American Journal of Obstetrics and Gynecology (2004) 191, 1678e87

www.ajog.org

The novel antimicrobial peptide b3-defensin is producedby the amnion: A possible role of the fetal membranes ininnate immunity of the amniotic cavity

Irina A. Buhimschi, MD,a,* Maher Jabr, MD,a Catalin S. Buhimschi, MD,a

Anelia P. Petkova, MD,b Carl P. Weiner, MD,c Ghassan M. Saed, PhD,b

Department of Obstetrics, Gynecology, and Reproductive Sciences, Yale University, New Haven, Conn,a Departmentof Obstetrics and Gynecology, Wayne State University, Detroit, Mich,b and Department of Obstetrics, Gynecology andReproductive Sciences, University of Maryland School of Medicine, Baltimore, Md c

Received for publication June 7, 2003; revised February 26, 2004; accepted March 30, 2004

KEY WORDSPreterm laborChorioamnionitis

Fetal membranesAntimicrobialPrematurity

Background: Innate immunity evolved to eliminate microorganisms before, or after their entry

into the tissues, but before enough antigen is available to activate an adaptive, immune response.Innate immunity is so successful that the majority of encountered microbes are neutralized. Theb-defensins are antimicrobial peptides produced by skin and mucosal surfaces and are an integralpart of the innate immune system. The ability of the amnion cells, which are epithelial derivatives,

to produce antimicrobial b-defensins has not been explored.Objective: This study was undertaken to test the hypothesis that amnion cells synthesizeb-defensins under either basal or stimulated conditions.

Methods: Amnion epithelial FL cells (ATCC CCL 62) were cultured in Ham’s F12 andDulbecco’s modified Eagle medium plus 10% fetal calf serum until confluence, then replated into24-well plates at 1.5 million cells per well. Cells from triplicate wells were harvested after 1, 3, 6,

and 24 hours of exposure to microbial wall components (lipopolysaccharide [LPS]: 1 mg/mL orpeptidoglycan [PG]: 10 mg/mL). Reverse transcription real-time polymerase chain reaction wasperformed with the use of human-specific primers for b1, b2, b3, and b4 defensins to compare

basal messenger RNA (mRNA) levels of defensins and in response to treatment. b-actin was usedfor standardization. Protein expression was investigated by immunofluorescence of the cells inculture, and by immunohistochemistry in paraffin sections of human fetal membranes frompregnancies with or without histologic chorioamnionitis.

Results: Amnion FL cells expressed mRNA for all known b-defensins with b3-defensin mRNAlevels significantly higher compared with others (P ! .001, 1-way analysis of variance[ANOVA]). b3 was the only b-defensin whose mRNA was upregulated in response to the

Supported by the March of Dimes (Basil O’Connor Scholar Research Award No. 5-FY00-601, I.A.B.) and National Institutes of Health: RO3-

HD40266 (I.A.B.) and RO1-HL49041-9 (C.P.W.).

* Reprint requests: Irina A. Buhimschi, MD, Department of Obstetrics, Gynecology and Reproductive Sciences, Yale University School of

Medicine, 333 Cedar St, PO Box 208063, New Haven, CT 06520-8063.

E-mail: irina.buhimschi@yale.edu.

0002-9378/$ - see front matter � 2004 Elsevier Inc. All rights reserved.

doi:10.1016/j.ajog.2004.03.081

Buhimschi et al 1679

microbial mimics LPS (1-way ANOVA, P = .019) and PG (1-way ANOVA, P = .011).

Immunofluorescence confirmed that b3-defensin protein was present in cultured amnion cells,

and upregulated in response to PG and LPS in distinct cells. Similarly, in tissue sections of human

fetal membranes amnion epithelium was intensely positive for b3-defensin protein by

immunohistochemistry. Conspicuous b3-defensin staining was also detected in the chorio-

decidua.

Conclusion: Amnion cells have the ability to produce b-defensins. The b3-defensin appears to be

the predominant epithelial defensin expressed. Its induction by microbial mimics suggests that the

amniotic epithelium may play a role in the innate immunity of the amniotic cavity.

� 2004 Elsevier Inc. All rights reserved.

Innate immunity functions to eliminate potential in-fection by either blocking the entry of microorganismsinto the tissues, or after entry but before enough antigenbecomes available to elicit an adaptive immune response.Thus, innate immunity prevents the establishment ofinfection and indirectly of its associated inflammatoryresponse. From an evolutionary perspective, this ancientdefense system appeared before separation of vertebratesand invertebrates. In contrast, adaptive (acquired) im-munity emerged just 450 million years ago.1 A funda-mental characteristic of innate immunity is its ability torecognize a wide variety of microorganisms: Gram-pos-itive andGram-negative bacteria, yeast, and even viruses.Epithelial cells are the first cells to encounter infectingmicroorganisms and, as such, have developed severalmechanisms for microbial protection. In many verter-brates, the primary defense mechanism of epithelial cellsis the ability to produce defensins that are small, 3 to 5 kdcationic peptides that can form pores in the cytoplasmicmembrane of intruding microorganisms.2,3

Two types of defensins are recognized in humans:alpha-defensins,producedbyneutrophils,T lymphocytes,and Paneth cells of the small intestine, and beta-defensinsproduced by epithelial cells. Four b-defensins have beendescribed to date. The first (human b1-defensin [HBD-1])was isolated from hemofiltrates4 and subsequently detec-ted in breast milk5 and urine where it is derived from thekidney and female urogenital tract.6 The b1-defensin isconstitutively expressed, and is not upregulated in thepresence of bacteria.7 The second beta-defensin (HBD-2)was first identified in keratinocytes8 and lung epithelium.9

Unlike b1, b2-defensin is induced by contact with micro-organisms such asPseudomonas aeruginosa, or inflamma-tory stimuli such as tumor necrosis factor a (TNF-a) andinterleukin-1b (IL-1b).9 Increased expression ofb2-defen-sin in patients with psoriasis10 is linked to an increasedresistance to skin infections,11 whereas decreased pro-duction or activity may explain a 30% increase in the rateof skin infections in patients with atopic dermatitis,a chronic inflammatory disease of the skin.11 Yet, in theboth conditions, there is a disrupted skin barrier.

In 2000 through 2001, several groups identified a thirdbeta defensin (human b3-defensin, HBD-3),12-14 fol-

lowed a short time later by a fourth beta-defensin(human b4-defensin, HBD-4)15 Both b3 and b4-defen-sins are inducible, but by distinct stimuli. b3-defensinmessenger RNA (mRNA) is induced in response toinflammatory cytokines, whereas b4 induction is medi-ated by protein kinase C activation.15

With the use of proteomics and SELDI-TOF (surface-enhanced-laser-desorption-ionization time-of-flight)technology, our group has previously described a specificprotein profile in the amniotic fluid of women withintrauterine inflammation leading to preterm delivery.16

We discovered that some of these proteins were antimi-crobial peptides involved in innate immunity. We furtherbecame intrigued by the possibility that the intraamni-otic cavity may be protected from ascending infection byproduction of defensins. We reasoned that the amnioncells, which are epithelial derivatives, interposed betweenthe septic cervico-vaginal environment and the normallysterile environment, could be an excellent candidate sitefor defensin production. Herein, we test the hypothesisthat the amnion epithelium synthesizes beta defensinsunder both basal and stimulated conditions.

Material and methods

Cell culture and treatments

FL cells, originally derived from normal human amnion(ATCCCCL-62), were obtained from theAmerican TypeCulture Collection (Rockville,Md).17Cells were grown at37(C in 5%CO2 and 95% air, in a mixture of Ham’s F12and Dulbecco’s modified Eagle medium (F12/DMEM)(1:1 [vol/vol]), supplemented with 10% nonheat inacti-vated donor calf serum containing 2 mmol/L L-gluta-mine, 50 U/mL penicillin, and 5 mg/mL streptomycin(Invitrogen,Carlsbad, Calif). The cellswere cultured in 75cm2 flasks. When 80% confluence of cells was achieved,they were seeded into 12-well plates at 1.5! 106 cells perwell in F12/DMEM with 10% fetal calf serum (FCS).After 24 hours, nonadherent cells were removed bya single change with serum-free medium, which was thenreplaced with 4 mL/well fresh F12/DMEM with 10%FCS. The cells were allowed to equilibrate for 1 hour,

1680 Buhimschi et al

after which triplicate wells were treated with either 1 mg/mL Escherichia coliederived lipopolysacharide (LPS)(026:B6, Sigma, St Louis, Mo), or 10 mg/mL peptidogly-can (PG) (Fluka, Milwaukee, Wis).18 Serum in the mediaserved as a source of LPS-binding protein and solubleCD-14 necessary when investigating stimulation of non-myeloid cells by LPS.19 Control wells received equivalentvolumes of sterile water. After 1, 3, 6, and 24 hours, thecells from each of the 3wells were harvested separately forRNA isolation. The experiment described previously wasperformed twice.

For immunofluorescence experiments, cells wereseeded onto 25 ! 75-mm slide chambers (Nunc, Ro-chester, NY) at a density of 105 cells per chamber. Thechambers (4 in each group) were then treated with eitherLPS or PG for 24 hours, while controls received anequivalent water volume.

Total RNA isolation was performed by using RNAaffinity columns as instructed by the manufacturer(RNeasy Mini Isolation kit, Quaigen, Valencia, Calif).RNA was quantitated by the measurements of absor-bance at 260 nm (A260). RNA purity was determined bythe ratio of absorbance at 260 nm and 280 nm (A260/A280). Ratios between 1.8 and 2.1 were consideredacceptable. RNA integrity was assessed with 1% aga-rose gel electrophoresis and ethidium bromide staining.Samples with either smearing or invisible 28S or 18Sbands were considered unacceptable.

Primer design

Optimal oligonucleotide primer pairs for real-time poly-merase chain reaction (rtPCR) amplification were de-signed with the aid of the computer program, BeaconDesigner 2.0 (Premier Biosoft, Palo Alto, Calif). Table Ishows the human oligonucleotide primers, which amplifyvariable portions of the coding regions that were used.Because quantitative application of this method is con-tingent on the analysis of the PCR products during theexponential amplification phase before the plateau, cyclerelationships, and dilutional curves for complementaryDNA (cDNA) for each molecule and the housekeepinggene b-actin were determined (data not shown).20

Reverse transcription and rtPCR (RT-rtPCR)

RNA concentration was adjusted to 6 mg total RNA per20 mLDEPCwater for each sample, heated to 68(C for 10minutes in the presence of 4 mL oligo dT primer, and thenrapidly chilled on ice. A master mix solution containing 8mL 5! first strand buffer, 4 mL of 0.1 mol/L dithiothrei-trol, 2 mL RNAse inhibitor, and 2 mL Superscript II (200U, Life Technologies, Carlsbad, Calif) was added and thereaction incubated at 42(C for 1 hour. Aliquots from thecDNA reaction were PCR amplified in 25 mL reactions asfollows: 2.5 mL of 10! PCR buffer, 1 mmol/L from eachdNTP, 200 nmol/L of forward and reverse primer, 2.5 mL

Syber green, and 2.5 U Taq polymerase enzyme. Thereaction was initiated by heat denaturation at 95(C for 2minutes, annealing the primers for 30 seconds at 64(C,and extension for 30 seconds at 72(C. This was repeatedfor a maximum of 40 cycles with the SmartCycler(Cepheid, Sunnyvale, Calif). In rtPCR, we monitoredproduct accumulation by using the fluorescent dye Sybergreen, allowing for higher sensitivity compared withconventional endpoint PCR detection. The more abun-dant a particular cDNA is in the sample, the fewer PCRcycles are needed to enter the exponential amplificationphase of the PCR. The relative abundance of defensinmRNAwas derived from the ratio between the number ofPCR cycles required for the housekeeping gene b-actin,and the number of cycles needed for each of the 4 b-defensin genes of interest. The principle of themethod hasbeen described elsewhere.21

Immunofluorescence

At the end of a 24-hour incubation with and withoutbacterial products, adherent cells were fixed with 3%freshly depolymerized paraformaldehyde for 30 mi-nutes, followed by repeated washings with phosphatebuffer saline solution (PBS). After blocking with 0.15mol/L glycine for 10 minutes, the cells were permeabi-lized with 0.1% Triton X-100 for 10 minutes and thenincubated with polyclonal primary antibody (rabbitanti-HBD-3; Novus Biologicals, Littleton, Colo) in a di-lution of 1:250 for 1 hour, followed by incubation with10 mg/mL fluorescent secondary antibody (goat anti-rabbit Texas red conjugated; Jackson Immunoresearch,West Grove, Pa). After thorough washing in PBS, thenuclei were counterstained with 40,6-diamidino-2-phe-nylindole (DAPI) for 2 minutes. Slides were mounted inan aqueous mounting medium containing antifade re-agents and examined for red (b3-defensin) and bluefluorescence (DAPI). On one slide in each treatmentgroup, the first antibody was omitted and served asnegative control. Images were acquired with a digitalcamera with the same exposure time.

Patient samples

Fetal membranes and placenta for immunohistochemi-cal study were obtained immediately after delivery. Allpatients provided written informed consent. The respec-tive Human Investigation Committees at the Universityof Maryland and Wayne State University approved theprotocols. The inclusion criteria were either spontane-ous delivery at term and negative placental pathology(control group, n = 5), or preterm vaginal delivery afterspontaneous labor or preterm premature rupture ofmembranes and positive placental pathology indicatinghistologic chorioamnionitis (n = 8). Fetal membraneand placental specimens were fixed in 10% neutralbuffered formalin and embedded in paraffin. The sections

Buhimschi et al 1681

Table I Primer sequences used in RT-rtPCR

Gene name and accession no. Primers sequences Product size (bp)

b1-defensin (HBD-1: NM_005218) Sense: GCCTCTCCCCAGTTCCTGAA 82Antisense: GCAGAGAGTAAACAGCAGAAGGTA

b2-defensin (HBD-2: NM_004942) Sense: TGTGGTCTCCCTGGAACAAAAT 105Antisense: GTCGCACGTCTCTGATGAGG

b3-defensin (HBD-3: AF217245) Sense: CTTCTGTTTGCTTTGCTCTTCCT 138Antisense: CTGTTCCTCCTTTGGAAGGCA

b4-defensin (HBD-4: AF202031) Sense: CACTCTACCAACACGCACCTAG 133Antisense: CGCAACTGGAACCACACACT

b-actin (NM_0011101.2) Sense: AAGCAGGAGTATGACGAGTCCG 559Antisense: GCCTTCATACATCTCAAGTTTGG

were also stained with hematoxylin and eosin (placenta)or Masson trichrome (fetal membranes), and examinedby a clinical pathologist for the presence of histologicevidence for chorioamnionitis. This diagnosis requiredthe presence of inflammatory cells in the subchorioniclayer, amniotic layer, or chorionic plate on biopsy speci-mens from at least 2 regions.

Immunohistochemistry

The 6-mm sections were deparaffinized in xylene andrehydrated with graded ethanol to potassium-PBS solu-tion, pH 7.2. After antigen retrieval with trypsin (Sigma),the sections were pretreated with 1% hydrogen peroxidefor 15minutes, followed by overnight incubation (at 4(C)with rabbit anti-HBD-3 antibody (1:300) and then a 1-hour incubation at room temperature with biotinylatedgoat antirabbit immunoglobulin G (IgG) (Jackson im-munochemicals: 1:600). Immunohistochemical stainingwas performed with avidin-biotin staining (VectastainElite ABC; Vector Laboratories, Burlingame, Calif) with3,30-diaminobenzidine/nickel sulfate as the chromogensolution. The tissue sections were dehydrated in gradedethanols, cleared, and mounted. Specific staining wasevaluated semiquantitatively by examining 6 fields perslide and subjectively scoring on a scale from 0 (nostaining) to 5 (intense blue-black staining) the intensityof the chromogen deposited in the amnion epitheliumversus chorio-decidua. Amedian score was computed foreach patient.

Data presentation and statistical analyses

After first confirming the normal distribution of resultsby using the Kolmogorov-Smirnov and Shapiro-Wilk Wtest, statistical comparisons were performed with Stu-dent t tests, 1-way or 2-way analysis of variance(ANOVA), followed by post hoc StudenteNewman-Keuls tests as appropriate. The results are presented asmean and SEM for continuous variables and as medianand range for discontinuous variables. P values of lessthan .05 are considered to indicate a statistically signif-icant difference between means.

Results

Quantification of b-defensin mRNA levels inFL cells

RT-rtPCR revealed that under basal, unstimulatedconditions, the mRNA for the b3 and b4-defensins weresignificantly more abundant than the mRNAs for theb1 and b2. The b3-defensin mRNAs was significantlymore abundant than all other defensins (ratio b-actin/b-defensin: b1: 0.59 G 0.06; b2: 0.61 G 0.04; b3: 0.91 G0.05; b4: 0.76 G 0.02; 1-way ANOVA F = 12.3, P !.001, Figure 1).

Changes in b-defensin mRNA levels in FL cellstreated with LPS or PG

The percentage of increase in b-defensin mRNA levelsafter continuous exposure to the microbial mimics LPSand PG was compared among the 4 defensins. Levelswere compared with the unstimulated levels at time 0 foreach of the 4 genes of interest. A 2-way ANOVA withLPS or PG on 1 level and the percentage of change inthe 4 b-defensin mRNA levels at 3 hours revealed thatboth LPS and PG significantly increased b3-defensinmRNA levels (2-way ANOVA F = 6.26, P = .002),whereas levels for the other mRNAs remained un-changed. There was no difference between the 2 stimu-lants (2-way ANOVA F = 1.33, P = .259, Figure 2).

The level of b3 mRNA (Figure 3) after LPS was timedependent (1-way ANOVA F = 4.76, P = .019) withthe level of change at 3 hours being maximal (notdifferent from the level at 6 hours, P = .821, signifi-cantly higher compared with 1 hour [P = .04] orprestimulation [P = .027]), suggesting that 3 hourswas an appropriate experimental time to identifychanges in mRNA levels in response to LPS. Theresponse to PG was similar (1-way ANOVAF = 5.609, P = .011), again with 3 hours being maxi-mal, not different from 6 hours (P = .992), but greaterthan 1 hour (P = .010) (Figure 3). A 2-way ANOVAwith the type of microbial constituent as 1 factor and the

1682 Buhimschi et al

Figure 1 A, Representative fluorescence curves of RT-rtPCR for b1, b2, b3, b4-defensins, and b-actin mRNA expression in

cultured CCL-62 amnion cells. B, Relative abundance of b-defensin mRNA in unstimulated CCL-62 cells. On the y-axis is the ratiobetween the number of PCR cycles needed for b-actin amplification and the number of cycles needed for amplification of each of the4 b-defensins. Data are presented as mean G SEM. (1-way ANOVA, followed by StudenteNewman-Keuls post hoc tests, n = 4 in

each group).

time of stimulation as the second factor revealed a highlystatistically significant effect of both PG and LPS withthe time of exposure (F = 10.177, P ! .001), whichagain was not different between LPS and PG itself(F = 0.841, P= .368). There was no statistically signif-icant interaction between stimulus and time (P= .874).

FL cells in culture express b3-defensin protein

To confirm the increase in mRNA levels observed inCCL-62 amnion cells in culture translated into increasedprotein expression, we performed immunofluorescenceon unstimulated and stimulated cultured cells by usingthe same stimulation protocols as for mRNA inductionexperiments. We observed positive red fluorescence inCCL-62 even in unstimulated conditions, demonstratingbasal expression of defensin 3 in amnion cells (Figure 4,A). Figure 4, B, illustrates the negative control slideswith nuclei blue after DAPI counterstain. Figure 4, Cand D, show representative fields from the slide cham-bers exposed to 1 mg/mL LPS and 10 mg/mL PG,respectively. In both conditions, some cells appearbrightly fluorescent. Of note, only some cells respondedwith increased b3-defensin protein expression afterexposure to the microbial wall constituents LPS andPG, although this was purportedly a homogenous cellpopulation.

b3-Defensin is present in humanfetal membranes

To demonstrate biologic relevance, we performed im-munohistochemical staining for b3-defensin on paraffin

sections of human fetal membranes from 8 patients whodelivered preterm spontaneously (gestational age: 29.4G 1.3 weeks) and had clinical symptoms of chorio-amnionitis confirmed by placental pathology. The con-trol patients (n = 5) delivered spontaneously at term

Figure 2 Relative abundance of b-defensin mRNA (percent

from unstimulated levels at time 0) in CCL-62 cells at 3 hoursof stimulation with LPS (1 mg/mL) or PG (10 mg/mL). Dataare presented as mean G SEM. (*P ! .01, 2-way ANOVA).

The number of experimental replicates are as follows: LPS:b1 n = 3, b2 n = 5, b3 n = 5, b4 n = 5; PG: b1 n = 3, b2n = 4,b3n = 5,b4n = 5;CRL:b1n = 3,b2n = 5,b3n = 6,

b4 n = 6.

Buhimschi et al 1683

(gestational age: 37.1 G 0.2 weeks) without clinical orpathologic evidence of chorioamnionitis. The medianstaining score for the patients with chorioamnionitis wasamnion: 3.5 (range 1-5), chorio-decidua: 3 (range 0-4),(Figure 5, B, C, and D) and was significantly differentfrom patients without chorioamnionitis: amnion:0 (range 0-1), chorio-decidua: 1 (range 0-1) (Figure 5,A; t test P! .001). Slides in which the first antibody wasomitted were clear of any staining (Figure 5, E).

Comment

The results of the current study, interpreted in light ofour earlier work, indicate that (1) FL cells, originallyderived from normal amnion, express mRNA for epi-thelial b-defensins; (2) b3-defensin is the predominantamniotic epithelial defensin; and (3) b3-defensin mRNAand protein are induced by exposure to Gram-positiveand Gram-negative microbial wall constituents. To-gether, these findings suggest that the amnion partic-ipates in innate immunity by producing b3-defensin,a recently identified member of the epithelial defensinfamily of peptides.

The epithelial defensin system is likely to be impor-tant in the control of infections and adaptive inflamma-tory responses of the female reproductive tract as it mayfunction to prevent colonization of the upper genitaltract during pregnancy. Yet, more important is thefinding that it is the b3 form, which has larger signifi-

Figure 3 Time course of the relative abundance (percentfrom unstimulated levels at time 0-CRL) of b3-defensinmRNA in CCL-62 cells after 1, 3, and 6 hours of stimulationwith LPS (1 mg/mL) or PG (10 mg/mL). Data are presented asmean G SEM. The number of experimental replicates are as

follows: LPS: 1 hour n = 3, 3 hours n = 5, 6 hours n = 4;PG: 1 hour n = 3, 3 hours n = 5, 6 hours n = 4; CRL (0hour) n = 5.

cance over other constitutive (b1) or inducible (b2, b4)epithelial defensins. Of the 4 epithelial defensins, b3 isthe most potent, is much more basic, has a wider andstronger bactericidal activity against Gram-positive,Gram-negative bacteria (including Staphylococcus au-reus, vancomycin-resitant Enterococcus faecium) as wellas the yeast Candida albicans, and is salt insensitive,a trait especially important in the amniotic fluid. Severalof these properties are explained by the unique ability ofdefensins to form dimers at low concentrations.22 Thehigh salt concentration and alkaline pH of amnioticfluid would be unfavorable for other cationic antimi-crobial peptides whose activity is generally lost in highsalt solutions.

The b3-defensin has been reported only recently asa new antimicrobial defensin of epithelial origin.12-14 Jiaet al12 have shown that the PCR signal for b3-defensinwas absent in fetal lung tissue but was expressed aftertreatment with interleukin-1b. Harder et al14 also in-dicated that b3-defensin is upregulated by bacteria andTNF-a. These reports suggest an inducible nature of theb3-defensin gene. Interestingly, the promotor analysis ofb3-defensin gene did not reveal any NFkB consensuselements up to 2900 base pairs (bp) of the 50 end of thecoding region, in contrast to b2-defensin.12 However,the region contains several consensus sequences foractivator protein-1 (AP-1), interferon gamma responseelements, and nuclear factor interleukin-6 (NF-IL-6)response elements suggesting that b3-defensin may beregulated by inflammatory stimuli but in an NFkB-independent manner.12 Our findings in amnion cells areprovocative because the most abundant isoform is thenovel b3-defensin and it appears to be induced bybacterial products. However, the molecular mechanismsresponsible for the genomic regulation of b3-defensin inhuman amnion require further investigation.

It is well recognized that colonization of the lowergenital tract by a variety of microorganisms may lead toascending intrauterine infection and inflammation lead-ing to preterm labor.23,24 However, the chain of eventslinking maternal or fetal exposure to microorganismsand intra-amniotic inflammation to preterm delivery hasyet to be elucidated. It is possible that the production ofendogenous antibiotic peptides by fetal membranes maybe critical in maintaining sterility in the intra-amnioticenvironment, and that the loss or ineffectiveness of thesefactors could be responsible for the microbial coloniza-tion of the amniotic cavity. Kjaergaard et al25 observedthat amniotic membranes (both chorion and amnion)have antibacterial properties against a diverse panel ofbacteria. Our study suggests that at least 1 responsiblefactor is b3-defensin.

We are struck by the heterogeneity of b3 staining inamnion cells reflecting that they are distinct entities bothin vitro and in vivo. In preliminary experiments, wecould not link the b3-defensin pattern of expression and

1684 Buhimschi et al

Figure 4 Immunofluorescence of b3-defensin in CCL-62 cells in culture (A) and after stimulation with LPS (1 mg/mL, C) or PG(10 mg/mL, D). B, Negative control with omitted primary antibody. Nuclei are counterstained with DAPI. (Original magnification:!200.)

Buhimschi et al 1685

Figure 5 Imunohistochemistry of b3-defensin in human fetal membranes in a representative slide from a control patient (A) anda patient with chorioamnionitis that delivered preterm (B, C). D, Morphologic (Masson trichrome) staining of the section from thepatient with chorioamnionitis indicating presence of inflammatory cells in the amnion and chorio-decidua. E, Negative control with

omitted primary antibody (same section as in B). The slides are not counterstained. (Original magnifications: A, B, D: !200; C:!400; E: !100.)

apoptosis (not shown). A reasonable explanation for theheterogeneity in staining of the amnion cell line is thatthe cells are differentiating in culture and that differen-tiated cells express higher levels of b3-defensin. More-over, the similar aspect observed in the amnion lining byimmunohistochemistry indicates that the possible differ-entiation may also occur in vivo. This observation mayalso suggest that the degree of defensin synthesis is anintrinsic property of some amnion cells rather thana reflection of the surrounding environment and sup-

ports the anisotropic nature of the amnio-chorion asdescribed previously by others.26

Analogous with other studies, a possible limitation ofour study rests on the use of human amnion FL cells(ATCC CCL-62). The process of immortalization andprolonged culture conditions may have induced a geno-typic or phenotypic drift in the cell line. The FL cell linewas originally thought to be derived from normal am-nion, but subsequently, and based on isoenzyme analysis,HeLa marker chromosomes, and DNA fingerprinting,

1686 Buhimschi et al

it was concluded the line was contaminated with HeLacells (http:\\www.atcc.org). This limitation extends notonly to the FL cell line but also to the WISH cell line(ATCC CCL-25). Both are routinely used to model invitro the function of the amnion.27-29 However, a studypublished by Kniss et al30 demonstrated that althoughHeLa and WISH cells are genetically identical, theyexhibit distinct phenotypical properties. The alternativemodel, primary amnion cell culture, also has disadvan-tages such as the mixed nature of each preparation,limited lifespan of the culture, and the potential micro-bial contamination problems. Further, a distinct alter-ation of some enzymatic activities in primary amnioncell culture has been attributed to either differentiationin vitro or to an artifact of primary culture condi-tions.26 Although recognizing the limitations of themodels available for in vitro study of the amnion, ourimmunohistochemistry results confirming the expres-sion of b3-defensin in situ in human fetal membranes,strongly support the validity of the observations withFL cells.

The demonstration that women with intra-amnioticinflammation have an increased innate immune responsethat translates into a characteristic proteomic profilewhose presence was highly correlated with the whiteblood cell count in the amniotic cavity may seem at oddswith the idea that impaired innate immunity may allowthe causative infectious agent to create an infection inthe first place. However, the antimicrobial peptides weidentified using SELDI (neutrophil defensins and theS100 proteins calgranulin A and C) were mainly ofinflammatory and not epithelial cell origin.16 Second, itis now clear that it is not the infectious process per sethat is deleterious to the fetus, but rather the excessiveinflammation that accompanies an intra-amniotic in-fection. In support of this observation we find that insome pregnancies ending with preterm delivery, there isan outpouring of inflammatory mediators and freeradicals in the amniotic cavity in the absence of a pos-itive culture result.31 This suggests either that the cultureconditions are suboptimal for the infectious cause, orand perhaps most probably, that the endogenous anti-bacterial mediators were successful neutralizing thecausative agent but remained behind. Third, in theinstance of a microorganism ascending from the vagina,there may be a time sequence and even cross-talkbetween the induction of epithelial and neutrophildefensins that are peptides distinct in cellular originand regulation. Under normal conditions, neutrophilsare generally absent from the amniotic fluid. Thus, thepresence of mediators of inflammatory cell origin in theamniotic cavity is a consequence of the inflammatorycell chemotaxis within the amniotic cavity. In support ofthis concept, there is evidence that b-defensins act alsoas chemokines bridging the innate and adaptive immunesystems.32

In conclusion, the amnion participates in innateimmunity by producing b3-defensin. The productionof this endogenous antibiotic peptide may be critical inmaintaining the intra-amniotic environment sterile andpreventing the generation of the adaptive inflammatoryresponse probably in most pregnancies.

Acknowledgments

We thank Dr Randall Armant for his assistance withoptimizing the immunofluorescence procedure and DrMichael Diamond for his support and assistance withthe preparation of the manuscript.

References

1. Hoffmann JA, Kafatos FC, Janeway CA, Ezekowitz RA. Phylo-

genetic perspectives of innate immunity. Science 1999;284:1313-8.

2. Lehrer RI, Barton A, Daher KA, Harwig SS, Ganz T, Selsted ME.

Interaction of human defensins with Escherichia coli: mechanism

of bactericidal activity. J Clin Invest 1989;84:553-61.

3. Ganz T, Lehrer RI. Defensins [review] Curr Opin Immunol

1994;6:584-9.

4. Bensch KW, Raida M, Magert HJ, Schulz-Knappe P,

Forssmann WG. hBD-1: a novel beta-defensin from human

plasma. FEBS Lett 1995;368:331-5.

5. Tunzi CR, Harper PA, Bar-Oz B, Valore EV, Semple JL, Watson-

MacDonell J, et al. Beta-defensin expression in human mammary

gland epithelia. Pediatr Res 2000;48:30-5.

6. Valore EV, Park CH, Quayle AJ, Wiles KR, McCray PB Jr.,

Ganz T. Human beta-defensin-1: an antimicrobial peptide of

urogenital tissues. J Clin Invest 1998;101:1633-42.

7. Zhao C, Wang I, Lehrer RI. Widespread expression of beta-

defensin hBD-1 in human secretory glands and epithelial cells.

FEBS Lett 1996;396:319-22.

8. Harder J, Bartels J, Christophers E, Schroder JM. A peptide

antibiotic from human skin. Nature 1997;387:861.

9. Harder J, Meyer-Hoffert U, Teran LM, Schwichtenberg L,

Bartels J, Maune S, et al. Mucoid Pseudomonas aeruginosa,

TNF-alpha, and IL-1beta, but not IL-6, induce human beta-

defensin-2 in respiratory epithelia. Am J Respir Cell Mol Biol

2000;22:714-21.

10. Ong PY, Ohtake T, Brandt C, Strickland I, Boguniewicz M,

Ganz T, et al. Endogenous antimicrobial peptides and skin

infections in atopic dermatitis. N Engl J Med 2002;347:1151-60.

11. Christophers E, Henseler T. Contrasting disease patterns in

psoriasis and atopic dermatitis. Arch Dermatol Res

1987;279:S48-51.

12. Jia HP, Schutte BC, Schudy A, Linzmeier R, Guthmiller JM,

Johnson GK, et al. Discovery of new human beta-defensins using

a genomics-based approach. Gene 2001;263:211-8.

13. Garcia JR, Jaumann F, Schulz S, Krause A, Rodriguez-Jimenez J,

Forssmann U, et al. Identification of a novel, multifunctional beta-

defensin (human beta-defensin 3) with specific antimicrobial

activity: its interaction with plasma membranes of Xenopus

oocytes and the induction of macrophage chemoattraction. Cell

Tissue Res 2000;306:257-64.

14. Harder J, Bartels J, Christophers E, Schroder JM. Isolation and

characterization of human beta -defensin-3, a novel human in-

ducible peptide antibiotic. J Biol Chem 2001;276:5707-13.

15. Garcia JR, Krause A, Schulz S, Rodriguez-Jimenez FJ, Kluver E,

Adermann K, et al. Human beta-defensin 4: a novel inducible

peptide with a specific salt-sensitive spectrum of antimicrobial

activity. FASEB J 2001;15:1819-21.

Buhimschi et al 1687

16. Buhimschi IA, Christner R. Biomarkers of intra-amniotic inflam-

mation. WO Pat. 2004/04328 A2, Pat. Appl. 60/426, 096 filed

November 14, 2002.

17. Fogh J, Lund RO. Continuous cultivation of epithelial cell strain

(FL) from human amniotic membrane. Proc Soc Exp Biol Med

1957;94:532-7.

18. Carl VS, Brown-Steinke K, Nicklin MJ, Smith MF Jr. Toll-like

receptor 2 and 4 (TLR2 and TLR4) agonists differentially regulate

secretory interleukin-1 receptor antagonist gene expression in

macrophages. J Biol Chem 2002;277:17448-56.

19. Tapping RI, Tobias PS. Cellular binding of soluble CD14 requires

lipopolysaccharide (LPS) and LPS-binding protein. J Biol Chem

1997;272:23157-64.

20. Saed GM, Zhang W, Diamond MP. Molecular characterization of

fibroblasts isolated from human peritoneum and adhesions. Fertil

Steril 2001;75:763-8.

21. Gibson UE, Heid CA, Williams PM. A novel method for real time

quantitative RT-PCR. Genom Res 1996;6:995-1001.

22. Schibli DJ, Hunter HN, Aseyev V, Starner TD, Wiencek JM,

McCray PB Jr., et al. The solution structures of the human beta-

defensins lead to a better understanding of the potent bactericidal

activity of HBD3 against Staphylococcus aureus. J Biol Chem

2002;277:8279-89.

23. McGregor JA, French JI. Bacterial vaginosis in pregnancy. Obstet

Gynecol Surv 2000;55(Suppl 1):S1-19.

24. Dudley DJ. Pre-term labor: an intra-uterine inflammatory

response syndrome? J Reprod Immunol 1997;36:93-109.

25. Kjaergaard N, Hein M, Hyttel L, Helmig RB, Schonheyder HC,

Uldbjerg N, et al. Antibacterial properties of human amnion

and chorion in vitro. Eur J Obstet Gynecol Reprod Biol 2001;

94:224-9.

26. Okita JR, Sagawa N, Casey ML, Snyder JM. A comparison

of human amnion tissue and amnion cells in primary culture

by morphological and biochemical criteria. In Vitro 1983;

19:117-26.

27. Xue S, Brockman DE, Slater DM, Myatt L. Interleukin-1 beta

induces the synthesis and activity of cytosolic phospholipase A2

and the release of prostaglandin E2 in human amnion-derived

WISH cells. Prostaglandins 1995;49:351-69.

28. Dutheil N, Malhomme O, Provost N, Becquart P, Burguette T,

Schlehofer JR, et al. Presence of integrated DNA sequences of

adeno-associated virus type 2 in four cell lines of human embryonic

origin. J Gen Virol 1997;78:3039-43.

29. Minemoto Y, Gannon J, Masutani M, Nakagama H, Sasagawa T,

Inoue M, et al. Characterization of adriamycin-induced G2 arrest

and its abrogation by caffeine in FL-amnion cells with or without

p53. Exp Cell Res 2001;262:37-48.

30. Kniss DA, Xie Y, Li Y, Kumar S, Linton EA, Cohen P, et al.

ED(27) trophoblast-like cells isolated from first-trimester chorionic

villi are genetically identical to HeLa cells yet exhibit a distinct

phenotype. Placenta 2002;23:32-43.

31. Buhimschi IA, Buhimschi CS, Weiner CP. Protective effect of N-

acetylcysteine against fetal death and preterm labor induced by

maternal inflammation. Am J Obstet Gynecol 2003;188:203-8.

32. Yang D, Chertov O, Bykovskaia SN, Chen Q, Buffo MJ,

Shogan J, et al. Beta-defensins: linking innate and adaptive

immunity through dendritic and T cell CCR6. Science 1999;

286:525-8.