:IBD,inflammatoryboweldisease;CD,Crohn’sdisease;UC,ulcerativecolitis;
NOD2, nucleotide-binding oligomerization domain 2; PC, Paneth cell; sPLA2, secretory
phospholipase A2; HD5?6, human ?-defensins 5?6; TG, transgenic.
§§To whom correspondence may be addressed. E-mail: eduard.stange@rbk.de or
clbevins@ucdavis.edu.
© 2005 by The National Academy of Sciences of the USA
www.pnas.org?cgi?doi?10.1073?pnas.0505256102PNAS ?
December 13, 2005 ?
vol. 102 ?
no. 50 ?
18129–18134
MEDICAL SCIENCES

Page 2

dipeptide and lipopolysaccharide (28). Two murine models have
underscored the functional importance of PC ?-defensins in innate
immunity. Mice rendered deficient in their ability to process PC
?-defensinsprecursorsdonotproducemature?-defensinspeptides
and are highly susceptible to challenges with orally administered
bacterial pathogens (29). Similarly, a protection against a normally
lethal Salmonella infection in HD5 transgenic (TG) mice demon-
strates that this peptide has a biologically significant effect on
pathogenic microbes in the intestinal lumen (30).
Inviewoftheprominentroleofdefensinsasmucosalantibiotics,
the highly abundant expression of ?-defensins in PCs, the mucosal
expressionoftheCDsusceptibilitygeneproductNOD2inPCs,and
thelikelyroleofbacteriainIBDpathogenesis,wehypothesizedthat
?-defensinsmayplayacentralroleinCDpathophysiology(10,31).
A deficiency in the innate immune defense provided by defensins
might allow bacteria to adhere to the CD mucosa and trigger an
inflammatory response. When we initially tested this hypothesis by
comparing PC ?-defensin mRNA in ileal biopsies from controls
with those from CD patients, but without regard for location of
disease, no statistical differences in levels were observed (32).
However,segregationoftheCDspecimensaccordingtotheclinical
phenotype of either ileal or colonic involvement revealed a signif-
icant difference in defensin mRNA expression, such that only
patients with ileal involvement, especially those harboring NOD2?
CARD15 mutations, had decreased ileal levels of HD5 and HD6
mRNA (32).
Herein, we report a decrease in antimicrobial activity and a
specific reduction of PC ?-defensins peptides in CD of the ileum.
The decrease of PC ?-defensins could not be attributed simply to
a nonspecific response to inflammation in ileum and was not
observed in ileal mucosa of patients with either CD limited to the
colon (Crohn’s colitis) or UC. We established a functional conse-
quence of changes in PC ?-defensin expression by examining the
compositionofinherentmicrobiotainaHD5TGmousemodeland
observedchangesinthemicrobiotaattributabletoHD5expression
levels. We propose that a PC ?-defensin deficiency may be a key
factor in the pathogenesis of ileal CD through its compromise of
innate immunity. This view of disease pathophysiology would
provide a justification for seeking alternative therapeutic strategies
(10,20,33)aimedtobolsterprotectiveinnateimmunemechanisms
and restore the host–microbe balance at the intestinal mucosa.
Methods
Patients and Patient Material. Surgical specimens of ileal mucosa
and pouch biopsies were obtained at the Cleveland Clinic Foun-
dation. Ileal biopsies were obtained and processed at the Robert
Bosch Hospital. The protocols were approved by the respective
Institutional Review Boards at these locations. The diagnosis at
both institutions was based on standard criteria using clinical,
radiological, endoscopic, and histopathological findings (34). Ex-
clusion criteria included the diagnoses of backwash ileitis, indeter-
minate colitis, concurrent cytomegalovirus or Clostridium difficile
infection, CD of the pouch, chronic pouchitis, and nonsteroidal
antiinflammatory drug-induced pouchitis.
Supporting Information.FurtherdetailsareprovidedinSupporting
Text, Tables 1–4, and Fig. 5, which are published as supporting
information on the PNAS web site.
Real-Time PCR. Real-time PCR was performed by using single-
stranded cDNA from tissue (or gene-specific plasmids as con-
trols) with specific oligonucleotide primer pairs (Table 2) in a
temperature cycler equipped with a fluorescence detection
monitor (LightCycler, Roche Diagnostics, Mannheim, Ger-
many), as described (32).
Protein Isolation and Immunoblot Analysis. Protein extracts from
ileal mucosa were isolated from randomly selected controls, CD
NOD2?CARD15 wild-type and CD NOD2?CARD15 SNP13 pa-
tients, as described (35). Protein expression of HD5, ?-1-trypsin
inhibitor, lysozyme, and sPLA2in the patient samples was quanti-
fied by immunoblotting, as described (36).
Antimicrobial Activity in Ileal Mucosal Biopsies. Cationic proteins
from ileal mucosal biopsies were isolated by using a weak cation
exchange matrix, as described (35). Assays were normalized to
protein concentration, as determined by Bradford assay. Midloga-
rithmic growth phase suspensions of E. coli (American Type
Culture Collection 25922) and Staphylococcus aureus (American
Type Culture Collection 25923) were incubated with the cationic
protein fraction at 37°C in a final volume of 100 ?l of 1:6 diluted
Schaedler Broth (BD Biosciences, Sparks, CA) (37). Bacterial
suspensions incubated with vehicle (0.01% acetic acid) served as
negative controls. After 120 min, bis-(1,3-dibutylbarbituric acid)t-
rimethine oxonol Molecular Probes), a dye sensitive to membrane
potential,wasaddedataconcentrationof1?g?ml.Bacterialpellets
were isolated by centrifugation, resuspended in 300 ?l of FACS-
Flow (BD Biosciences) and analyzed by flow cytometry by using a
FACSCalibur (BD Biosciences). A total of 30,000 events were
analyzedineachsample.Theantimicrobialactivitywasdetermined
as percentage of depolarized bacteria compared with untreated
controls (37).
Histologic Analyses. HD5 immunohistochemistry and histologic
staining were performed in parallel sections. Immunohistochem-
istry on ileal tissue was performed as described (38) and phloxine
tartrazinehistologicstainingasdescribed(39).Hematoxylin?eosin-
stained paraffin sections from CD and non-IBD controls were
blindly scored for inflammation by a gastrointestinal pathologist
(R.E.P.).
Analysis of HD5 TG Mice. Bacterial microbiota was examined in an
HD5 TG mouse model (30). All animal studies were approved by
theinvolvedInstitutionalAnimalCareandUseCommittees.Insitu
hybridization on representative samples of HD5 TG mice and
human small intestine was performed as described (30). Intestinal
bacteria from the mouse intestine were fixed and analyzed as
described (40). Briefly, an aliquot of fixed bacteria was hybridized
to a Texas red-labeled Bact338 oligonucleotide probe (an oligonu-
cleotide sequence common to all bacteria). The bacteria were then
washed and mounted for viewing under oil by using an epifluores-
cencephotomicroscope.Fluorescentimageswerecapturedbyusing
METAMORPH software (Universal Imaging, Downingtown, PA).
Statistics. All statistical analyses of quantitative RT-PCR grouped
data were performed nonparametrically by using the U test of
Wilcoxon,Mann,andWhitney.ValuesofP?0.05wereconsidered
statistically significant. For illustration, mean values are presented
together with their standard error. HD5 protein expression data
were subjected to t test and ANOVA analysis by using SIGMASTAT
software, Ver. 2.0 (SPSS, Chicago).
Results
We measured PC antimicrobials and other PC products in the ileal
mucosa of four groups: controls, UC patients, CD patients with
solely colonic disease (colitis), and CD patients with ileal disease
(ileitis).PCsarelocatedatthebaseofthecryptsofLieberku ¨hn(Fig.
1A), and cross sections of these crypts show prominent eosinophilic
PC granules (Fig. 1B), known to be rich in antimicrobial peptides
(9), including HD5 (Fig. 1B). In surgical resection specimens from
CD ileitis patients, the ileal expression of HD5 (Fig. 1C) and HD6
mRNA (Fig. 5) was significantly reduced compared with non-IBD
controls. Similar decreases were observed in endoscopic biopsy
specimensfromCDileitispatientsversuscontrols(datanotshown).
In contrast, PC defensin expression in ileal biopsies was unchanged
in patients with either colonic CD (32) or UC (E.F.S., unpublished
18130 ?
www.pnas.org?cgi?doi?10.1073?pnas.0505256102 Wehkamp et al.

Page 3

data). The ileal mRNA expression of two other PC antimicrobials
(lysozyme and sPLA2) and five other PC products [?-1-antitrypsin,
pancreatic secretory trypsin inhibitor, hepatoma-specific protein?
pancreatitis associated protein, trypsin 2 (anionic trypsin), and
trypsin 3 (mesotrypsin)] was not significantly changed in CD ileitis
patients versus controls (Fig. 5), whereas pancreatic stone protein
was increased (Fig. 5).
Because CD patients with NOD2?CARD15 mutations are pre-
disposed to ileal involvement (3, 4), we analyzed the expression of
the ileal PC products in samples with different NOD2?CARD15
genotypes at three loci (SNP8, SNP12, and SNP13). The ileal levels
of HD5 mRNA were comparably low in ileal CD either with SNP8,
SNP12, or wild-type NOD2?CARD15 genotypes as compared with
non-IBDcontrols(Fig.1Cvs.D),withaconsistent3-foldreduction
in each group. In contrast, patients with NOD2?CARD15 SNP13
mutations had further reduced expression levels of HD5 mRNA
versus the other genotypes (P ? 0.03), resulting in a 9-fold differ-
ence compared with non-IBD controls (P ? 0.0005). HD6 expres-
sion in CD ileitis was similarly reduced to ?5-fold lower than
non-IBD controls (P ? 0.018), but the difference within the CD
group (between wild-type and SNP13 NOD2?CARD15 genotypes)
was not statistically significant (P ? 0.15). None of the other PC
products showed significant differences in expression levels when
comparing subgroups of NOD2?CARD15 genotypes (Table 3).
Compared with non-IBD controls, Western blot analysis of
mucosal tissue extracts (Fig. 1E) showed less HD5 peptide in
patients with ileal CD (P ? 0.038). Along with reductions of HD5
mRNA (Fig. 1 C and D), tissue concentrations of HD5 peptide are
reduced in ileal CD with wild-type NOD2?CARD15, but even
further diminished in case of a SNP13 mutation (Fig. 1 E and F).
In contrast to HD5, Western blot analysis of ?-1-antitrypsin,
lysozyme, and sPLA2showed no decrease in the ileal CD samples
as compared with non-IBD controls (Fig. 1G). Immunohistochem-
istryinninerepresentativesamples(threecontrolsandsixilealCD)
identified PCs as the source of HD5 (as shown in Fig. 1B).
To test whether the observed decrease in PC ?-defensins is a
direct consequence of inflammation, we examined HD5 and HD6
expression levels with respect to mucosal inflammation. Histologic
Fig. 1.
(illustration by David R. Schumick Illustration, Elyria, OH). (B) Phloxine tartrazine staining of small intestinal ileal mucosa (Left), showing antimicrobial
peptide-richgranules.ImmunohistochemicallocalizationofthePC?-defensinHD5(Right).(Upper)Normalcontrol;(Lower)CD.(Scalebars,25?m.)(C)Expression
of HD5 mRNA in surgical specimens from controls and patients with ileal CD. The mRNA copy number per 10 ng of total RNA was determined with quantitative
real-time RT-PCR using external standards. [Scale bars represent means (?standard error).] The significance values are based on the Mann–Whitney test (*, P ?
0.05). (D) Expression of HD5 in patients with ileal CD with respect to NOD2?CARD15 genotype. Note that compared with non-IBD controls, all NOD2?CARD15
genotypesubgroups(wildtypeandmutated)ofilealCDpresentedherehavesignificantlyreducedHD5mRNAlevels,asshowninC.Dataexpressedandanalyzed
as in C. (E) (Upper) Coomassie blue-stained SDS gel containing protein extracts (12 ?g per lane) from ileal mucosa of controls (lanes 1–3) and ileal CD patients
(lanes 4–9). NOD2?CARD15 genotype analysis detected SNP-13 mutation in ileal CD patients (lanes 7–9); wild-type sequence was found in other samples (lanes
1–6).StandardsarerecombinantproHD5(openarrow).(Lower)ImmunoblotanalysisofHD5peptideinilealtissuesamples.BlotusingHD5antibody(36,39)was
from replicate gel as in A with 0.6 ?g per lane protein loading. The difference in mean values between controls (lanes 1–3) and ileal CD specimens (lanes 4–9)
was significant by t test analysis (P ? 0.038). (F) Quantification of HD5 peptide in ileal tissue samples. Scale bars represent the percentage (?standard error) of
HD5 peptide amounts in CD specimens as compared with nondisease control samples, which was set as 100%. HD5 peptide concentrations in ileal samples were
determined by quantitative comparison of immunoblot bands from tissue to serial dilutions of recombinant HD5 peptide on the same gel?membrane (data not
shown).SignalwasquantifiedbyusingaVersaDoc1000BioRadimagingsystem.(G)ImmunoblotblotanalysisofsPLA2,lysozyme,and?-1-antiprotease.(Upper)
CoomassiebluestainingofanSDS-TricinePAGEgel(12?g?lane),asdescribedinE.(Lower)AnalysisfromreplicategelasinEwith1.2?gperlaneproteinloading
with sPLA2, lysozyme, and ?-1-antiprotease antibodies.
PC ?-defensins in controls and IBD patients. (A) Illustration of the position of PCs at the base of the crypt of Lieberku ¨hn in the small intestinal mucosa
Wehkamp et al.PNAS ?
December 13, 2005 ?
vol. 102 ?
no. 50 ?
18131
MEDICAL SCIENCES

Page 4

assessment of the degree of inflammation by a gastrointestinal
pathologist(R.E.P.)showedclearcorrelationwiththeexpressionof
the proinflammatory cytokine IL-8 (Fig. 2A), an indicator of
mucosal inflammation. As compared with non-IBD controls, the
levels of HD5 mRNA were similarly decreased in all patients with
ileal CD, irrespective of whether the mucosal inflammation was
absent, moderate, or severe (Fig. 2B). Similarly, HD6 showed the
same pattern of decrease, independent of inflammation status
(data not shown). We next examined the concentrations of PC
?-defensin mRNA in mucosal biopsies of ileal pouches, which are
ileal stool reservoirs surgically constructed after the total removal
of the colon. The ileal pouch mucosa occasionally becomes in-
flamed,aconditioncalledpouchitis(41).PC?-defensinmRNAdid
not show significant changes in pouchitis specimens compared with
those from noninflamed pouches (Fig. 2 C and D), thus showing
that this inflammation does not affect expression levels.
In ileal CD compared with non-IBD controls, we found reduced
total antimicrobial activity against E. coli and S. aureus in ileal
mucosal biopsies (Fig. 3A). The magnitude of the reduced activity
paralleled the lower levels of PC antimicrobial peptide expression
in CD ileitis (Fig. 3B Left), which is principally attributable to the
reduction in HD5 expression. We were not able to unequivocally
attribute the reduced antimicrobial activity in mucosal biopsies
observed in these assays to HD5 content, however, because cur-
rently available antibodies do not neutralize HD5 activity (data not
shown).
In view of recently reported locus polymorphisms that result in
variable numbers of defensin genes per genome (42, 43), we sought
to determine whether the numbers of HD5 or HD6 genes were
reduced in any of 20 CD patients (NOD2?CARD15 wild-type, n ?
10; and NOD2?CARD15 SNP13 mutation, n ? 10), who showed
diminished defensin levels. The analysis showed two gene copies
per diploid genome for both ?-defensins in all tested CD patients,
identical to what was seen in non-IBD controls (n ? 10) (data not
shown).
To study possible in vivo consequences of the changes in PC
?-defensin expression levels we observed in ileal CD, we turned to
our recently described murine HD5 TG model (30). We reasoned
that by comparing heterozygous and homozygous HD5 TG mice,
wecouldtestwhethera2-folddifferenceinHD5expression,similar
to that observed in Crohn’s ileitis versus controls (?3-fold), had a
biological effect on luminal microbes. The TG mice express HD5
inmousePCs(Fig.4A)atlevelscomparabletothoseofhumanPCs
(Fig.4B),whilenotappearingtoaltertheexpressionofendogenous
mouse PC defensins (30). Luminal microbiota obtained from the
small intestines of wild-type, heterozygous, and homozygous HD5
TG mice were labeled with a fluorescent oligonucleotide probe
designed to hybridize to all known bacterial species (TR-Bact 338,
ref. 40). Morphologically, we observed a graded transition in the
composition of bacterial microbiota, from predominantly small
bacilli and cocci in the wild-type mice, to a mixed population of
bacilli and fusiform bacterial species in the heterozygous TG mice,
and finally a population of predominantly fusiform bacteria in the
homozygous TG mice (Fig. 4C). These findings, together with
unpublished studies (N.H.S., unpublished observations), suggest
that the composition of the commensal bacterial population in the
Fig. 2.
Correlation of mucosal inflammation and the proinflammatory cytokine IL-8
mRNA levels in ileal CD and non-IBD controls. Hematoxylin?eosin-stained
paraffin sections from specimens were assessed for mucosal inflammation by
agastrointestinalpathologistwhowasuninformedaboutsamples[noinflam-
mation (0), moderate (1), and severe (2)]. IL-8 mRNA expression levels, ex-
pressedasmRNAcopiesper10ngofRNA[means(?standarderror)shownfor
each group]. Inset shows the average inflammation score for controls (n ? 8)
and ileal CD patients (n ? 25). (B) Expression of HD5 mRNA in specimens from
non-IBD controls and patients with ileal CD grouped as in A. Data expressed
as mRNA copy number per 10 ng RNA (means ? standard error), from same
specimens as analyzed in Fig. 1C. (C) Expression of IL-8 mRNA in specimens
from normal and inflamed ileal pouches. Biopsy specimens were from normal
(n ? 7) and inflamed pouch mucosa (pouchitis, n ? 7). IL-8 mRNA expression
levels are expressed as in A. The significance values based on the Mann–
Whitney test (*, P ? 0.05). (D) Expression of HD5 mRNA in the same specimens
as in C. A similar pattern was observed for HD6 (data not shown).
Mucosal inflammation and expression of PC defensin HD5. (A)
Fig.3.
copynumbersinilealmucosa.(A)Antimicrobialactivityinmucosalbiopsiesof
ileal CD and non-IBD controls. Protein extracts were incubated with cultures
of either E. coli (Left) or S. aureus (Right), and bacterial killing was assessed by
usingaflowcytometricassay.BiopsiesfromilealCD(allNOD2wildtype)were
obtained from either macroscopically inflamed (open circles) or uninflamed
(filled diamonds) regions of the ileum. (B) Expression levels of four PC anti-
microbials(ontheleft)andsixnonantimicrobialPCproductsinilealspecimens
ofcontrolsandilealCD.DataarecompiledfromexperimentsdescribedinFigs.
1 and 5 and are expressed as mRNA copy number per 10 ng of RNA.
QuantitativeanalysisofantimicrobialactivityandPCmRNAtranscript
18132 ?
www.pnas.org?cgi?doi?10.1073?pnas.0505256102Wehkamp et al.

Page 5

small intestine is affected by TG expression of HD5 and, more
importantly, that modest changes in expression levels of HD5
resulted in readily detectable differences in the composition of
luminalbacteria.Itisknownthatbacterialmorphologycanchange,
depending on the bacterial environment. Therefore, it is possible
that some of the observed changes may represent morphological
changes in response to the small intestinal microenvironment. In
either case, the presence of HD5 results in alterations in the
microbiota in a dose-dependent manner.
Discussion
A unifying hypothesis for the pathogenesis of IBD is that in
geneticallysusceptibleindividuals,intestinalmicrobescontributeto
the initiation and perpetuation of chronic mucosal inflammation
(1).Underhealthyconditions,thereisacomplexinterplaybetween
commensal microbes and the intestinal mucosa, resulting in the
establishmentofadelicatebalance(forreviews,seerefs.44and45).
A perturbation of this balance is proposed to contribute to the
pathogenesis of IBD (23, 24). Herein, we report a decrease of
antimicrobial activity in the ileal mucosa of CD. We attribute this
finding to a reduction in PC ?-defensin expression, but additional
factorsmaycontributealso.BecauselevelsofallotherexaminedPC
products are unchanged or even increased in CD ileitis, the
decrease in HD5 and HD6 may be the result of a specific defect in
PC?-defensinregulation.However,thedecreasewasnotuniformly
observed in all CD patients. Rather, CD patients with ileal disease
specifically showed the decrease. In contrast, patients with CD
limited to the colon had normal levels of PC ?-defensins in the
ileum, and levels were also unchanged in UC. Furthermore, the
observed decrease in PC ?-defensins seen in ileal CD specimens
was not a consequence of the presence or intensity of mucosal
inflammation. Finally, we found, in an in vivo murine model, the
appearance of the luminal microbiota is altered by differences in
HD5 expression comparable to those seen in ileal CD vs. non-IBD
controls. We therefore propose that a specific deficiency of PC
defensins characterizes CD of the ileum, and this deficit helps to
define on a molecular level the phenotypic localization of disease
to the ileum. This deficit affects the antibacterial host defense
capacity of the intestinal mucosa and may initiate and?or perpet-
uate the chronic inflammation that characterizes this disease.
The specific decrease in PC ?-defensins in ileal CD raises the
question of whether this decrease is because of a primary genetic
defect. This would seem unlikely, because the defensin locus at
chromosome 8p23 has not been identified as a significant suscep-
tibility locus in CD (2). Furthermore, the decrease in two different
?-defensins (HD5 and HD6), which are located ?126 kb apart and
flank the chromosomal position of HNP1-4 (46), argues against a
single discrete causative mutation. However, locus polymorphisms
that alter gene copy numbers of the clustered defensin genes have
been reported (42, 43). Our data revealed that HD5 and HD6 gene
copy numbers are unchanged in all of the tested CD patients (n ?
20) compared with controls, suggesting decreased gene copy num-
ber is not a common cause for decreased expression. It remains
possible, but speculative, that null mutations in either of the PC
?-defensin genes might lead to a decrease in HD5 or HD6
expressions of similar magnitude to those reported here and result
in the phenotype of ileal CD.
An alternative explanation for the observed decrease in PC
?-defensins is that a mutation in an IBD susceptibility gene, such
as NOD2?CARD15, could influence PC ?-defensin expression.
Consistent with this notion, we found a more pronounced decrease
of HD5 and HD6 expression in samples harboring the NOD2?
CARD15 SNP13 mutation. The selective reduction in PC ?-defen-
sin expression would suggest that the other PC antimicrobials,
lysozyme and sPLA2, which are not similarly affected by mutations
in NOD2?CARD15, are not similarly regulated. Although there is
Fig.4.
analyzed by in situ hybridization by using an antisense probe (Left). Hybridization of HD5 probe to section of human ileum (Inset). Wild-type mouse shows no
hybridizationtoHD5antisenseprobe(datanotshown).Senseprobe(Right)andRNaseApretreatmentcontrols(datanotshown)werenegativeforhybridization
signal. Arrows point to dense signals that overlie PCs. Counterstain was hematoxylin?eosin. (Scale bar, 20 ?m.) (B) Expression of HD5 mRNA in ileal specimens
fromheterozygousandhomozygousHD5TGmice.DataexpressedasmRNAcopynumberper10ngoftotalRNAdeterminedwithquantitativereal-timeRT-PCR
using external standards. [Scale bars represent means (?standard error).] (C) FISH analysis of luminal microbes in mouse ileum. Representative hybridization
analysis with TR-Bac338 probe (detecting all bacteria) is shown for wild-type mice (Left), HD5 TG heterozygote mice (Center) and HD5 TG homozygote mice
(Right).
Bacterialmicrobiotainwild-type,heterozygous,andhomozygousHD5TGmice.(A)ExpressionandlocalizationofHD5mRNAinTGmousesmallintestine
Wehkamp et al. PNAS ?
December 13, 2005 ?
vol. 102 ?
no. 50 ?
18133
MEDICAL SCIENCES