Original

os

ara

nica

3; a

e 31

Porphyromonas gingivalis is a Gram-negative anaerobicchallenge and contributing to the chronic nature of thisdisease.

ferences as the multi-cellular complexity and level ofdifferentiation of the oral mucosa between OMMs and mono-layer cultures of oral epithelial cells may influence our under-standing of the hostepathogen interaction. The main

* Corresponding author. Academic Unit of Oral & Maxillofacial Pathology,

School of Clinical Dentistry, University of Sheffield, Claremont Crescent,

Sheffield S10 2TA, UK. Tel.: 44 114 2717957.E-mail address: i.douglas@sheffield.ac.uk (C.W.I. Douglas).

Microbes and Infectionbacterium that significantly contributes to the pathogenesis ofperiodontitis. This inflammatory disease affects the tissues thatsurround and support the teeth and is the leading cause oftooth-loss worldwide [1]. P. gingivalis can invade oralepithelial cells in vivo [2] and in vitro [3e5], and has beendetected in a number of locations within the oral cavity,including the buccal [6] and gingival mucosa [2]. Bacterialinternalisation by epithelial cells is thought to promote itspersistence within oral tissues by evasion of host immune

There has been recent and continued interest in the uti-lisation of in vitro three-dimensional (3D) cultures of full-thickness epithelium as a more anatomically representativemodel than two-dimensional, monolayer cultures [7]. Theculture of organotypic mucosal models (OMMs) has beenwidely reported in the literature although they have been littleused for the study of host interactions with periodontal path-ogens [8e10] and monolayer cultures have remained the mainepithelial model of choice [11,12].

The oral mucosa may provide a protective role to the effectsof internalised periodontal bacteria [13]. Therefore, such dif-1. Introduction responses and therapeutic agents, so prolonging the hostAbstract

Porphyromonas gingivalis is a Gram-negative, keystone pathogen in periodontitis that leads to tissue destruction and ultimately tooth loss.The organism is able to infect oral epithelial cells and two-dimensional (monolayer) cultures have been used to investigate this process.However, recently there has been interest in the use of three-dimensional, organotypic mucosal models to analyse infection. These models arecomposed of collagen-embedded fibroblasts overlain with multilayers of oral epithelial cells. In this study we report for the first time significantdifferences in the response of oral mucosal models to P. gingivalis infection when compared to monolayer cultures of oral epithelial cells.Intracellular survival (3-fold) and bacterial release (4-fold) of P. gingivalis was significantly increased in mucosal models compared withmonolayer cultures, which may be due to the multi-layered nature and exfoliation of epithelial cells in these organotypic models. Furthermore,marked differences in the cytokine profile between infected organotypic models and monolayer cultures were observed, particularly for CXCL8and IL6, which suggested that degradation of cytokines by P. gingivalis may be less pronounced in organotypic compared to monolayer cultures.These data suggest that use of oral mucosal models may provide a greater understanding of the host responses to P. gingivalis invasion thansimple monolayer cultures. 2014 Institut Pasteur. Published by Elsevier Masson SAS. All rights reserved.

Keywords: Porphyromonas gingivalis; Periodontitis; Epithelial cells; Oral mucosa; Chemokines; Innate immunityCharacterisation and optimisation ofstudy Porphyromona

Abigail Pinnock, Craig Murdoch, Keyvan Moh

Integrated Biosciences, University of Sheffield, School of Cli

Received 28 October 201

Available onlin1286-4579/$ - see front matter 2014 Institut Pasteur. Published by Elsevier Mahttp://dx.doi.org/10.1016/j.micinf.2014.01.004article

rganotypic oral mucosal models togingivalis invasion

mzadeh, Simon Whawell, C.W. Ian Douglas*

l Dentistry, 19 Claremont Crescent, Sheffield S10 2TA, UK

ccepted 22 January 2014

January 2014

16 (2014) 310e319www.elsevier.com/locate/micinfsson SAS. All rights reserved.

InspironResaltar

InspironResaltar

InspironResaltar

or differences in the epithelial cells employed in these studies.sig-

nificant differences between host cell responses to P. gingivalis

2.1. Materials

All materials were purchased from Sigma, Poole, UK un-

2.2. Bacterial strains and culture conditions

P. gingivalis strains NCTC 11834 (National Collection of

istarrats at the end of a licenced study (by Mrs C Freeman, Uni-

wasassessed in monolayers and OMMs using an antibiotic pro-

ndType Cultures) and W50 [22], were cultured on fastidiousanaerobe agar (FA; LabM Limited, Lancashire, UK) supple-mented with 10% horse blood (Oxoid, Fisher Scientific,Loughborough, UK) at 37 C in an anaerobic atmosphere(80% N2, 10% H2, 10% CO2; MiniMACS Anaerobic Work-station, Don Whitley Scientific, UK).

2.3. Epithelial cell culture conditions

Normal oral keratinocytes (NOK) were isolated frombuccal biopsies as previously described [23]. Briefly, biopsieswere obtained with written, informed consent from patientsattending the Charles Clifford Dental Hospital, Sheffield,under ethical approval (reference number 09/H1308/66).Biopsies were washed and incubated overnight at 4 C with0.1% (w/v) trypsin. The epithelium and the connectivetissue layers were separated by gentle scraping. Primaryepithelial cells were seeded into a tissue culture flask alongwith 5 105 lethally irradiated murine 3T3 fibroblasts(XCELLentis, Gent, Belgium). Primary fibroblasts were iso-lated from the connective tissue layer by incubation with0.05% (w/v) type I collagenase (GibcoBRL, Paisley, Scotland)less otherwise stated. The murine monoclonal P. gingivalisantibody (clone 1B5) was a kind gift from Prof. M. Curtis,Barts and The London School of Medicine, London, UK [21].infection when oral epithelial cells are grown as monolayers oras OMMs. Our data provides further evidence to support theimportance of using OMMs to study bacterial internalisationand subsequent end-point analyses (e.g. chemokine profiling).

2. Methods and materialsThis study investigates these aspects further and reportsconsequences of P. gingivalis interaction with the epitheliuminclude intracellular bacterial survival, release of chemokinesand cytokines by epithelial cells, and host cell death [20] all ofwhich are features likely to contribute to the initiation and/orpropagation of the disease. However, much of the literatureregarding these host cell responses is conflicting. For example,the time of intracellular survival of P. gingivalis variesconsiderably in the literature from 6 h [14] to 4e8 days[15,16,29]. In addition, some studies have shown that P. gin-givalis-infected epithelial cells display increased production ofchemokines and cytokines [17,19], whereas others show che-mokine/cytokine paralysis or degradation [18]. These conflict-ing data may be due to the use of different strains ofP. gingivalis

A. Pinnock et al. / Microbes aand then cultured in Dulbeccos Modified Eagles Medium(DMEM) GlutaMAX (Gibco, UK) supplemented withtection assay as previously described [26], or a modifiedversion of this method for analysis of OMMs. Monolayers andOMMs were incubated in DMEM GlutaMAX and HamsF-12 (3:1) for 1 h at 37 C and then blocked with 2% (w/v)bovine serum albumin (BSA) in DMEM GlutaMAX for1 h at 37 C. P. gingivalis suspended in the sameDMEM GlutaMAX and Hams F-12 medium (3:1) wasapplied at a multiplicity of infection (MOI) of 100 (mono-layer) or 2 107 cells/300 ml (OMM; determined to be anequivalent MOI 100 for an identical surface area of monolayercultured cells), were incubated at 37 C with monolayers andOMMs for 1.5 or 4 h respectively in 5% CO2 (preliminaryassays indicated that the viability of P. gingivalis in this at-The number of viable intracellular P. gingivalis cellsversity of Sheffield) as previously described [24]. Extractedcollagen was dissolved in 0.1 M acetic acid at 4 C, freeze-dried and re-dissolved in 0.1 M acetic acid to a stockconcentration of 8 mg ml1, and stored at 4 C. Keeping allreagents on ice, models were cultured using a protocol adaptedfrom Ref. [25]. Human primary buccal fibroblasts at a con-centration of 1 106 per model, in CDMEM, were added to asolution of 13.8 mg ml1 DMEM, 2.25 mg ml1 sodium bi-carbonate, 2 mM HEPES, 6.3 mM NaOH, 8.5% (v/v) FCS,2.1 mM L-glutamine and 5.28 mg ml1 rat-tail type I collagen.The resultant fibroblast-containing suspension was incubatedat 37 C for 1e2 h until solidified. Once solid, OMMs werecompletely submerged in supplemented CDMEM for 2 daysfollowing which either NOK or the H357 cell line were seededat a density of 1 106 cells/model in supplemented CDMEMfor 1e2 days to allow keratinocyte adhesion. OMMs wereraised to the air-to-liquid interface for approximately 7 days orleft completely submerged in supplemented CDMEM for 2e3days.

2.5. Antibiotic protection assayRat-tail type I collagen was isolated from the tails of W10% foetal calf serum (v/v, FCS), 625 ng ml1 amphotericinB, 50 U ml1 penicillin and 50 U ml1 streptomycin (com-plete medium (CDMEM)). Keratinocytes were cultured inCDMEM supplemented additionally with 0.1 mM adenine,5 mg ml1 insulin, 5 mg ml1 transferrin, 5 mg ml1 triiodo-thyronine, 0.4 mM hydrocortisone and 10 ng ml1 epidermalgrowth factor. Primary keratinocytes were passaged up to 3times, each time seeding with irradiated 3T3 fibroblasts, andprimary fibroblasts were passaged up to 9 times. The H357epithelial cell line, originally isolated from a squamous cellcarcinoma of the tongue (a kind gift from Prof. Stephen Prime,University of Bristol, UK), was also cultured in supplementedCDMEM.

2.4. Oral mucosal models (OMMs)

311Infection 16 (2014) 310e319mosphere was unaffected; post-invasion viability: anaerobic3.77 0.58%, aerobic 4.46 0.33%, p 0.15, n 3).

subjected to high temperature antigen retrieval in citratefor

30 min at room temperature. The murine monoclonal P. gin-

valis NCTC 11834 (MOI 100 in serum-free medium) washt at

5% CO , 37 C. Preliminary data revealed that fluorescently

1.5 h and 4 h respectively. Models containing intracellular P.with

200 mg ml metronidazole (to kill the external adherent

chemiluminescence signal was detected using a Compact X4care,

Gloucester, UK). Signal intensities were analysed using

cytokeratin 13, 14 and E-cadherin when compared to sectionsof P.

gingivalis into air-exposed and submerged H357 and NOK

nd2

labelling the P. gingivalis cells did not significantly decreasetheir viability or cell invasion capability (post-invasion viablecount: unlabelled bacteria 4.22 0.96%; FITC-labelled3.27 0.32%, p 0.18). OMMs were then fixed in 10%(v/v) PBS-buffered formalin and embedded in optimum cut-ting temperature (OCT) (FisherScientific), at approximately43 C. Sections (10 mm) were prepared using a MicromHM560 cryostat (ThermoScientific), at 20 to 30 C, andmounted on microscope slides. Slides were flooded with

1added to air-exposed H357-OMMs and incubated overniggivalis antibody (clone 1B5) was diluted in horse serum (1:50)and incubated with the sections overnight at 4 C in a hu-midified atmosphere. Slides were washed in PBS and incu-bated with biotinylated secondary antibody for 30 min(VECTASTAIN Elite ABC-Peroxidase Kits; Vector Labora-tories, Peterborough, UK) prepared according to the manu-facturers instructions. After washing, slides were incubatedwith avidin biotinylated enzyme complex (ABC) reagent(Vector Laboratories) for 30 min. Finally, slides were washedand 3,30-diaminobenzidine tetrahydrochloride (DAB) (VectorLaboratories) substrate was added. Slides were counterstainedwith haematoxylin, using a Leica ST4020 Small LinearStainer (Leica Microsystems, Milton Keynes, UK) andmounted using DPX non-aqueous mounting medium (Merck,Nottingham, UK). An isotype mouse IgG1 control (Dako,Copenhagen, Denmark) antibody (1:50) was used to confirmthe specificity of the primary antibody.

2.7. Immunofluorescence

Plate-cultured P. gingivalis NCTC 11834 were washed 3times in PBS and labelled with 5-(6)-carboxyfluorescin suc-cinylester (Invitrogen) in PBS (0.4 mg ml1) for 30 min in thedark at 4 C. After washing, fluorescently-labelled P. gingi-buffer. Sections were then blocked with 100% horse serumFollowing incubation, epithelial cells were washed three timeswith PBS to remove non-adherent bacteria and external,adherent bacteria were killed using 200 mg ml1 metronidazolefor 1 h. After washing the metronidazole from the cells, steriledistilled water was used to osmotically lyse the epithelial cells.OMMs were mechanically lysed by homogenisation (TissueRuptor, Qiagen, West Sussex, UK). The number of exfoliatedepithelial cells in the culture medium was determined using ahaemocytometer prior to homogenisation.

2.6. Immunohistochemistry

OMMs were paraffin embedded, sectioned (4 mM) andmounted onto glass slides (VWR International, Lutterworth,UK). Sections were deparaffinised, rehydrated, peroxidaseactivity quenched with 3% hydrogen peroxide and sections

312 A. Pinnock et al. / Microbes a1 mg ml Hoechst 33342 (ThermoScientific, Northumber-land, UK) for 2 min, washed and then mounted usingOMM was assessed visually by immunohistochemistry andimmunofluorescence (Fig. 2). Staining indicated that P. gin-givalis was able to penetrate the epithelium in both H357 andNOK OMM. Staining of air-exposed H357 OMM sectionsof normal healthy oral mucosa (data not shown). InvasionQuantity One software (Bio-Rad, UK) and calculated relativeto their internal positive controls after subtraction of thenegative background intensity.

2.9. Statistical analysis

Statistical analyses were performed using a Students un-paired two-tailed t test.

3. Results

3.1. Invasion of monolayer and OMM by P. gingivalis

The OMMs used in this study displayed a well-organisedarchitecture consisting of a multi-layered epithelium and afibroblast-populated connective tissue layer (Fig. 1). OMMsgenerated from NOK exhibited a more stratified epitheliumthan those generated with H357 cells. In addition, the thick-ness of the epithelium could be successfully controlled byexposing the surface of the OMM to air or keeping it sub-merged with medium (Fig. 1). Both NOK and H357 OMMsshowed similar patterns of expression of the epithelial markersAutomatic X-Ray Film Processor (Xograph Healthbacteria) for 4 h and the conditioned medium assessed for thepresence of chemokines using the RayBio Human Inflam-mation Antibody Array 3 (Insight Biotechnology Ltd, Mid-dlesex, UK) according to the manufacturers instructions. Thegingivalis were incubated in serum-free medium1ProlongGold anti-fade reagent (Invitrogen, Paisley, UK).Staining was visualised using the Zeiss Axiovert 200 invertedfluorescence microscope with Axiovision imaging software(Zeiss, Ltd).

2.8. Chemokine array

Previous studies assessing the detection of chemokinerelease from epithelial cells treated with P. gingivalis havedescribed a local chemokine paralysis effect [18]. Therefore,NOK monolayers and NOK-OMM were pre-stimulated with25 ng ml1 TNF-a (Peprotec, UK) for 4 h to increase thelevels of chemokine release prior to the addition of P. gingi-valis, allowing for easier assessment of changes in chemokinedetection. Following pre-stimulation, both monolayers andOMMs were incubated with P. gingivalis NCTC 11834 for

Infection 16 (2014) 310e319(Fig. 2A) showed a greater level of P. gingivalis penetrationthan in NOK OMM (Fig. 2B). P. gingivalis is more likely to

cult

y su

ndFig. 1. Haematoxylin and eosin staining of air-exposed and submerged OMM

fibroblast-populated collagen matrix at the air-to-liquid interface or completel

A. Pinnock et al. / Microbes ainvade junctional epithelium where pocket deepening isoccurring and since this epithelium is only a few cells thick,the opportunity to penetrate through the epithelium is likely tobe greater. We found that P. gingivalis was able to penetratecompletely through the thin epithelium and was localised inthe connective tissue in submerged OMM that are only a fewlayers thick (Fig. 2C). Whereas, bacteria were observed onlyin the upper layers of the thicker epithelium in the OMMgrown at an air-to-liquid interface (Fig. 2D).

Comparing monolayer cultures of H357 cells with mono-layers of NOK, the percentage invasion by P. gingivalis (usingan antibiotic protection assay) was slightly higher in the H357cultures than was seen with the NOK monolayer cultures butthis was not statistically significant (3.19 1.68% and2.15 0.90% respectively, Fig. 3A). While some work hasbeen done looking at P. gingivalis invasion of OMM [8e10],there are no reports comparing P. gingivalis invasion in thiswith that which occurs in monolayer cultures. Consequently,we undertook such a comparison and found that the percent-age invasion of H357 cell OMMs by P. gingivalis strain NCTC11834 was 3.38 0.45% which was not significantly differentfrom that obtained with H357 monolayer cultures(3.19 1.68%; Fig. 3A). However, invasion of NOK OMM byP. gingivalis NCTC 11834 was significantly lower than inva-sion into H357 OMM (0.98 1.02% versus 3.38 0.45%respectively; p < 0.05; Fig. 3A). These data confirm the ob-servations of the immunostaining studies (Fig. 2) and suggeststhat there may be differences between the two cell types intheir expression of cell surface proteins/receptors importantured using H357 and NOK. The H357 cell line and NOK were cultured on a

bmerged in culture medium.313Infection 16 (2014) 310e319for invasion of this bacterium or the rate of epithelial celldesquamation at the OMM surface. It must be noted thatbecause the invasion experiments of monolayer and OMMwere performed over different periods of time (e.g. 1.5 h formonolayer and 4 h for OMM), only the total percentage in-vasion was assessed. This is because invasion of monolayerswas greatest after 1.5 h, whereas invasion of OMM byP. gingivalis was greatest after 4 h (data not shown). Thisallowed for comparison of maximal percentage invasionvalues between the two culture systems.

The levels of P. gingivalis invading the thinner-layered,submerged H357 OMM was reduced but not statisticallydifferent from the invasion of H357 OMM grown at an air-to-liquid interface (1.28 0.96% and 2.30 1.74% respectively;Fig. 3B).

3.2. Intracellular survival and bacterial release of P.gingivalis from H357 monolayer and OMM cultures

To compare the intracellular survival of P. gingivalis overtime in both H357 monolayer and OMM cultures, an anti-biotic protection assay was performed and the epithelialcultures were further incubated and assessed for intracellularsurvival by releasing bacteria from lysed epithelial cells.H357 cells were used for this assay because monolayer andOMM cultures using these cells were shown to have the sameor higher percentage P. gingivalis invasion than NOK cul-tures (Fig. 3A). Consequently, H357 cells were deemed abetter model to use because with the greater number of

nd314 A. Pinnock et al. / Microbes ainternalised bacteria the possibility of detecting any changesin intracellular survival over time would be higher. Using thismodel, the intracellular viability of P. gingivalis was found todecrease in a time-dependent manner in both monolayer andOMM to reach zero after 18 h for monolayer cultures but wasmore prolonged with the OMM (still 0.001% at 48 h;Fig. 4A). This prolonged, though low, viability of P. gingi-valis within organotypic cultures suggests that features ofOMM structure or physiology may contribute to the sus-tained survival of this bacterium.

Periodontitis is a cyclical disease and so patients mayexhibit active disease followed by periods of quiescence. It hasbeen proposed that P. gingivalis may escape or be releasedfrom epithelial cells to cause re-infection of the periodontaltissues [27]. To assess this using H357 monolayer and OMMcultures, cells were briefly treated with metronidazolefollowing bacterial invasion, to kill external P. gingivalis andthen cultured for a further 3, 6, 18 and 24 h. Viable P. gingi-valis were looked for extracellularly in the culture mediumwith maximal values being detected at 6 h for monolayer and24 h for OMM cultures (Fig. 4B). At these time points thelevels of P. gingivalis within the extracellular medium weresimilar for both monolayer and OMM with 23.34 8.52% and20.82 2.14% of the total intracellular extracellular bac-teria respectively. Therefore, the release of intracellularP. gingivalis into the culture medium appeared to be occurring

Fig. 2. Detection of P. gingivalis cells within cultured OMMs. P. gingivalis cells (un

OMM (A&D), air-exposed NOK-OMM (B) or submerged H357-OMM (C) for 4 h

and counter-stained with haematoxylin (AeC) or OMMs were cryosectioned and H

superficial layers of air-exposed H357-OMM (A) and deeper within the OMMs

submerged H357-OMM (C). Staining of P. gingivalis within NOK-OMM (B), arrow

three independent experiments.Infection 16 (2014) 310e319in both monolayer and OMM systems. This may not be due toactive release by epithelial cells alone but also to exfoliation ofthe infected epithelial cells into the culture medium. Indeed,exfoliated cells were detectable in the medium of bothmonolayer and OMM at similar levels (10.0 2.50% and10.0 2.80% respectively, of the original epithelial countfrom monolayer and on the surface of OMM) over a 48 hperiod (Fig. 4C). It should be noted that P. gingivalis could notgrow in the culture medium used for the monolayer or OMMculture systems (data not shown).

3.3. Chemokine/cytokine release from monolayer andOMM in response to intracellular P. gingivalis

Periodontitis is an inflammatory disease and the release ofpro-inflammatory chemokines and cytokines by the oralepithelium has previously been demonstrated [28]. However,the literature is ambiguous concerning the contribution ofP. gingivalis to the chemokine load, and in particular whateffect intracellular P. gingivalis have on the detection of thesechemokines/cytokines [8,29]. Here, the contribution of intra-cellular P. gingivalis (following killing of external bacteriawith metronidazole) on chemokine/cytokine release wasdetermined using a semi-quantitative antibody array. NOKswere chosen for this assay, rather than H357, because theresponse of NOK in terms of the chemokine/cytokine release

labelled (AeC) or FITC-labelled (D)) were incubated with air-exposed H357-

. OMMs were then stained using a P. gingivalis-specific antibody (clone 1B5)

oechst used to stain the epithelial cell nuclei (D). Staining was detected in the

(arrows in C&D). Penetration of P. gingivalis into the connective tissue of

s indicate staining of intracellular bacteria. Images are representative of at least

ndA

B

3

4

5

6

ion

(%)

0

1

2

3

4

5

6

H357 NOK

Inva

sion

(%)

MonolayerOMMP

J

nd- P. gingivalis

AA B C D E F G H I

316 A. Pinnock et al. / Microbes amonolayers [8,13], it was hypothesised that organotypiccultures of the oral epithelium could provide more insightinto bacterial invasion and host responses of oral epithelialcells. In this study we have shown that there are markeddifferences in the total percentage invasion, intracellularviability, bacterial release and cytokine profile following

Monolayer

OMM

Chemokine/Cytokine

Also known

CXCL8 Interleukin 8

IL-1 Interleukin 1

IL-6 Interleukin 6

CCL2 Monocyte chemotactic protein 1 (MC

CXCL10 Interferon-gamma-inducible protein 1

CCL5 Regulated upon activation, normal Tsecreted (RANTES)

TIMP-2 Tissue inhibitor of metalloprotease 2

TNF- Tumour necrosis factor

A B C D E F G H I J

B

C

A B C D E F1 Pos Pos Neg Neg Eotaxin Eotaxin 22 Pos Pos Neg Neg Eotaxin Eotaxin 23 IL-1 IL-2 IL-3 IL-4 IL-6 IL-6sR4 IL-1 IL-2 IL-3 IL-4 IL-6 IL-6sR5 IL-13 IL-15 IL-16 IL-17 IP-10 MCP-16 IL-13 IL-15 IL-16 IL-17 IP-10 MCP-17 RANTES TGF-1 TNF- TNF- s TNF R1 s TNF RII8 RANTES TGF-1 TNF- TNF- s TNF R1 s TNF RII

Fig. 5. Cytokine immunoblot of NOK-monolayer and NOK-OMM stimulated with

pre-stimulated with TNF-a for 4 h. Monolayers and OMMs were exposed to TNF-

OMMs were washed and incubated with metronidazole for 4 h during which time cy

dot analysed. Using the internal positive control (Pos), the relative density was ca

Immunoblots (A), location of cytokine antibodies on the array membranes (B), f

representative of duplicate experiments.+ P. gingivalis K L A B C D E F G H I J K L

Infection 16 (2014) 310e319epithelial cell invasion of monolayer and OMM byP. gingivalis.

Calculating the percentage invasion of bacteria intoepithelial cells is commonly used to investigate trends inexperimental data; for example, to investigate which strains ofbacteria invade cells at a greater level than others and the

as Fold Change

Monolayer OMM

-0.7 0.5

2.3 -0.1

-0.1 0.3

P-1) -0.1 -0.6

0 (IP-10) -0.1 -0.1

cell expressed and -0.1 -0.6

-0.1 -0.1

-0.1 -0.1

K L A B C D E F G H I J K L

G H I J K LGCSF GM-CSF ICAM-1 IFN- I-309 IL-1GCSF GM-CSF ICAM-1 IFN- I-309 IL-1IL-7 IL-8 IL-10 IL-11 IL-12 p40 IL-12 p70IL-7 IL-8 IL-10 IL-11 IL-12 p40 IL-12 p70

MCP-2 M-CSF MIG MIP-1 MIP-1 MIP-1MCP-2 M-CSF MIG MIP-1 MIP-1 MIP-1

PDGF-BB TIMP-2 Blank Blank Neg PosPDGF-BB TIMP-2 Blank Blank Neg Pos

TNF-a and P. gingivalis NCTC 11834. NOK monolayer and NOK OMM were

a with or without P. gingivalis for 1.5 h and 4 h respectively. Monolayers and

tokines were released. A cytokine array was performed and the density of each

lculated between eP. gingivalis and P. gingivalis for monolayer and OMM.old changes between infected and non-infected blots (C). Results shown are

ndeffect of therapeutic agents on bacterial survival. In the last15e20 years, epithelial cell invasion by P. gingivalis has beenshown by many research groups [3,4]. The most commonlyused epithelial cell in these studies has been KB (now shownto be a HeLa contaminated cell line) or NOK. NOK culturedfrom the oral mucosa retain some in vivo characteristicscompared with immortalised oral epithelial cells and cancercell lines, indicating that the use of NOK may be a morerepresentative in vitro model [3]. The levels of bacterial in-vasion shown in the present study for NOK and H357 cells aresimilar to those previously reported in the literature [12] butthey differed when grown as OMM. Indeed, in terms ofmodelling the oral mucosa more closely, organotypic modelshave recently been championed [8,9]. This is because of theirmulti-layered nature and the presence of a fibroblast-containing scaffold that is comparable with the oral mucosa.In this study, we found that the total percentage bacterial in-vasion of H357 monolayer and OMM cultures were similar,suggesting that either H357 monolayer or H357 OMM may besuitable to study experimental trends in invasion data. Despitethis, significant differences were shown for bacterial survivaland chemokine/cytokine release between monolayer andOMM, suggesting that while monolayer cultures may besuitable to investigate trends in experimental data, OMM maybe important for elucidating experimental end-points, such asprotein release, receptor expression and cellular viability.NOK OMM did not support as high a level of P. gingivalisinvasion as H357 OMM, which might be due to differences inthe expression, activation and/or polarity of epithelial cellsurface receptors, cellular differentiation and/or intracellularsignalling pathways of these two cell types.

Immunohistochemistry of OMM tissue sections indicatedthat P. gingivalis has the capacity to infiltrate the upper mostlayers of the epithelium of air-exposed NOK, and in particular,H357 OMM. This is in agreement with previously reportedliterature that shows the cellular invasion of P. gingivalisthrough multiple epithelial layers [8e10,30]. In addition, weobserved that P. gingivalis could enter the connectivecomponent of OMM but only in models containing a thin layer(4e5 cells thick) of epithelium, which is similar in thicknessto junctional epithelium at the base of the gingival pocket. Themovement of P. gingivalis through the epithelium and into theconnective tissue layer is particularly important becausein vivo the blood vessels are situated close to the epitheliallayers, presenting a risk of low level bacteraemia, which mayplay a role in atherosclerosis, endocarditis or other systemicdiseases [31]. Therefore, it appears that the fewer epithelialcell layers the bacterium is required to move through, thegreater the opportunity for the pathogen to breach host de-fences and to cause damage.

Intracellular survival is important for the continuation ofdisease; for example, intracellular pathogens may be expelledby cells allowing them to re-infect neighbouring cells. Currentdata is conflicting as to the length of time P. gingivalis cansurvive within human cells. Intracellular persistence has been

A. Pinnock et al. / Microbes areported for anything up to 4e8 days [15,16,29]. However, weshow that for both monolayer and OMM the intracellularviability of P. gingivalis significantly declines after 6 h toalmost zero by 48 h. This is in accordance with the findings ofLi et al. [14], who reported similar levels of intracellularP. gingivalis in KB (HeLa), endothelial and smooth musclecells. These authors showed a decline in intracellular bacteriato almost zero after 24e48 h in the numbers of intracellular P.gingivalis after 90 min incubation [14]. This rapid decline inintracellular viability may be due to the experimental condi-tions employed (e.g. the culture medium or the aerobic at-mosphere necessary to maintain viability of the human cells)or it is possible that some P. gingivalis cells enter an uncul-tivable state after several hours of intracellular occupancy[14]. Interestingly, we observed that the intracellular viabilityand persistence of P. gingivalis in OMM was greater than inmonolayers. Dickinson et al. [9] recently showed that P.gingivalis cells are able to spread from cell-to-cell. Since theOMM is a multi-layered epithelium, the opportunity to con-tact fresh cells may be greater in this model compared to cellsgrown as monolayers and so this may account for the apparentincreased survival in the OMM. As well as survival, weobserved release of P. gingivalis from both monolayers andOMM cultures and although we observed epithelial exfolia-tion in both model systems, which is in accordance withpreviously reported literature [32], it is uncertain as to whetherthis is sufficient to explain the presence of P. gingivalis in theculture medium over time or whether this represents activerelease. P. gingivalis has recently been reported to be activelyreleased from epithelial cells [33] but by either mechanism,bacteria would be able to seed either local or distant sites inthe oral cavity. Intracellular replication was not found usingthe experimental techniques employed in this study.

The first line of host defence against bacterial challenge isthe release of pro-inflammatory chemokines and cytokines.Here, we investigated whether there were similarities or dif-ferences in the pro-inflammatory chemokine/cytokine releasein the presence of intracellular P. gingivalis produced bymonolayer and OMM. In accordance with previous studies,we observed a decrease in the majority of chemokines/cyto-kines detected from both the monolayer [34,35] and OMMcultures in the presence of this organism [9]. This attenuationof the host response has been suggested to be important forthe survival of P. gingivalis at the site of infection bydepressing the host immune response [18]. Chemokine andcytokine release was measured in infected monolayer andOMM after removal of extracellular P. gingivalis followingtreatment with metronidazole. Protease inhibitors were notemployed because we found that these dramatically decreasedthe epithelial cell viability over this time period. Notwith-standing these issues, not all chemokines and cytokines werereduced upon prolonged exposure to P. gingivalis. Indeed, forOMM, there was a slight increase in the detection of CXCL8and IL-6 compared with the un-stimulated control andmonolayer cultures. This is in accordance with previouslyreported data from an organotypic culture model using palatalepithelial cells [8]. This lack of attenuation of these chemo-

317Infection 16 (2014) 310e319kines in OMM compared with monolayers may be due to theaccumulation of these molecules in the multi-cellular layers of

ndthe OMM and consequent relative protection against the ac-tion of bacterial proteases or perhaps due to intrinsic differ-ences between the physiology of the two model systems.While it is possible that the action of gingipains might bereduced in the aerobic conditions of these tissue culture sys-tems, previous workers have reported that most of the pro-teolytic activity of P. gingivalis cultures was retained in thepresence of oxygen [36] and that gingipains are active intissue culture systems incubated aerobically [8]. We cannotrule out that some of the cytokines secreted may have origi-nated from oral fibroblasts within the OMM. However, ex-periments comparing stimulated OMM containing fibroblaststo those without fibroblasts show that these cells contributeonly a small increase in CXCL8 production (approximately0.4-fold) but no apparent contribution to the secretion of IL-6(data not shown). Therefore, these data suggest that the fi-broblasts may only play a minor role in chemokine/cytokinecontribution in the OMM and their presence seems unlikely toexplain the lack of attenuation of these cytokines comparedwith monolayer cultures. Additionally, the chemokine/cyto-kine contribution from fibroblasts, as part of OMM, is anintrinsic feature that qualifies these models as more repre-sentative of the oral tissues.

In summary, we have shown that monolayer cultures maybe suitable to determine trends in experimental data on cellinvasion by P. gingivalis. However, when requiring absoluteend-point values, the use of organotypic models are a morerelevant in vitro model of the cellular microenvironmentencountered by P. gingivalis in vivo.

Acknowledgements

The authors would like to thank GlaxoSmithKline forfunding this work.

References

[1] Choi JI, Seymour GJ. Vaccines against periodontitis: a forward-looking

review. J Periodontal. Implant Sci 2010;40:153e63.[2] Rautemaa R, Jarvensivu A, Kari K, Wahlgren J, DeCarlo A,

Richardson M, et al. Intracellular localization of Porphyromonas gingi-

valis thiol proteinase in periodontal tissues of chronic periodontitis pa-

tients. Oral Dis 2004;10:298e305.

[3] Lamont RJ, Chan A, Belton CM, Izutsu KT, Vasel D, Weinberg A.

Porphyromonas gingivalis invasion of gingival epithelial cells. Infect

Immun 1995;63:3878e85.[4] Belton CM, Izutsu KT, Goodwin PC, Park Y, Lamont RJ. Fluorescence

image analysis of the association between Porphyromonas gingivalis and

gingival epithelial cells. Cell Microbiol 1999;1:215e23.

[5] Duncan MJ, Nakao S, Skobe Z, Xie H. Interactions of Porphyromonas

gingivalis with epithelial cells. Infect Immun 1993;61:2260e5.

[6] Rudney JD, Chen R, Sedgewick GJ. Actinobacillus actino-

mycetemcomitans, Porphyromonas gingivalisand Tannerella forsythensis

are components of a polymicrobial intracellular flora within human

buccal cells. J Dent Res 2005;84:59e63.

[7] MoharamzadehK, Colley H,Murdoch C,HearndenV, ChaiWL, Brook IM,

et al. Tissue engineered oral mucosa. J Dent Res 2012;91:642e50.[8] Andrian E, Grenier D, Rouabhia M. In vitro models of tissue penetration

and destruction by Porphyromonas gingivalis. Infect Immun 2004;72:

318 A. Pinnock et al. / Microbes a4689e98.[9] Dickinson BC, Moffatt CE, Hagerty D, Whitmore SE, Brown TA,

Graves DT, et al. Interaction of oral bacteria with gingival epithelial cell

multilayers. Mol Oral Microbiol 2011;6:210e20.

[10] Wayakanon K, Thornhill MH, Douglas CWI, Lewis AL, Warren NJ,

Pinnock A, et al. Polymersome-mediated intracellular delivery of anti-

biotics to treat Porphyromonas gingivalis-infected oral epithelial cells.

FASEB J 2013;27:4455e65.

[11] Saito AKE, Inagaki S, Imamura K, Kita D, Lamont RJ, Ishihara K.

Porphyromonas gingivalis entry into gingival epithelial cells modulated

by Fusobacterium nucleatum is dependent on lipid rafts. Microb Pathog

2012;53:234e42.

[12] Stafford P, Higham J, Pinnock A, Murdoch C, Douglas CW, Stafford GP,

Lambert DW. Gingipain-dependent degradation of mammalian target of

rapamycin pathway proteins by the periodontal pathogen Porphyromonas

gingivalis during invasion. Mol Oral Microbiol 2013;28:366e78.

[13] Dabija-Wolter G, Sapkota D, Cimpan MR, Neppelberg E, Bakken V,

Costea DE. Limited in-depth invasion of Fusobacterium nucleatum into

in vitro reconstructed human gingiva. Arch Oral Biol 2012;57:344e51.

[14] Li L, Cohen MR, deCarlo A, Kozarov E. Intracellular survival and

vascular cell-to-cell transmission of Porphyromonas gingivalis. BMC

Microbiol 2008;8:26.

[15] Houalet-Jeanne S, Pellen-Mussi P, Tricot-Doleux S, Apiou J, Bonnaure-

Mallet M. Assessment of internalization and viability of Porphyromonas

gingivalis in KB epithelial cells by confocal microscopy. Infect Immun

2001;69:7146e51.

[16] Madianos PN, Papapanou PN, Nannmark U, Dahlen G, Sandros J. Por-

phyromonas gingivalis FDC381 multiplies and persists within human

oral epithelial cells in vitro. Infect Immun 1996;64:660e4.

[17] Sandros J, Papapanou P, Dahlen G. Porphyromonas gingivalis invades

oral epithelial cells in vitro. J Periodontal Res 1993;28:219e27.[18] Darveau RP, Belton CM, Reife RA, Lamont RJ. Local chemokine pa-

ralysis, a novel pathogenic mechanism for Porphyromonas gingivalis.

Infect Immun 1998;66:1660e5.

[19] OBrien-Simpson NM, Pathirana RD, Walker GD, Reynolds EC. Por-

phyromonas gingivalis RgpA-Kgp proteinase-adhesin complexes pene-

trate gingival tissue and induce proinflammatory cytokines or apoptosis

in a concentration-dependent manner. Infect Immun 2009;77:1246e61.

[20] Yao L, Jermanus C, Barbetta B, Choi C, Verbeke P, Ojcius DM, et al.

Porphyromonas gingivalis infection sequesters pro-apoptotic Bad

through Akt in primary gingival epithelial cells. Mol Oral Microbiol

2010;25:89e101.

[21] Curtis MA, Hanley SA, Aduse-Opoku J. The rag locus of Porphyromonas

gingivalis: a novel pathogenicity island. J Periodontal Res 1999;34:

400e5.

[22] Curtis M, Slaney J, Carman R, Johnson N. Identification of the major

surface protein antigens of Porphyromonas gingivalis using IgG antibody

reactivity of periodontal case control serum. Oral Microbiol Immunol

1991;6:321e6.

[23] Colley HE, Hearnden V, Jones AV, Weinreb PH, Violette SM, MacNeil S,

et al. Development of tissue engineered models of oral dysplasia and

early invasive oral squamous carcinoma. Br J Cancer 2011;105:

1582e92.

[24] Rajan N, Habermehl J, Cote MF, Doillon CJ, Mantovani D. Preparation

of ready-to-use, storable and reconstituted type I collagen from rat tail

tendon for tissue engineering applications. Nat Protoc 2007;1:2753e8.

[25] Dongari-Bagtzoglou A, Kashleva H. Development of a highly repro-

ducible three-dimensional organotypic model of the oral mucosa. Nat

Protoc 2006;1:2012e8.

[26] Suwannakul S, Stafford GP, Whawell SA, Douglas CW. Identification of

bistable populations of Porphyromonas gingivalis that differ in epithelial

cell invasion. Microbiology 2010;156:3052e64.

[27] Socransky SS, Haffajee AD, Goodson JM, Lindhe J. New concepts of

destructive periodontal disease. J Clin Periodontol 1984;11:21e32.

[28] Stadnyk A. Cytokine production by epithelial cells. FASEB J

1994;8:1041e7.

[29] Eick S, Reissmann A, Rodel J, Schmidt KH, Pfister W. Porphyromonas

Infection 16 (2014) 310e319gingivalis survives within KB cells and modulates inflammatory

response. Oral Microbiol Immunol 2006;21:231e7.

[30] Papapanou PN, Sandros J, Lindberg K, Duncan MJ, Niederman R,

Nannmark U. Porphyromonas gingivalis may multiply and advance

within stratified human junctional epithelium in vitro. J Periodontal Res

1994;29:374e5.

[31] Hayashi C, Viereck J, Hua N, Phinikaridou A, Madrigal AG,

Gibson III FC, et al. Porphyromonas gingivalis accelerates inflammatory

atherosclerosis in the innominate artery of ApoE deficient mice.

Atherosclerosis 2011;215:52e9.[32] Dierickx K, Pauwels M, van Eldere J, Cassiman JJ, van Steenberghe D,

Quirynen N. Viability of cultured periodontal pocket epithelium cells and

Porphyromonas gingivalis association. J Clin Periodontol

2002;29:987e96.

[33] Takeuchi H, Furuta N, Morisaki I, Amano A. Exit of intracellular Por-

phyromonas gingivalis from gingival epithelial cells is mediated by

endocytic recycling pathway. Cell Microbiol 2011;13:677e91.

[34] Stathopoulou PG, Benakanakere MR, Galicia JC, Kinane DF. The host

cytokine response to Porphyromonas gingivalis is modified by gingi-

pains. Oral Microbiol Immunol 2009;24:11e7.

[35] Huang GTJ, Kim D, Lee JKH, Kuramitsu HK, Haake SK. Interleukin-8

and intercellular adhesion molecule 1 regulation in oral epithelial cells by

selected periodontal bacteria: multiple effects of Porphyromonas gingi-

valis via antagonistic mechanisms. Infect Immun 2001;69:1364e72.

[36] Diaz PI, Rogers AH. The effect of oxygen on the growth and physiology

of Porphyromonas gingivalis. Oral Microbiol Immunol 2004;19:88e94.

319A. Pinnock et al. / Microbes and Infection 16 (2014) 310e319

Characterisation and optimisation of organotypic oral mucosal models to study Porphyromonas gingivalis invasion1 Introduction2 Methods and materials2.1 Materials2.2 Bacterial strains and culture conditions2.3 Epithelial cell culture conditions2.4 Oral mucosal models (OMMs)2.5 Antibiotic protection assay2.6 Immunohistochemistry2.7 Immunofluorescence2.8 Chemokine array2.9 Statistical analysis

3 Results3.1 Invasion of monolayer and OMM by P. gingivalis3.2 Intracellular survival and bacterial release of P. gingivalis from H357 monolayer and OMM cultures3.3 Chemokine/cytokine release from monolayer and OMM in response to intracellular P. gingivalis

4 DiscussionAcknowledgementsReferences