J Periodontol December 2009

Gingival Transcriptome Patterns During Induction and Resolution of Experimental Gingivitis in Humans
Steven Offenbacher,* Silvana P. Barros,* David W. Paquette,* J. Leslie Winston, Aaron R. Biesbrock, Ryan G. Thomason, Roger D. Gibb, Andy W. Fulmer, Jay P. Tiesman, Kenton D. Juhlin, Shuo L. Wang, Tim D. Reichling, Ker-Sang Chen, and Begonia Ho

Background: To our knowledge, changes in the patterns of whole-transcriptome gene expression that occur during the induction and resolution of experimental gingivitis in humans were not previously explored using bioinformatic tools. Methods: Gingival biopsy samples collected from 14 subjects during a 28-day stent-induced experimental gingivitis model, followed by treatment, and resolution at days 28 through 35 were analyzed using gene-expression arrays. Biopsy samples were collected at different sites within each subject at baseline (day 0), at the peak of gingivitis (day 28), and at resolution (day 35) and processed using whole-transcriptome gene-expression arrays. Gene-expression data were analyzed to identify biologic themes and pathways associated with changes in gene-expression proles that occur during the induction and resolution of experimental gingivitis using bioinformatic tools. Results: During disease induction and resolution, the dominant expression pathway was the immune response, with 131 immune response genes significantly up- or downregulated during induction, during resolution, or during both at P <0.05. During induction, there was signicant transient increase in the expression of inammatory and oxidative stress mediators, including interleukin (IL)-1 alpha (IL1A), IL-1 beta (IL1B), IL8, RANTES, colony stimulating factor 3 (CSF3), and superoxide dismutase 2 (SOD2), and a decreased expression of IP10, interferon inducible T-cell alpha chemoattractant (ITAC), matrix metalloproteinase 10 (MMP10), and beta 4 defensin (DEFB4). These genes reversed expression patterns upon resolution in parallel with the reversal of gingival inammation. Conclusions: A relatively small subset (11.9%) of the immune response genes analyzed by array was transiently activated in response to biolm overgrowth, suggesting a degree of specicity in the transcriptome-expression response. The fact that this same subset demonstrates a reversal in expression patterns during clinical resolution implicates these genes as being critical for maintaining tissue homeostasis at the biolmgingival interface. In addition to the immune response pathway as the dominant response theme, new candidate genes and pathways were identied as being selectively modulated in experimental gingivitis, including neural processes, epithelial defenses, angiogenesis, and wound healing. J Periodontol 2009;80:1963-1982. KEY WORDS Gene expression; gene expression regulation; gingivitis; inammation; transcriptomes.

* Department of Periodontology, School of Dentistry, University of North Carolina at Chapel Hill, Chapel Hill, NC. Procter & Gamble, Mason, OH.

ver 40 years ago, Loe et al.1 performed a classic experimental gingivitis study in which subjects refrained from daily tooth cleaning, and over a period of 3 weeks, the dental microbial biolm (plaque) accumulated, and clinical gingivitis developed. Reinstitution of oral hygiene measures reversed plaque accumulation and led to a rapid resolution of gingival inammation and a return to gingival health. At the time, these experiments2,3 provided the rst direct human evidence of the microbial etiology of gingivitis. More recently, the experimental gingivitis model was adapted to include the use of intraoral stents that cover selected teeth to be worn only during routine oral hygiene to permit the study of localized changes in biolm overgrowth and inammation.4 The effects of the stent-induced biolm overgrowth (SIBO) model are short-term, with
doi: 10.1902/jop.2009.080645

1963

Immune Response Adaptation to Stress of Biolm Emergence

Volume 80 Number 12

reversible clinical changes, and because the oral cavity is easily accessible, it represents an ideal model to study the host response to the commensal biolm.5-8 During the gingivitis-induction phase, the SIBO model enables the examination of the cellular and molecular aspects of the host response as it adapts to a microbial stress elicited by overgrowth and emergence of the naturally occurring subgingival biolm. The purpose of this investigation was to perform whole-transcriptome gene-expression analyses of gingival tissues biopsied at different time points during the induction and resolution phases of gingival inammation. This model permitted us to examine the reversible host adaptation to the biolm maturation that resulted in gene-expression patterns that changed for each subject from baseline during the induction phase, and which were reversed during the resolution phase, paralleling the localized changes in clinical signs of inammation. Thus, the goal of the study was to create a descriptive summary of the signicant geneexpression changes during the induction and resolution of gingivitis and to characterize the major gene-expression pathways using bioinformatics. This analysis provides a survey of the associations between gene-expression and clinical signs over time. In this report, we describe the SIBO model, serving as a mucosal immunity stress test, to characterize the early stress responses to the overgrowth of the commensal microbes at the biolmgingival interface and provide insight into the range of molecular responses that are likely to be involved with the maintenance of homeostasis between the local host tissue and the biolm. MATERIALS AND METHODS Overall Design of the Study Eighteen subjects were enrolled in this study conducted in Chapel Hill, North Carolina, and 14 subjects (10 females and 4 males; aged 20 to 63 years) completed the experimental protocol between March 27, 2006 and August 16, 2006. Eighteen normal, systemically healthy individuals with mild gingivitis were enrolled in an experimental gingivitis study that included an initial prophylaxis with a 4-week stent-induced, biolm-overgrowth, experimental gingivitis induction phase followed by a 1-week hygiene-resolution phase. Four weeks duration was chosen, rather than the traditional 3-week (full-mouth) induction phase initially described by Loe et al.,1 to enable the full expression of gingivitis within most subjects using a half-mouth biolm overgrowth model. Gum tissue was collected at three time points: 1) baseline; 2) 4 weeks after the use of an oral stent to induce biolm emergence and a state of clinical inammation; and 3) 1 week after a prophylaxis and the reinstitution of oral
1964

hygiene to reverse the state of gingivitis. Tissue biopsy samples were extracted for RNA, and suitable gene-chip data was obtained for 14 subjects. In this article, the analyses are limited to exploring geneexpression patterns that co-vary with the overall clinical responses over time. Clinical Protocol The protocol and informed consent were approved by the Institutional Review Board at the University of North Carolina at Chapel Hill. After obtaining written informed consent, the subjects were screened for eligibility. Eligibility criteria included: adult male or females; age range: 18 to 65 years; 20 teeth (excluding third molars); and at least some mild gingivitis, dened as 20 periodontal sites with bleeding upon gentle probing. Exclusion criteria included major medical conditions and pregnancy. There were three phases to the study: during the hygiene phase (days -14 to 0), subjects with naturally occurring gingivitis (20 sites with bleeding on probing [BOP] without any probing depth [PD] >5 mm) were brought to optimal gingival health with mechanical scaling, tooth polishing, and oral hygiene instructions. Each subject was issued toothpaste, oss, and a power toothbrush. During the no-hygiene phase (days 0 to 28), subjects refrained from all oral hygiene procedures in two quadrants (up to eight teeth in each arch) but were permitted to continue plaque-control procedures in the two remaining quadrants using the dentifrice and oss dispensed at screening. During daily brushing procedures, subjects wore removable, customized acrylic stents to cover the two experimental quadrants to encourage subject compliance. The resolution phase (days 28 to 35) began when, after clinical examinations and biologic sampling, all subjects received scaling, tooth polishing, and oral hygiene instructions to reestablish gingival health. Subjects were followed another 7 days for biologic sampling, and if resolution was not complete at day 35, subjects received additional hygiene instructions and were followed until they exhibited clinical resolution of the induced gingivitis. Single interproximal biopsy samples of gingival epithelium and connective tissues were obtained from subjects on days 0, 28, and 35. Additional details of the experimental design, clinical measurements, tissue sampling, and processing of biologic samples are described in the online Journal of Periodontology (supplementary Methods, section M1). Description of the transcriptome analyses, quality control assessment of genechip array data, statistical methods for determining signicant visit differences in expression patterns, correlations with clinical signs and methods for
Oral-B Professional Care, Braun, Kronberg, Germany.

J Periodontol December 2009

Offenbacher, Barros, Paquette, et al.

Figure 1.
Signicant increases in all clinical signs from baseline during 4-week gingivitis induction with return to baseline at resolution (day 35) for the teeth that were under the stent . Clinical parameters during the oral hygiene phase (days -14 to 0), 28-day SIBO phase (days 0 to 28), and resolution phase (days 28 to 35). A) PI, (B) GI, and (C) BI. Error bars represent one standard error of the mean. Index values are dened in supplementary Methods, section M1.

theme extraction are described in supplementary Methods, section M2. Transcriptome Analyses: Filtering Genes Based on Patterns of Expression We chose a priori to take advantage of the repeatedsampling design of our study (in which each subject was analyzed three times) to identify those genes and gene pathways that demonstrated either signicant upregulation or downregulation during the gingivitisinduction phase and during resolution after treatment. Thus, we sought to examine for changes in gene expression over time for each subject and to pool those changes to identify trends in gene expression. We created the gene-set bins using the following denitions: 1) the induced gene set = genes that displayed either a statistically signicant upregulation or downregulation during the induction phase comparing baseline day 0 to day 28; 2) the resolution gene set = genes that displayed either a statistically signicant upregulation or downregulation during the resolution phase comparing day 28 to day 35; 3) the sustained gene set = induced gingivitis genes that did not show a statistically signicant return toward baseline values by the resolution visit at day 35 (i.e., those genes that remained either elevated or suppressed after treatment); 4) the transient gene set = induced genes (dened in denition 1 above) that showed a statistically signicant return toward baseline values by the resolution visit at day 35 and, therefore, were highly correlated with the changes in gingival indices; and 5) the healing gene set = genes that remained unchanged from baseline during the induction phase but either signicantly increased or decreased in expression during the resolution phase (at day 35) as a result of treatment (at day 28).

RESULTS Study Execution and Clinical Outcomes Overall, the subjects displayed a statistically signicant change in clinical inammation (gingival index [GI] and bleeding index [BI]) and plaque scores (plaque index [PI]) that increased from baseline (day 0) to a maximum at 4 weeks (day 28) with a return to baseline at 5 weeks (day 35) (Fig. 1) for those teeth that were under a stent. For example, the mean GI changed from a baseline value of 0.78 0.09 to 1.34 0.07 at day 28, returning to 0.83 0.09 at resolution (P <0.0001 and P = 0.0003 [paired-difference t test], respectively). This change in clinical phenotype was induced and closely coupled to the increase in biolm overgrowth, as reected in the parallel shifts in plaque scores transitioning from 0.21 0.06 at baseline to 1.18 0.13 at 4 weeks and returning to 0.17 0.06 at resolution (P <0.0001 and P <0.0001, respectively). Gene-Expression Patterns The time-dependent changes in expression levels of specic transcripts and pathways organized into three broad categories of response are described in the online Journal of Periodontology (Fig. 1 in supplementary Results, section R1). Overall, compared to baseline, there were 4,108 probe sets at induction and 3,026 probe sets at resolution that showed statistical signicance, with 3,984 showing signicant changes from induction to resolution (at the a = 0.05 level). Overall, there were a total of 9,194 genes that demonstrated signicant time-dependent changes in this model. For the gene chip that was used in this study, 2,731 probe sets (95% condence
HG-U133 Plus 2.0 GeneChip, Affymetrix, Santa Clara, CA.

1965

Immune Response Adaptation to Stress of Biolm Emergence

Volume 80 Number 12

Figure 2.
Transient response gene set that covaried with clinical signs of inammation during the induction and resolution phases of the SIBO model. Pathways were constructed using 1,106 probe sets representing 853 genes that demonstrated an up/down response and 707 probe sets representing 598 genes that displayed a down/up expression pattern during induction and resolution, respectively. A theme is represented as a node in the network with colored lines highlighting the interactions appearing between nodes. If two themes share genes, then a line was drawn to connect the two nodes. For the more complex themes that contain more than one subtheme (i.e., immune response, neural processes, wound healing, bone, vasculature, epithelial tissue, and broblasts and extracellular matrix [ECM]), the theme is represented as a node cluster, where the subthemes are gathered in a box, with common color coding. In this high-level view of the theme network, linkages between the immune response and other themes (except neural processes) are turquoise, whereas linkages between neural processes and other themes (except the immune response) are magenta. Linkages between the immune response and neural processes are coded orange.

internal: 2,631 to 2,830) were expected to show statistical signicance due to chance alone at the a = 0.05 level. Genes that signicantly changed during induction but returned to baseline were designated transient response genes (supplementary Fig. 1). Genes that signicantly changed expression during induction but did not return to baseline by resolution were designated sustained response genes, and those that remained unchanged during induction but had either increased or decreased expression during resolution were termed healing response genes. Detailed ndings of those three categories of gene responses are presented in the online Journal of Periodontology (supplementary Results, section R1). Within the transient response gene set, we identied 24 major gene-expression themes summarized as
1966

a theme network (Fig. 2). The network highlights the dramatic extent to which components of the immune response and neural processes are linked to each other and to the other main themes extracted from these data. The genes represented in Figure 2 are listed in supplementary Table 1. The immune response was the dominant thematic pathway with 19 nodes represented, followed by neural processes with 15 nodes. In addition to linkages with neural processes, the immune response linked extensively to epithelial tissue (six nodes) and wound healing (six nodes) with activation of hostbacterial and hostviral pathways, as well as calcium ion signaling pathways. Fewer immune response pathways were linked to vasculature (eight nodes; in red in Fig. 2) and chemotaxis. The role of neural activation

J Periodontol December 2009

Offenbacher, Barros, Paquette, et al.

pathways with 15 nodes linking broadly to epithelial tissues, vasculature, and wound healing, represents a novel and unanticipated nding for gingivitis. Neural processing was moderately upregulated, including pathways associated with transmission of nerve impulses, nerve development, and maturation, as well as neurodegenerative disorders. Fibroblast and extracellular matrix activation (10 nodes; in green in Fig. 2) was not driven primarily by immune response modulators but, rather, was modulated by a variety of pathways from epithelial, zinc, vasculature, and wound-healing signals. The major system-based, gene ontology, categoric themes that were signicantly modulated during the stress of biolm-induced overgrowth included the immune response, tissue morphology, hematologic development and function, and organ development. There are ve dominant molecular and cellular functional biologic themes that emerged during induction, including cell movement, cell-tocell signaling and interaction, molecular transport, cell cycle, and amino acid metabolism. The majority of the biologic themes overlapped with the immune response category, conrming the central importance of this category in the model. For this reason, the immune response gene set was selected for primary analysis of the transient-gene responses. These genes, associated with the immune response during induction, appear in the online Journal of Periodontology (supplementary Results, section R2, Table 2). The strong modulation of the immune response included subthemes of hostbacterial and hostviral interactions, chemotaxis and transendothelial migration, oxidative stress, and wound healing. In supplementary Table 2, the immune response genes that were associated with either induction or resolution, using a false-discovery rate <0.05 and a P value <0.05, are listed and organized by cellular compartmentalization as sorted by ingenuity pathway analyses (IPA) (see supplementary Methods, theme extraction, section M2). There were 57 immune response genes (50 transient and seven sustained) that demonstrated signicantly altered expression during induction and resolution. In addition, there were 100 genes that showed statistically signicant changes during resolution, including 26 genes that are common to induction, to create an immune response gene set of 131 genes that signicantly changed during experimental gingivitis. The immune response genes that showed signicant changes and the fold differences during induction and resolution appear in supplementary Table 2. The 131 immune response genes signicantly upor downregulated during induction, during resolution, or during both at P <0.05 are listed in supplementary

Table 2 and graphically depicted in Figures 3A and 3B to highlight the changes that occurred during induction and resolution, respectively. The 57 genes associated with the immune response that demonstrated signicant modulation during induction are shown in Figure 3A. This represents 11.9% of the 477 immune response genes on the array (GO:0006965), suggesting some degree of specicity in the activation of the immune response. Genes with a signicantly increased expression during induction compared to baseline are shaded red, whereas genes that showed signicant suppression are shaded green in Figure 3A. Those genes that did not show statistical signicance during induction but signicantly changed during resolution are shaded gray in Figure 3A, and the opposite shading is shown in Figure 3B (i.e., those genes that changed during induction and did not change during resolution are shaded gray in Fig. 3B). Some genes that were signicant in the pathway but not independently statistically signicant are shaded gray in both gures and increased the total number of transcripts depicted from 131 to 199. Although the changes during induction or resolution of those genes shaded gray were not statistically signicant, the magnitude and the direction of the change often reversed from induction to resolution. Supplementary Table 2 can serve as a legend key to understand the gene responses depicted in Figures 3A and 3B, as it shows the magnitude of change and P values during induction and resolution for all 131 immune response genes. Furthermore, each gene was illustrated to indicate the compartmentalization of the gene product (either extracellular, plasma membrane, cytoplasmic, mitochondrial, or nuclear, as identied using IPA), and the different shapes indicate the nature of the gene product (e.g., enzyme, receptor, and transcriptional factor). The changes in expression in the 57 genes that occurred during induction can be followed during the resolution phase depicted in Figure 3B. Comparing Figures 3A to 3B (or supplementary Table 2) shows that gene transcripts like CXCL11 and epidermal growth factor (EGF), which were suppressed (green) during induction, became signicantly increased during resolution (red). Other genes, like CCL5 and interleukin (IL)-1 alpha (IL1A), that increased during induction were signicantly downregulated during resolution. As mentioned, all genes that changed during induction shifted toward baseline levels at resolution (Fig. 3A). Certain genes, like IL1B showed a non-signicant increase during induction (increase of 1.72-fold; P = 0.06), but the decrease during resolution demonstrated statistical signicance (0.76; P = 0.02); thus, IL1B changes likely contributed to the observed changes in clinical signs.
1967

Immune Response Adaptation to Stress of Biolm Emergence

Volume 80 Number 12

Figure 3.
Changes in the patterns of immune response gene expression during induction and resolution of gingivitis. A) Induction phase. B) Resolution. The color coding reects the changes in gene expression during induction (day 28 versus day 0) and resolution (day 35 versus day 28). Red reects increased expression, gray reects a non-signicant change, and green reects downregulation. The illustration depicts the localization of the gene products by cellular compartmentalization: extracellular, plasma membrane, cytoplasm, mitochondrial, and nuclear. Genes are clustered by category: cytokines, chemokines, growth factors and adipokines, adhesion and extracellular matrix (ECM) proteins, microbial pattern recognition molecules, immunoglobulin domains and T-cell activation proteins, complement component binding, and arachidonic acid metabolite synthesis and receptors. Values for fold changes during induction and resolution for individual gene products appear in supplementary Results, section R2, Table 1. The shape of the gene product indicates the nature of the gene product (i.e., enzyme, receptor, transcriptional factor). = cytokine/growth factor; = enzyme; = G-protein coupled receptor; = growth factor; = kinase; = ligand-dependent nuclear receptor; = peptidase; = phosphatase; = transcription regulator; = transmembrane receptor; = transporter; = unknown.

A Deep Dive Into the Immune, Oxidative Stress, and Wound Healing Response Themes The major gene ontology categories associated with the SIBO response include the immune response gene set described above involving the functional pathways of chemotaxis, transendothelial migration of leukocytes, hostbacterial and hostviral interaction, and T- and B-cell activation. The second dominant category response relevant to experimental gingivitis was the response to stimuli that includes oxidative stress and wound healing. Gene changes that were associated with these two major themes are organized
1968

based upon functional pathways as shown in Table 1 (for up/down genes) and Table 2 (for down/up genes). These tables summarize functional attributes and pathway involvement for each gene product and were organized based upon pathway and listed in decreasing magnitude of fold change and P values within each pathway. Relevant genes subdivided by thematic pathways appear in Tables 1 and 2 and were presented for chemotaxis, transendothelial migration, hostbacterial and hostviral interaction, T- and B-cell activation, oxidative stress, and wound healing/matrix metabolism. Gene lists were organized by 1)

J Periodontol December 2009

Offenbacher, Barros, Paquette, et al.

Figure 3.
(Continued).

theme, 2) P value, and 3) fold change. For genes with more than one transcript, each one is shown separately. DISCUSSION To the best of our knowledge, this article is unique in that it provides a dynamic analysis of the periodontal tissue gene-expression pattern during the induction and resolution of stent-induced experimental gingivitis in humans. The repeated-sampling design, collecting tissues at baseline (day 0), at the peak of stent-induced gingivitis (day 28), and at resolution (day 35), enabled us to ascertain the changes in gene-expression patterns within each subject. During biolm overgrowth, the dominance in those gene-activation pathways that were associated with the immune response, neural processes, vasculature activation, and specic cell signaling was reported. An advantage of this study design was that these samples were collected from the same individuals

over time, enabling us to evaluate how gene-expression patterns relate to individual clinical changes and whether these changes can be generalized across patients. The time-dependent changes in gene expression seen in the model provided an analytical approach for evaluating the dynamic transcriptome changes into the transient, sustained, and healing response gene sets. The gene-expression experiments described in this article were conducted using rigorous experimental design parameters based upon signicant biologic replication and robust qualitycontrol metrics.9,10 However, the generalizability of this study is limited because there were relatively few subjects involved, and potential race-, gender-, and age-specic differences in gene-expression patterns could not be fully considered. A second limitation of the study is that mRNA levels were not conrmed by polymerase chain reaction (PCR) methods. This conrmatory step is often performed on a specic subset of selected genes of interest
1969

Immune Response Adaptation to Stress of Biolm Emergence

Volume 80 Number 12

Table 1.

Gingival Tissue Response Genes Displaying Increased Expression During Induction of Experimental Gingivitis and Decreased Expression at Resolution
Gene PROK2 CSF3 Name {Gene Ontology Pathway} Prokineticin 2 {1,7} CSF-3 (GCSF) {1} D28/D0 4.55* 2.27

D35/D28 0.93* 0.64

Function Regulates angiogenesis. Initiates differentiation, proliferation, and survival of PMNs and macrophages. Receptor that controls proliferation and differentiation of various hematopoietic cell types. Regulates angiogenesisendothelial sprouting activity. Leukocyte integrin facilitates leukocyte migration across endothelial barrier. Belongs to the chemokine-like factor gene superfamily, a novel family that links the chemokines and the transmembrane 4 superfamilies of signaling molecules. PMN chemoattractant. Initiates cytokine cascade and vascular adhesion molecule expression. Promotion of inammatory inltrate and T-cell chemotaxis. Initiates cytokine cascade and vascular adhesion molecule expression. PMN hematoregulatory cytokine suppresses hematopoietic progenitor cell proliferation. Chemoattractant for lymphocytes and monocytes (immune response and chemotaxis). Mediates leukocyte migration and adhesion. Receptor for colony stimulating factor 3, a cytokine that controls the production, differentiation, and function of granulocytes. Phagocytic granule protein. PMN hematoregulatory cytokine suppresses hematopoietic progenitor cell proliferation.

CSF2RA

CSF-2 receptor alpha {1}

2.13

0.83

ANGPTL1 ITGB2

Angiopoietin-like 1 {1,3,6} Beta integrin 2 {1,2,4,7}

2.07 1.84 1.44 1.32 1.81

0.79 0.79 0.75 0.9 0.73

CMTM2

CKLF-like MARVEL TD {1,3}

IL8 IL1A CCL5 IL1B CXCL3

IL-8 {1,2,3,7} IL-1 alpha {1,3,7} RANTES {1,2} IL-1 beta {1,3,7} GRO gamma {1,2}

1.81 1.79 1.76 1.22 1.72 1.43* 1.71*

0.96 0.7 0.75 0.83 0.77 0.83 0.85*

CCL3L3

Chemokine ligand 3 like 3 {1,2}

1.62

0.66

CX3CR1 CSF3R

Chemokine receptor 1 {1} Colony stimulating factor 3 receptor (granulocyte) {1}

1.55 1.51* 1.31

0.92 0.88 0.86*

S100A8 CXCL2

S100 calcium binding protein A8 {1} GRO beta/MGSA-beta {1}

1.32 1.31*

0.89 0.65

1970

J Periodontol December 2009

Offenbacher, Barros, Paquette, et al.

Table 1. (continued )

Gingival Tissue Response Genes Displaying Increased Expression During Induction of Experimental Gingivitis and Decreased Expression at Resolution
Gene CCL4 ENPP2 KAL1 CXCL12 Name {Gene Ontology Pathway} MIP-1beta {1} Ectonucleotide pyrophosphatase/ phosphodiesterase 2 (autotaxin) {1} Kallmann syndrome 1 sequence {1} Stromal cell derived factor 1 {1} D28/D0 1.28* 1.26 1.23 1.17 1.09* 2.03* 1.63 D35/D28 0.72

Function Attracts CD4 T lymphocytes. Production of physiologically active lysophosphatidic acid. Neural cell adhesion and axonal migration. Trafcking and homeostasis of T-lymphocytes induced by LPS, IL-1, and TNF-alpha. Cell adhesion protein on the cell membrane. Has a role in vascular remodeling, regulates catenin turnover at adherens junctions, and correlates with cellular migration. Receptor for intercellular adhesion molecules, and mediates leukocyteendothelial cell binding. Mediates signal from phospholipase C. Extracellular matrix adhesion receptor for bronectin and cellcell adhesion receptor. Mediator of PMN transendothelial migration. Regulatory subunit of cyclic AMPdependent protein kinase. Activates NF-kappa-B in response to IL-18. Binds to tenascin and integrins, and has a role in cell adhesion and proliferation. Epithelial surface protection. Acute phase protein that induces chemotaxis and adhesion molecule expression. Receptor for prostaglandin E2. Recognizes pathogen associated molecular patterns, ligands include ssRNA viral components.
+

0.89 0.82* 0.8 0.86 0.67 0.53

PCDH9 CDH11

Protocadherin 9 {2} Cadherin 11 {2}

ITGAL

Alpha L integrin {2,4}

1.3 1.25 1.3* 1.27 1.25 1.26 1.27 1.26* 1.17* 1.17

0.83 0.78 0.79 0.79 0.86 0.88* 0.78 0.79 0.8 0.79

PRKCB1 ITGA4

Protein kinase C, beta 1 {2,6,7} Integrins LFA1 and alpha-4 {2}

PAK3 PRKAR2B IL18R1 VCAN

P21 activated kinase 3 {2} Protein kinase, cAMP-dependent, regulatory, type II, beta {2} Interleukin-18 receptor 1 {2} Versican {2}

MUC4 SAA1

Intestinal mucin {3,7} Serum amyloid A1 {3,7}

1.94* 1.78 1.73 1.71 1.46

0.99 0.93 0.92* 0.92 0.76

PTGER3 TLR7

Prostaglandin E receptor 3 {3,7} Toll-like receptor 7 {3}

1971

Immune Response Adaptation to Stress of Biolm Emergence

Volume 80 Number 12

Table 1. (continued )

Gingival Tissue Response Genes Displaying Increased Expression During Induction of Experimental Gingivitis and Decreased Expression at Resolution
Gene IRAK1BP1 Name {Gene Ontology Pathway} Interleukin-1receptor associated kinase 1 binding protein 1 {3} Prostaglandin D2 synthase 21 kDa {3} D28/D0 1.31 1.1* 1.09 1.28

D35/D28 0.88* 0.82 0.88 0.88*

Function Prevents overproduction of proinammatory cytokines by the innate immune system. Neuromodulator and/or trophic factor in the central nervous system. Expressed in peripheral leukocytes. Optimizes immune recognition, facilitating contacts between helper T lymphocytes and antigen presenting cells and between cytolytic effectors and target cells. Procytokine activation, homolog for IL-1beta convertase. Stabilizes T-cell antigen receptor/CD3 complex at the T-cell surface. Unknown. Binds actin, and has a role in cytoskeleton rearrangement. Transduction of T-cell receptor mediated activation. Unknown. T-cell receptor. Transcriptional regulator activates gene transcription during mitogenesis and differentiation. Regulates T-cell receptor signaling.

PTGDS

PTGER4 CD2

Prostaglandin E receptor 4 {3} CD2 molecule {3}

1.25 1.24

0.8 0.85

CASP4 TRAT1 TRBV21-1 CORO1A LCK TRBV19 TRA EGR1

Caspase 4 {3} T-cell receptor associated TM adaptor 1 {4} T-cell receptor variable 21-1 {4} Coronin, actin binding protein {4} Lymphocyte-specic protein tyrosine kinase {4} T-cell receptor variable 19 {4} T-cell receptor variable alpha {4} Early growth response 1 {4}

1.19 1.52 1.5 1.45

0.91 0.65 0.75 0.83

1.39* 1.34* 1.39* 1.38 1.34

0.89* 0.85 0.79 0.73 0.78

PTPN22

Tyrosine phosphatase type 22 {4}

1.34* 1.28 1.28 1.34 1.3 1.27* 1.25

0.76 0.73 0.81 0.83 0.84 0.81

TARP NLRC3 TRAC CD3D

T-cell r gamma constant 2 {4} NLR-CARD domain 3 {4} T-cell receptor alpha constant {4} CD3d molecule, delta (CD3-TCR complex) {4}

T-cell recognition of foreign antigen. Modulates T-cell activation. Unknown. Transitions thymocytes from immature precursors to the nal mature CD4 + or CD8 + singlepositive T cell. Mediates signaling in T cells. A high-mobility group box containing transcription factor implicated in blood glucose homeostasis.

0.85

SIT1 TCF7L2

Signal threshold regulating tm 1 {4} Transcription factor 7-like 2 {4}

1.25 1.2

0.91 0.9

1972

J Periodontol December 2009

Offenbacher, Barros, Paquette, et al.

Table 1. (continued )

Gingival Tissue Response Genes Displaying Increased Expression During Induction of Experimental Gingivitis and Decreased Expression at Resolution
Gene FCGR2A Name {Gene Ontology Pathway} CD32 (IgG Fc fragment) {5} D28/D0 1.87

D35/D28 0.88

Function PMN receptor for IgG activates PTK and PMN degranulation and respiratory burst. PMN receptor for IgG activates PTK and PMN degranulation and respiratory burst. Associated with pre-B-cell acute lymphoblastic leukemias. Regulates apoptosis and cellular redox pathways, and has enhanced antioxidant capacity. B cell-specic scaffold protein and LYN tyrosine kinase substrate that promotes tyrosine phosphorylation of inositol 1,4,5-trisphosphate receptors. Mediates the selective uptake of IgG in mothers milk. Regulates B-cell development. Tissue-specic and differentiation stage-specic DNA-binding protein that participates in the regulation of the pre-B and B lymphocyte-specic MB1 gene. Related to DNA transposons found in fungi and nematodes. I kappa B family of proteins.

1.28 FCGR3B CD16b (IgG Fc fragment) {5} 1.44

0.84 0.74

PBX1 BCL2

Pre-B cell leukemia transcription factor 1 {5} B cell CLL /lymphoma 2 {5}

1.35 1.34 1.33 1.28* 1.26

0.81 0.79 0.76 0.85 0.78

BANK1

B-cell scaffold protein with ankyrin repeats 1 {5}

FCGRT TNFSF13 EBF1

Fc fragment of IgG receptor, transporter, alpha {5} Tumor necrosis factor ligand 13 {5} Early B cell factor 1 {5}

1.23 1.21 1.19

0.9 0.9* 0.81

TIGD7 NFKBIZ

Tigger transposable element derived 7 {5} Nuclear factor of kappa light polypeptide gene enhancer in B cells inhibitor, zeta {5} Immunoglobulin heavy constant alpha 1 {5}

1.19 1.14* 0.84 1.1

0.79 0.79 1.11* 0.77

IGHA1

Predominant Ig class in body secretions, such as saliva, tears, bronchial secretions, and nasal mucosal secretions, to defend against local infection. PMN; degrades superoxide and hydrogen peroxide. An amiloride-sensitive diamine oxidase. Signal transducing molecule. Inhibits growth of certain human tumor cells and stimulates proliferation of human broblasts and other normal and tumor cells.

SOD2 ABP1 LIFR AREG

Superoxide dismutase 2 {6} Amiloride binding protein 1 {6} Leukemia inhibitor factor receptor alpha {6} Amphiregulin (schwannoma-derived growth factor) {6}

1.68 1.47 1.24 1.23 1.23

0.61 0.91 0.85* 0.86 0.92*

1973

Immune Response Adaptation to Stress of Biolm Emergence

Volume 80 Number 12

Table 1. (continued )

Gingival Tissue Response Genes Displaying Increased Expression During Induction of Experimental Gingivitis and Decreased Expression at Resolution
Gene GSR OXR1 P2RX5 CR1 CD69 Name {Gene Ontology Pathway} Glutathione reductase {6} Oxidation resistance 1 {6} Purinergic receptor P2X 5 {7} Complement receptor 1 {7} CD69 molecule {7} D28/D0 1.13

D35/D28 0.92

Function Leads to peroxide reduction through oxidation reduction of l-ascorbate. Protects against oxidative damage. Ligand gated ion channel. Regulates complement activation. Earliest inducible cell-surface glycoprotein acquired during lymphoid activation is involved in lymphocyte proliferation and functions as a signal transmitting receptor in lymphocytes, NK cells, and platelets. Functions in absorption and distribution of iron in the body. Mobilizes intracellular calcium ions to mediate platelet shape change and aggregation. Vascular endothelial cells; also, internalize and degrade OxLDL through the OLR1 receptor. Involved with macrophage-mediated cellular proliferation; mitogenic for broblasts. Mediates the degradation of extracellular matrix and basement membrane proteins for physiologic processes that require tissue remodeling, including wound healing. Catalyzes two reactions in the formation of leukotrienes. Constitutive apolipoprotein of high-density lipoprotein. Has a role in pathogen recognition and innate immunity activation. Stimulates the growth and differentiation of various tissues.

1.08 2.74 1.99 1.5*

0.91* 0.79 0.73 0.64

TF P2RY12

Transferrin {7} Purinergic receptor P2Y, G-protein coupled, 12 {7} Oxidized low-density lipoprotein (lectin-like) receptor 1 {7} Heparin binding EGF {7}

1.45 1.44*

0.84* 0.9*

OLR1

1.41*

0.9

HBEGF

1.39 1.3 1.29

0.79 0.8 0.78

MMP25

MMP-25 {7}

ALOX5 SAA4 TLR10 IGF1

Arachidonate 5-lipoxygenase {7} Serum amyloid A4, constitutive {7} Toll-like receptor 10 {7} Somatomedin C {7}

1.25* 1.23 1.23* 1.22*

0.8 0.93 0.97* 0.78*

D = day; PMN = polymorphonuclear leukocyte; TNF = tumor necrosis factor; AMP = adenosine 39, 59-monophosphate; NF = nuclear factor; Ig = immunoglobulin. Although individual genes may not have demonstrated statistical signicance, gene pathways were signicant at a <0.001. Gene ontology pathways appear as numbers 1 through 7 in brackets {}: {1} = chemotaxis; {2} = transendothelial migration; {3} = hostbacterial interaction; {4} = T-cell activation; {5} = B-cell activation; {6} = oxidative stress; {7} = wound healing/matrix metabolism. Fold changes during induction are shown in column D28/D0, and resolution changes are indicated under D35/D28. Multiple values are presented for genes with more than one transcript. * Statistical signicance of the individual gene-expression changes relative to baseline (P 0.10). Statistical signicance of the individual gene-expression changes relative to baseline (P 0.05). Statistical signicance of the individual gene-expression changes relative to baseline (P 0.01). Statistical signicance of the individual gene-expression changes relative to baseline (P >0.10).

1974

J Periodontol December 2009

Offenbacher, Barros, Paquette, et al.

but is increasingly deemed redundant for pathway analyses, especially when adequate array quality control is performed.10 Because the goal of this effort was not to dene specic biomarkers but to map relevant biologic pathways contributed by hundreds of gene-expression changes, crossvalidation with single-gene platforms such as quantitative reverse transcription PCR was impractical. Condence in the results of these experiments is substantiated by: 1) the number of replicates used in each experiment; 2) the coordinate regulation of multiple genes in each pathway; and 3) previous strong correlations of data on the gene chip used in this study to alternative quantitative PCR platforms9,10 (also, data not shown). Additional points regarding the limitations of the study design and the interpretation of transcriptome data depicted in supplementary Figure 1 appear in the online Journal of Periodontology (supplementary Discussion, section D1). Immune Response Adaptation to the Stress of Biolm Emergence Hostbacterial signaling: Chemotaxis, epithelial and dendritic cell activation, and vasculitis. During induction, many genes that are chemotactic for leukocytes, especially neutrophils and monocytes, were upregulated with signicant increases in IL8, CCL4 (MIP-1b), CCL5 (RANTES), CMTM2 (CKLF-like MARVEL TD) and moderate upregulation of CXCL12 (stromal cell derived factor 1), CXCL2 (Grob), and CXCL3 (Grog) (Table 1). In addition to these upregulated chemokine ligand genes, key receptor genes were induced, including CX3CR1 (chemokine receptor 1). These chemokines attract and activate leukocytes. The observed increase in these specic chemokine genes in combination with colony stimulating factor 3 (CSF3) and its receptors (CSF3R and CSF2ra) suggests a dominant role for selective neutrophil recruitment and activation. This concept was also supported and reected by increases in certain granulocyte/monocytic granule proteins such as SA100A8. The myeloid-related protein, S100A8 (calprotectin), is a product of neutrophils, monocytes, and activated macrophages and is a potent chemoattractant for neutrophils reported to be elevated within the gingival crevicular uid (GCF) in gingivitis.11,12 CX3CL1 (fractalkine) was upregulated 2.4fold and is chemotactic for T cells. The increases in membrane-bound CX3CR1 (fractalkine receptor) also served to tether leukocytes at areas of inammation and within the epithelium.13 The neutrophil chemokine CCL3L3 is not often expressed at sites of inammation. Levels of CCL3L3 expression by neutrophils increase dramatically when neutrophils become infected with intracellular pathogens such as Anaplasma phagocy-

tophilum.14 Increased expression of other cellular markers associated with either invasion of intracellular pathogens or internalization of bacteria was also present, such as of integrin beta-2 (also known as CD18). These ndings suggest that there are cellular signals to recruit neutrophils and other myeloid immunocytes and phagocytic cells and that they are activated by bacteria and/or bacterial products.15 The chemokine IL-8 was previously reported15 to increase in the gingival crevicular uid (GCF) in experimental gingivitis, and the increased expression levels during induction (1.8-fold) seen in the present article are in clear agreement with that report. Two additional chemokines, CCL4 (MIP-1b) and CCL5 (RANTES), previously reported in analyses of inamed periodontitis biopsy samples,16,17 were not specically linked to experimental gingivitis. These inammatory chemokines play a central role in eliciting the T helper 1 (Th1) immune response.18 It has been also shown that MIP-1a, MIP-1b, and RANTES have important functions beyond chemotaxis, as these molecules increase the cytolytic capacity of T and natural killer (NK) cells and costimulate T-cell proliferation.19 This is consistent with classical descriptions by Gemmell et al.17 of T-cell mediated cytotoxicity of resident broblasts serving to trigger apoptosis and local tissue destruction in gingivitis. It is also important to consider the temporal involvement of the type 1 (CCL) chemokines in the context of their functional role, as those chemokines, released in the initial days of infection, participate in cell adhesion and the early attraction for memory T lymphocytes and monocytes, which play a major role in the defense against the bacteria and bacterial products released during biolm overgrowth. Notably, two chemokines that are products of monocytic/dendritic cells (CXCL10 [IP-10] and CXCL11 [I-TAC]) were signicantly suppressed during induction, and these mediators are chemotactic for T cells (Table 2). DEFB4 is an antimicrobial peptide produced by epithelial tissue with chemotactic properties that was suppressed during induction (Table 2). Several endothelial chemotactic and proliferation growth factors that also signal for angiogenesis and endothelial proliferation were elevated, including prokineticin 2 (non-signicant trend, PKT2) and angiopoietin-like 1, and there was evidence of an increased expression of leukocyte binding proteins with enhanced b2-integrin expression. This is consistent with endothelial proliferation, angiogenesis, and endothelial priming for leukocyte adherence. The increases in IL-1 expression represent a wellestablished and consistent nding in both experimental and naturally occurring gingivitis.20-22 Elevations of IL-1a and -1b were reported in tissues and GCF in these conditions.20,21 Immunohistology examining
1975

Immune Response Adaptation to Stress of Biolm Emergence

Volume 80 Number 12

Table 2.

Gingival Tissue Response Genes Displaying Decreased Expression During Induction of Experimental Gingivitis and Increased Expression at Resolution
Gene CXCL10 Name {Gene Ontology Pathway} IP-10 {1,2} D28/D0 0.75* D35/D28 3.93* Function Has a chemotactic effect on activated T cells, TH1-mediated cells, and NK cells. Interferon inducible T-cell alpha chemoattractant. Monocyte chemoattractant. Functions in intracellular adhesion. Cofactor that inhibits the transcriptional activity of androgen receptor in a ligand-dependent manner. Functions in cell-to-cell recognition and adhesion. Degrades proteoglycans and bronectin. Antiadhesive properties balance FN adhesive properties, and has a role in tissue repair. Receptor of leukotriene B4, a potent chemoattractant that is primarily involved in inammation, immune responses, and host defense against infection. Plays a critical role in the guidance of growth cones during neuronal development. Protein expressed by synovial broblasts of rheumatoid arthritis patients. Selenium dependent glutathione peroxidase. Oxidoreductase activity. An NADPH:O(2) oxidoreductase avoprotein; is a component of the thyroid H(2)O(2) generator crucial for hormone synthesis at the apical membrane. Involved in chromatin regulation. Generates superoxide. Reduces aliphatic and aromatic aldehydes.

CXCL11 DEFB4 SDC1 PAK6

I-TAC {1,2} Beta 4 defensin {1} Syndecan 1 {2} p21-activated kinase 6 {2}

0.74* 0.66 0.95* 0.92

5.46 1.8 1.07 1.34

PCDH7 MMP10 TNC

BH protocadherin {2} Stromelysin 2 {2} Tenascin C {2}

0.81 0.69 0.68

1.29 2.38* 1.39*

LTB4R

Leukotriene B4 receptor {3}

0.91

1.25

SEMA4B

Immunoglobulin domain (semaphorin) 4B {5} Sema domain Ig {5}

0.9

1.37

SEMA3C

0.83*

1.3*

GPX2 DHRS7 DUOXA1

Glutathione peroxidase 2 {6} Dehydrogenase/reductase (SDR family) member 7 {6} Dual oxidase maturation factor 1 {6}

0.97 0.94 0.9

1.7 1.22 1.17

AOF1 NOX5 AKR1B10

Amine oxidase domain 1 {6} NADPH oxidase {6} Aldo-keto reductase B10 {6}

0.89 0.88* 0.86*

1.11* 1.32 1.23*

1976

J Periodontol December 2009

Offenbacher, Barros, Paquette, et al.

Table 2. (continued )

Gingival Tissue Response Genes Displaying Decreased Expression During Induction of Experimental Gingivitis and Increased Expression at Resolution
Gene PLA2G3 EGF COL6A3 COL5A3 F2RL1 GPR68 Name {Gene Ontology Pathway} Phospholipase A2 {6} Epidermal growth factor {6} Collagen type 6 alpha 3 {7} Collagen type 5 alpha 3 {7} Coagulation factor II (thrombin) receptor-like 1 {7} G protein-coupled receptor 68 {7} D28/D0 0.79* 0.73 0.92* 0.89* 0.88 0.88* D35/D28 2.68

Function Catalyzes hydrolysis of glycerphospholipids. Regulates cell growth, proliferation, and differentiation. Connective tissue. Connective tissue. Proteinase-activated receptor-2 for trypsin, thrombin-like molecules. Proton-sensing receptor that stimulates inositol phosphate formation. Intrinsic coagulation pathway initiator that binds to negatively charged surfaces including collagen. Extracellular matrix protein. Polyspecic organic cation transporters in the liver, kidney, and intestine, which are critical for the elimination of many endogenous amines as well as a wide array of drugs and environmental toxins. Induction of interferon alpha and beta. Transcription factor for antitrypsin, apolipoprotein CIII, and transthyretin genes. Alpha-1-antichymotrypsin precursor. Increases capillary permeability.

1.89 1.09* 1.15

1.26 1.3

F12

Coagulation factor XII {7}

0.87*

1.38*

SPON2 SLC22A3

Spondin 2, extracellular matrix protein {7} Solute carrier family 22 (extraneuronal monoamine transporter), member 3 {7}

0.87 0.86

1.28 1.33*

IRF7 HNF4A

Interferon regulatory factor 7 {7} Hepatocyte nuclear factor 4, alpha {7} Serpin (serine protease inhibitor) {7} Histamine receptor H1 {7}

0.82* 0.73

1.32* 1.85*

SERPINA3 HRH1

0.73 0.64

1.99 1.57

D = day; Ig = immunoglobulin. Although individual genes may not have demonstrated statistical signicance, gene pathways were signicant at a <0.001. Gene ontology pathways appear as numbers 1 through 7 in brackets {}: {1} = chemotaxis; {2} = transendothelial migration; {3} = hostbacterial interaction; {4} = T-cell activation; {5} = B-cell activation; {6} = oxidative stress; {7} = wound healing/matrix metabolism. Fold changes during induction are shown in column D28/D0, and resolution changes are indicated under D35/D28. * Statistical signicance of the individual gene-expression changes relative to baseline (P 0.05). Statistical signicance of the individual gene-expression changes relative to baseline (P 0.01). Statistical signicance of the individual gene-expression changes relative to baseline (P 0.10). Statistical signicance of the individual gene-expression changes relative to baseline (P >0.10).

cellular localization patterns of IL-1 protein expression within gingivitis tissues showed a dominance of IL-1a expression within the sulcular and pocket epithelial cellular zones, suggesting that the epithelium is a likely source of IL-1a, whereas monocytic and dentritic cells are likely responsible for IL-1b expression. Thus, the observed increase in IL1A gene expression was likely

largely due to epithelial activation, with endothelial, monocytic/dendritic, and broblastic activation being secondary sources of IL1A mRNA synthesis. The upregulation of the IL signaling cascade to include the receptor IL1R1, the adaptor protein MYD88, and the kinase IRAK1BP1, provided strong conrmation of the dominant role of this IL-1 cascade in modulating
1977

Immune Response Adaptation to Stress of Biolm Emergence

Volume 80 Number 12

the SIBO response.23,24 The marked increase in IL1A and IL1B expression may also serve to enhance endothelial expression of leukocyte adhesion molecules to promote leukocyte recruitment. Clearly, gingival tissue and GCF levels of IL-1a and -1b represent wellestablished biochemical markers of gingivitis.20-22 The large increase in the neural chemokine PKT2 with a 4.55-fold induction with the onset of inammation suggests a possible role of neural regulation of chemotaxis via innervations that terminate in vascular beds or, possibly, pocket epithelium. This concept was also supported by the signicant increase in Kallmann syndrome 1 sequence (KAL1) expression, which encodes a chemotactic peptide of neural origins. However, one major new nding in these chemokine data resides in the increased expression seen in the neuropeptide chemokines urotensin (Table 1), PKT2, and KAL1. These increases suggest the role of neural activation and local production of these chemoattractants, possibly at the innervation of vascular beds, to affect neutrophil and endothelial recruitment. Additional neural genes modulated during induction include PMP22 and SNCA. Although, in humans, unilateral denervation (e.g., as a result of trauma) was associated with more severe disease ipsilaterally, the potential role of neural processes has been largely unexplored.25 Increases in the neuropeptide substance P and neurokinin A were reported to be elevated in periodontitis,26 but this may reect macrophage activation. The clear signals for PKT2, as well as the dominance of the neural processing pathways shown in Figure 2, suggest that the role of neurogenic inammation and its potential activity in maintaining homeostasis in response to biolm overgrowth represent new areas for future investigation. Notably, the activation of neural pathways does not appear to represent a dominant transcriptome pathway in periodontitis.16 An extensive list of integrins, focal adhesion molecules, and motility modulating genes were also upregulated during induction, which is consistent with the activation of the transendothelial migration pathway. Protocadherin 9, which is an adherence molecule secreted by neuronal cells, showed the highest fold change during induction. Tenascin C plays an important role in cell adherence to extracellular matrix, is important for cell trafcking, and was specically suppressed (Table 2). Downregulation of matrix metalloproteinase 10 (MMP10) suggests an epithelial response that was impaired, as it is secreted at the free edge of ulcerated epithelium and is essential for reepithelialization.27 Downregulation of the chemokine genes CXCL10 and CXCL11, which produce important mediators of epithelial migration, further suggests that epithelial migration was suppressed during induction. These downregulatory shifts in epithelial signals that regulate reepithelialization were
1978

consistent with the increases in the clinical sign of BI during induction. These signals to the epithelium, many of which are of neural origin, appeared to stun normal epithelial migration to possibly diminish the normal epithelial barrier protective function. Finally, the defensin B peptide, which is chemotactic for neutrophils, was suppressed. Because defensin B is a major antimicrobial peptide produced by keratinocytes and usually upregulated by inammatory stimuli, this nding suggests that the biolm specically mediated the suppression of this peptide by the epithelium. As expected, there are several key genes associated with the hostbacterial interaction pathway that were activated during the induction phase (Tables 1 and 2). The increased expression of IL1A, IL1B, and IL8, as classic innate immune response markers, was revealed. In fact, the entire IL signaling pathway was upregulated to include the receptor IL1R1 (Table 1), the adaptor protein MYD88 (Fig. 3), the regulatory kinase IRAK1BP1 (Table 1), and the IL-1 converting enzyme (caspase 4 [CASP4]; Table 1) that cleaves pro-IL-1 and releases the nal IL1 gene product extracellularly. BCL2 (Table 1) and BAX (Fig. 3) were upregulated and probably serve as antiapoptotic signals preserving mitochondrial structure under periods of stress. Several inammatory biomarkers associated with arachidonate metabolism also increased, including the two prostaglandin E2 receptors (EP4 and EP3). PTGER4 is an epithelial signal that initiates skin immune responses by promoting the migration and activation of Langerhans cells.28 There was also a signicant increase in PGD2 synthase expression, which is a product of bone, neural and mast cells. Interestingly, phospholipase A2, which is responsible for releasing free arachidonic acid from phospholipids, was downregulated during induction. MUC4 is important for epithelial protection and was upregulated during induction and was signicantly reduced at resolution. The major microbial pattern-recognition receptor molecules (toll-like receptors [TLRs]) that were induced included TLR-7 (Table 1), which recognizes single stranded RNA (viral). There were also non-signicant increases in both TLR-4 and CD14, both of which bind lipopolysaccharides (LPSs) (Fig. 3). The epithelial barrier of the sulcus/pocket epithelial lining in cooperation with the interstitial dendritic cells and subbasal Langerhans cells appear to be likely sources for the early increases in expression of TLR-4, TLR-7, and TLR-10 because epithelial cell membranes express TLR-4, and dendritic cells express TLR-3, TLR-7, and TLR-10. Interestingly, several antiviral signaling pathway components were upregulated, including membrane proteins that recognize viral RNA molecules (TLR-7 and TLR-3 [Fig. 3]) and intracellular signaling molecules including VISA and transcriptional factors ITCH and IRF7. Thus,

J Periodontol December 2009

Offenbacher, Barros, Paquette, et al.

the antibacterial components and the antiviral or, more specically, the anti-RNA pathways were also activated. Leukocyte activation (T cell, B cell, and oxidative stress). After recruitment and tethering, T-cell expression of CD2 initiates T-cell contact with antigen-presenting cells and at mucosal surfaces within the lamina propria.29 In addition to the chemokine signals previously mentioned, T-cell recruitment and adhesion were facilitated by increased expression of alpha L integrin and serum amyloid A1 (Table 1). In this experimental gingivitis model, there were multiple T-cell receptors and adaptor proteins that were upregulated during induction and decreased upon resolution. T-cell activation was reected in the increased expression of T alpha, beta, and gamma chains and in the upregulation of SIT 1, which is a negative modulator of T-cell sensitization, as well as the T-cell marker TNFSF13, which is associated with immunoglobulin A isotype switching. Cytotoxic T cells were activated as reected in increased expression of the T-cell granule protein granulysin Y (GNLY). T-cell protein tyrosine phosphatase PTPN22 is an inhibitor of TCR signaling, perhaps increasing the antigenic load or threshold needed for T-cell activation.30 In addition to T-cell activation, certain B-cell pathways were also upregulated during SIBO, although these pathways were not unique to B cells and are shown for comparison purposes. The B-cell proliferation/differentiation factors EBF1 and PBX1 were upregulated. BCL2 is a prosurvival, antiapoptotic signal that was upregulated during induction, as were CD32 and CD16b (FCGR2A and FCGR3B, Table 1). However, these genes are not unique to B cells and were expressed on neutrophils, dendritic cells, and other immune cells. Many genes associated with oxidative stress were induced in SIBO. The neutrophil and macrophage chemoattractant and activator CSF3 also increased (2.27 fold) in expression during induction, as did the endothelial chemoattractant ANGPT. These two mediators are induced by hypoxia and oxidative stress, likely due to the activation of neutrophils within the tissues releasing free radicals and mediators of oxidative stress. This is supported by the marked increase in neutrophilic SOD and GSR expression, mitochondrial OXR1, and TNFSF13. As can be seen in the pathways associated with oxidative stress, many genes appear on other cross-tabulations in Table 1, with the important upregulation of SOD2, the mitochondrial form of superoxide dismutase, and glutathione reductase, both important cytoplasmic markers of oxidative stress. Enzymatic pathways indicating metabolism of products of free radicals and peroxide included glutathione peroxidases GPX1, GPX4, and selenium binding component SECISBP2 and thioredoxin reductase TXNRD1. Increases of glutathione were suggested by increased expression of glutathi-

one synthetases GSS, GCLM, and GCLC. In addition, there were several growth-factor responses characteristic of oxidative stress, including IGF1, HBEGF, ANGPL1, SEMA4B, and AREG that serve as angiogenic signals to recruit endothelial neovascularization. The activation of angiogenesis along with oxidative stress represents a dominant pathway. The upregulation of both ANGP and ANGPTL1 during induction provided stimulus for neovascularization promoting endothelial chemotaxis and proliferation as well as capillary budding. The downregulation of MMP10 expression during the gingivitis induction phase may also play a functional role in angiogenesis. MMP-10, also known as stromelysin-2, induces capillary tube regression and activates the zymogen form of MMP-1. Increased MMP-10 and activated MMP-1 act in concert to degrade the collagen of the basement-membrane matrix leading to capillary tube regression and apoptosis of endothelial cells.31 Thus, the reduced MMP10 expression during induction, with recovery during resolution, was consistent with the activation of the angiogenesis during inammation and subsequent capillary tube regression when the inammatory stimuli ceased. Response to wounding (extracellular matrix metabolism). Wound-healing pathways were activated, including the 5-lipoxygenase (ALOX5) and its activating protein (ALOX5AP) (Table 1), complement receptor 1, and MMP25. MMP-25 is a neutrophil membrane-associated MMP that provides the neutrophil with tissue-invasive properties. Scavenger receptor OLR1 was upregulated, and enhanced expression of the purinergic receptors also occurred. Collagen 5 alpha 3 and collagen 6 alpha 3, both constituents of the basal lamina, were downregulated during induction and rebounded during resolution. In a similar manner, two coagulation factors, XI and XII, showed a signicant down/ up pattern as well as IRF7, which is a transcriptional regulator responsible for interferon induction. The adipokine and MMP inhibitor Serpin A3 was also suppressed during induction and upregulated during repair. MMP25, specically expressed in the leukocyte lineage, was shown to be activated in response to bacterial challenge, rebounding during the resolution phase. MMP25 upregulation is capable of inactivating alpha-1 proteinase inhibitor, a major tissue inhibitor of the proteolytic enzymes released by activated neutrophils. MMP-25, when bound to neutrophils, can degrade vascular type IV collagen and bronectin, which facilitates the transendothelial migration of neutrophils to inammatory sites, and has been reported to be increased within the GCF in periodontal disease.32 CONCLUSIONS The conclusions of this study are presented in Table 3.
1979

Immune Response Adaptation to Stress of Biolm Emergence

Volume 80 Number 12

Table 3.

Novel Biologic Insights Gleaned Through Analysis of Transcriptome Changes in Experimental Gingivitis
Prediction Gingival inammation inuences the transcription of genes of the chemotaxis pathway and transendothelial migration of leukocytes with the role of the immune response as a dominant pathway theme. Verication Transcript levels of 38 genes were signicantly increased in the induction phase, returning to signicantly lower levels in the resolution phase. Key upregulated chemoattractant molecules included IL-8, CCL5 (RANTES), MIP-1b (CCL4), CCL3L3, CX3CR1, SAA1, and S100A8. Downregulated chemoattractant molecules included DEFB4, CXCL11, and CXCL10. Modulation of the immune response represented the dominant pathway, including a strong pathway interaction with the epithelium. Transcriptions of 16 genes associated with hostbacterial interaction were upregulated in the induction phase. Key upregulated genes included IL8, CSF3, IL1A, and IL1B. Key genes associated with angiogenesis were upregulated including PKT2, ANGPTL1, and CAD11. Increases in CD32, SOD2, and CD16beta with downregulation of SERPINE1 are consistent with neutrophil activation, phagocytosis, and intracellular pathogen processing. The hematopoietic growth factors CSF3, CSF3R, CSF2RA, and TNFSF13, which are secreted by Th1 cells, showed signicant activation in response to biolm overgrowth, returning to signicantly lower levels in the resolution phase. Direct evidence of T-cell activation was also supported by increases in TRAT1, TRBV21, TRBV19, and TRA-alpha. Direct evidence of selective regulation was internally consistent with the observation of the upregulation of HBEGF with a juxtapositional downregulation of EGF. Similar specicity was shown with the upregulation of IL8, CCL5, CCL3L3, and CCL4 versus the downregulation of CXC11, CXC10, and DEFB4. The second most dominant cellular activation pathway was associated with neural processes, representing a novel discovery. TLR-7 and TLR-3 increases during induction dominated over TLR-4 (LPS recognition), signaling the potential importance of exogenous or endogenous single- and double-stranded nucleic acid recognition. Subepithelial dendritic cell activation by increases in CCL5, CXCL12, CD14, TLR23, TLR7, and RARRES3 was highly signicant (P = 0.0017). Migration was impeded, and antimicrobial peptide secretion was depressed (DEFB4). Oxidative stress emerged as a dominant enzymatic tissue response with transient uctuations in peroxidases, glutathione synthase, thioredoxin reductase, and superoxide dismutase.

Genes related to hostbacterial interaction and activation of inammation are transcriptionally activated during gingivitis induction. Genes associated with increased vascularization and angiogenesis are increased during induction and decreased upon resolution. Neutrophil inltration is accompanied by activation of phagocytosis and degranulation. In the induction phase of gingivitis, there is a predominance of TH1 host response.

Overall, there is a specic and selective transcriptional regulation that actively seeks to reestablish a metastable inammatory response to reach equilibrium.

Whole-transcriptome analyses would reveal novel cellular and molecular pathways that have not been previously linked to gingivitis responses. Expression of microbial pattern recognition molecules would provide insight into the chemical nature of stimuli and the subsequent cascade. Epithelial integrity is critical as an immune barrier.

Additional inammatory processes can be identied in addition to soluble cytokine signaling.

1980

J Periodontol December 2009

Offenbacher, Barros, Paquette, et al.

ACKNOWLEDGMENTS This study was supported by grants RR00046 and UL1-RR025747 (National Institutes of Health/National Center for Research Resources, Bethesda, Maryland) and a grant from Procter & Gamble. Drs. Begonia Ho, Ryan G. Thomason, and Roger D. Gibb are with the Oral Care Technology Division; Ker-Sang Chen, Andy W. Fulmer, Kenton D. Juhlin, Shuo L. Wang, Tim D. Reichling, and Jay P. Tiesman are with the Global Biotechnology Division; Aaron R. Biesbrock is with the Health Care Research Center; and J. Leslie Winston is with the Oral Care Professional and Scientic Relations Division of Procter & Gamble. The authors acknowledge Beth Jewell-Motz, Senior Scientist, and John Dunavent, Statistical Analyst, of Procter & Gamble for assistance with study design, and David A. Barrow (University of North Carolina School of Dentistry) for editing the nal publication version. Steven Offenbacher has received lecture fees from Procter & Gamble. The authors report no conicts of interest regarding the results presented in this study. REFERENCES
1. Loe H, Theilade E, Jensen SB. Experimental gingivitis in man. J Periodontol 1965;36:177-187. 2. Loe H, Theilade E, Jensen SB, Schiott CR. Experimental gingivitis in man. 3. Inuence of antibiotics on gingival plaque development. J Periodontal Res 1967; 2:282-289. 3. Theilade E, Wright WH, Jensen SB, Loe H. Experimental gingivitis in man. II. A longitudinal clinical and bacteriological investigation. J Periodontal Res 1966; 1:1-13. 4. Burrell RC, Walters JD. Distribution of systemic clarithromycin to gingiva. J Periodontol 2008;79: 1712-1718. 5. Trombelli L, Scapoli C, Tatakis DN, Minenna L. Modulation of clinical expression of plaque-induced gingivitis: Response in aggressive periodontitis subjects. J Clin Periodontol 2006;33:79-85. 6. Scapoli C, Mamolini E, Trombelli L. Role of IL-6, TNFA and LT-A variants in the modulation of the clinical expression of plaque-induced gingivitis. J Clin Periodontol 2007;34:1031-1038. 7. Deinzer R, Weik U, Kolb-Bachofen V, Herforth A. Comparison of experimental gingivitis with persistent gingivitis: Differences in clinical parameters and cytokine concentrations. J Periodontal Res 2007;42: 318-324. 8. Trombelli L, Farina R, Minenna L, Carrieri A, Scapoli C, Tatakis DN. Experimental gingivitis: Reproducibility of plaque accumulation and gingival inammation parameters in selected populations during a repeat trial. J Clin Periodontol 2008;35:955-960. 9. Canales RD, Luo Y, Willey JC, et al. Evaluation of DNA microarray results with quantitative gene expression platforms. Nat Biotechnol 2006;24:1115-1122. 10. MAQC Consortium, Shi L, Reid LH, et al. The MicroArray Quality Control (MAQC) project shows inter- and intraplatform reproducibility of gene

11.

12. 13. 14. 15.

16. 17. 18. 19.

20. 21.

22.

23. 24. 25.

26.

27. 28.

29.

expression measurements. Nat Biotechnol 2006; 24:1151-1161. Kido J, Nakamura T, Kido R, et al. Calprotectin, a leukocyte protein related to inammation, in gingival crevicular uid. J Periodontal Res 1998; 33:434-437. Roth J, Vogl T, Sorg C, Sunderkotter C. Phagocytespecic S100 proteins: A novel group of proinammatory molecules. Trends Immunol 2003;24:155-158. Bazan JF, Bacon KB, Hardiman G, et al. A new class of membrane-bound chemokine with a CX3C motif. Nature 1997;385:640-644. Lee YS, Bak EJ, Kim M, Park W, Seo JT, Yoo YJ. Induction of IL-8 in periodontal ligament cells by H(2)0(2). J Microbiol 2008;46:579-584. Giannopoulou C, Cappuyns I, Mombelli A. Effect of inammation, smoking and stress on gingival crevicular uid cytokine level. J Clin Periodontol 2003; 30:996-1002. Demmer RT, Behle JH, Wolf DL, et al. Transcriptomes in healthy and diseased gingival tissues. J Periodontol 2008;79:2112-2124. Gemmell E, Yamazaki K, Seymour GJ. The role of T cells in periodontal disease: Homeostasis and autoimmunity. Periodontol 2000 2007;43:14-40. Moser B, Loetscher P. Lymphocyte trafc control by chemokines. Nat Immunol 2001;2:123-128. Taub DD, Turcovski-Corrales SM, Key ML, Longo DL, Murphy WJ. Chemokines and T lymphocyte activation: I. Beta chemokines costimulate human T lymphocyte activation in vitro. J Immunol 1996;156:2095-2103. Offenbacher S, Barros SP, Singer RE, Moss K, Williams RC, Beck JD. Periodontal disease at the biolm-gingival interface. J Periodontol 2007;78:1911-1925. Kinane DF, Winstanley FP, Adonogianaki E, Moughal NA. Bioassay of interleukin 1 (IL-1) in human gingival crevicular uid during experimental gingivitis. Arch Oral Biol 1992;37:153-156. Heasman PA, Collins JG, Offenbacher S. Changes in crevicular uid levels of interleukin-1 beta, leukotriene B4, prostaglandin E2, thromboxane B2 and tumour necrosis factor alpha in experimental gingivitis in humans. J Periodontal Res 1993;28:241-247. Serhan CN, Chiang N, Van Dyke TE. Resolving inammation: Dual anti-inammatory and pro-resolution lipid mediators. Nat Rev Immunol 2008;8:349-361. Van Dyke TE, Kornman KS. Inammation and factors that may regulate inammatory response. J Periodontol 2008;79(Suppl. 8):1503-1507. Zavarella MM, Leblebicioglu B, Claman LJ, Tatakis DN. Unilateral severe chronic periodontitis associated with ipsilateral surgical resection of cranial nerves V, VI, VII. J Periodontol 2006;77:142-148. Lundy FT, Mullally BH, Burden DJ, Lamey PJ, Shaw C, Linden GJ. Changes in substance P and neurokinin A in gingival crevicular uid in response to periodontal treatment. J Clin Periodontol 2000;27:526-530. Gill SE, Parks WC. Metalloproteinases and their inhibitors: Regulators of wound healing. Int J Biochem Cell Biol 2008;40:1334-1347. Kabashima K, Sakata D, Nagamachi M, Miyachi Y, Inaba K, Narumiya S. Prostaglandin E2-EP4 signaling initiates skin immune responses by promoting migration and maturation of Langerhans cells. Nat Med 2003;9:744-749. Pawlowski NN, Struck D, Grollich K, et al. CD2 deciency partially prevents small bowel inammation and 1981

Immune Response Adaptation to Stress of Biolm Emergence

Volume 80 Number 12

improves parasite control in murine Toxoplasma gondii infection. World J Gastroenterol 2007;13:4207-4213. 30. Gregersen PK. Gaining insight into PTPN22 and autoimmunity. Nat Genet 2005;37:1300-1302. 31. Saunders WB, Bayless KJ, Davis GE. MMP-1 activation by serine proteases and MMP-10 induces human capillary tubular network collapse and regression in 3D collagen matrices. J Cell Sci 2005;118:23252340. 32. Emingil G, Kuula H, Sorsa T, Atilla G. Gingival crevicular uid matrix metalloproteinase-25 and -26

levels in periodontal disease. J Periodontol 2006;77: 664-671. Correspondence: Dr. Steven Offenbacher, Center for Oral and Systemic Diseases, North Carolina Oral Health Institute, School of Dentistry, University of North Carolina at Chapel Hill, P.O. Box 14290, Durham, NC 27709. E-mail: steve_offenbacher@dentistry.unc.edu. Submitted December 15, 2008; accepted for publication March 25, 2009.

1982