Cell culture models of oral mucosal barriers: A review with a focus on applications, culture conditions and barrier properties

ABSTRACT

Understanding the function of oral mucosal epithelial barriers is essential for a plethora of research fields such as tumor biology, inflammation and infection diseases, microbiomics, pharmacology, drug delivery, dental and biomarker research. The barrier properties are comprised by a physical, a transport and a metabolic barrier, and all these barrier components play pivotal roles in the communication between saliva and blood. The sum of all epithelia of the oral cavity and salivary glands is defined as the blood-saliva barrier. The functionality of the barrier is regulated by its microenvironment and often altered during diseases. A huge array of cell culture models have been developed to mimic specific parts of the blood-saliva barrier, but no ultimate standard in vitro models have been established. This review provides a comprehensive overview about developed in vitro models of oral mucosal barriers, their applications, various cultivation protocols and corresponding barrier properties.

Keywords:

• saliva
• tight junctions
• desmosomes
• ABC-transporters
• SLC-transporter
• craniomaxillofacial surgery
• salivary gland
• tongue
• gingivitis
• periodontitis

Introduction

Ever since evolution allowed for cell type specialization and compartmentalization, the formation of biological barriers gained specific importance. The major interfaces between an organism and the external environment or between different compartments are made up of organized and continuous epithelial cell sheets. Important barrier-forming epithelia are building up the epidermis, the surfaces of the eyes, the surfaces of the digestive, respiratory, reproductive and urinary tracts and the ducts and secretory cells of different glands.¹ Depending on their location, epithelia accomplish several critical functions. Concretely, they play a major role in maintaining homeostasis and act as selective filters that regulate the transcellular movement of solutes between different compartments.² At the same time, they can act as barriers to protect underlying tissue from physical and chemical damage, bacterial infection, dehydration or heat loss.³ The physical barrier aspect of epithelia is defined by the cell membrane and cell junctions that tightly connect neighbor cells to each other and thereby regulate the movement of substances via the paracellular way. In addition, epithelial cell layers represent a transport barrier using influx as well as efflux proteins (transporter proteins or receptors) for small molecules as well as proteins and a metabolic barrier applying enzymes for the modification or conversion of molecules.

Studies on barrier functionality of several epithelial subtypes have shown that cellular barriers are no static constructs, but that they can be altered by several conditions. As an example, it is well known that inflammation and the release of proinflammatory cytokines such as interferon-γ (INFγ) and tumor necrosis factor-α (TNFα) lead to compromised permeability in in vitro models, e.g. for intestinal mucosa^4–6 or salivary glands.^7–9

Next to epithelial cell sheets, another example of an extensively studied biological barrier is the blood-brain barrier (BBB), which main component are brain capillary endothelial cells. In cooperation with other cell types such as astrocytes, pericytes or neurons, the BBB acts as a bidirectional filter controlling the exchange of substances at the interface of the blood and the fluids of the central nervous system (CNS).¹⁰

In contrast to other well characterized biological barriers such as the BBB, the gastrointestinal tract or pulmonary epithelia, less research has been done on cellular barriers which separate blood compartments from saliva. This blood-saliva barrier (BSB) is mainly defined by epithelia of the oral cavity and salivary glands. In addition to epithelial cells, these cell layers are infiltrated by other cell types such as Langerhans cells, melanocytes, Merkel cells or endothelial cells forming blood vessels that might contribute to barrier functionality.

Modelling epithelia of the oral and salivary glands in vitro by cell monolayers and complex tissue engineering approaches has been a major goal of recent studies. A plethora of in vitro models of the BSB has been developed, but no ultimate, standardized models are established neither for models of the oral cavity nor for salivary gland epithelia. Moreover, the epithelia of different regions in the oral cavity (tongue, gingiva, buccal) exhibit significant different barrier properties.¹¹ This is also valid for epithelia from salivary glands (acini, ductal cells). In addition, differences between the three major salivary glands (glandula submandibularis, parotis and sublingualis) as well as the hundreds of minor salivary glands are probable. A minority of the studies using in vitro BSB models are dealing with transport processes of molecules across the BSB. A prerequisite to interpret these reports properly is to understand the barrier properties of these models, which are also understudied. Moreover, cell culture conditions (growth medium, supplements, cell seeding density; submerged versus air-lift set-up, cell type and origin, mono – or multicultures, 2D or 3D) distinctly influence the resulting barrier properties of the used in vitro models. Therefore, there was an essential need for a comprehensive summary considering all the different parameters for in vitro models of the BSB, on the one hand to provide a general overview for readers who are interested in the topic, but also for researchers who apply and would like to compare or improve their in vitro models. The first chapter deals generally with transport routes across epithelial cell layers in relation to the BSB with some examples, the second chapter describes how the barrier functionality is assessed in in vitro models. These two chapters provide the fundamentals in order to understand and classify the data presented in chapters three and four about barrier studies with in vitro models of the epithelia of the oral cavity and the salivary glands. Each of these two chapters begins with a short anatomical overview and general considerations, before the detailed data about the in vitro models are presented and discussed.

Transport Routes across Epithelial Cell Layers

In general, permeation across epithelial barriers is largely achieved by simple passive diffusion (mostly paracellular), carrier-mediated diffusion, active transport or endocytosis.¹² The transport route is mainly determined by lipophilicity, charge and overall molecular geometry of the permeant.¹² For buccal mucosa, it is thought that the majority of tracers and peptide drugs is transported through the paracellular route by passive diffusion.^13,14

Transporter proteins

Active transport of xenobiotics via membrane transporters is an important aspect for the development of alternative drug delivery routes such as transbuccal drug transport, as they can determine pharmacokinetic, safety and efficacy profiles of drugs.¹⁵ During recent years, two major superfamilies of membrane transporters have been extensively studied, namely ATP-binding cassette (ABC) and solute carrier (SLC) transporters. They are key regulators that manage the movement of endogenous metabolites maintaining physiological homeostasis as well as xenobiotics such as drugs and toxins.¹⁶ To date, more than 400 ABC and SLC members have been identified in the human genome with expression patterns throughout the whole body.^15,17 Most notably, expression of both transporter families has been detected in barrier-forming epithelia of major organs such as kidney, liver, intestine, placenta and eye, as well as other body fluid-separating compartments such as the BBB.^18–23 On the mechanistic level, both transporter families act differently. ABC members represent ATP-dependent efflux transporters in all living organisms, whereas the ABC importer function seems to be restricted to prokaryotes.²⁴ In contrast, SLC members are mainly uptake transporters that do not rely on ATP hydrolysis.¹⁷ SLC and ABC transporters have been described to be polyspecific, i.e. to transport several substrates with different affinities and regulations in distinct tissues. Furthermore, membrane transporters show overlapping substrate specificities among the ABC and SLC superfamilies.²¹ Many ABC and SLC members have been described to be clinically relevant with several gene polymorphism linked to diseases (for reviews see e.g.,^{17,21,25,26,27}). However, the study of SLC transporters can be hindered using in vitro models, as many cell lines at higher passages lose transporter expression and activity.¹⁷

Considering BSB compartments, current data suggest that active transport across buccal mucosa seems to be rare.²⁸ In this context, active transport processes have not been extensively studied across the BSBs and therefore the endogenous function of these transporters are not well understood. Therefore, comprehensive investigations of active transport mechanisms across the BSB are still missing and their results might change the current view about their relevance for BSB functionality.

Cell junctions

The paracellular barrier functionality of polarized epithelia is strongly determined by tight junctions (TJs), adherens junctions (AJs) and desmosomes, which are the main constituents of a multiprotein complex referred to as apical junctional complex (AJC) located at the most apical end of the lateral plasma membrane.²⁹ Primarily, TJs act as paracellular gates that restrict diffusion of ions and solutes based on their molecular size and charge.³⁰ In epithelial cell layers, TJs consist of a narrow belt-like structure in the apical region of the lateral plasma membrane. TJs are found in tissues that are involved in polarized secretions, absorption functions, and maintaining barriers between blood and interstitial fluids. The proteins sealing the paracellular gaps belong to the claudin family. Currently, over 20 members of the human claudin family have been identified.³¹ Claudins can bind to other claudins of the neighbor cell either in a homo- or a heteromeric manner. Other proteins important for the structure of the TJs are occludin or tricellulin as well as the group of the junctional adhesion molecules (JAMs). Importantly, the extracellular parts of the TJs are linked via tight-junction associated proteins such as zonula occludens (ZO) 1 to 3 to the cytoskeleton enabling a direct signaling from outside the cells into the cytosol.³² The expression of TJ proteins is mostly investigated either by PCR or Western blot analysis at the mRNA or protein level. In order to understand changes on the functional level, permeability studies with paracellular marker molecules are accomplished and supported by imaging analysis to visualize the localization of TJ proteins. Many authors use TEM for ultrastructural analysis to ensure the distribution of TJ within the cell layers.^33,34 In the case of BSB models, some approaches have been made, but still no comprehensive studies were accomplished to understand the complex tight junction network of BSB models investigating it at the expression, at the localization and the functional level. The stratum corneum is the topmost additional and specialized layer which contributes to the paracellular barrier of the stratified parts of the oral mucosa. Lamellar bodies or so-called membrane coating granules are formed and released by keratinizing epithelial cells during the differentiation processes. These granules contain lipids (e.g. glucosylceramides), hydrolytic enzymes and proteins (e.g. corneodesmosin), which are released after the secretion of the granules and spread over the cell surfaces to support the formation of a thick cell envelope resistant against keratinolytic agents.^35,36 This cell envelope is water-impermeable and a very strong shield preventing paracellular permeability in keratinized stratified squamous epithelia of the oral mucosa. On the contrary, in non-stratified epithelial layers tight junctions might form the major paracellular barrier.

Endocytosis

In order to internalize molecules from the extracellular space, cells also use endocytic processes, where cargo material is engulfed by an invagination of the plasma membrane. In general, two main types of endocytosis are classified depending on the size of ingested material and in the following of formed endocytic vesicles. The ingestion of larger particles such as microorganisms is accomplished by phagocytosis, which is most efficiently performed by specialized phagocytic cells. In contrast, pinocytosis allows for the uptake of fluids and small solutes.³⁷ Endocytosis could be mediated via receptors or adsorption (onto the cellular surface due to electrostatic interactions). For example, multiple mechanisms for the internalization of Candida albicans or HIV in oral epithelial cells were shown. Fungal invasins derived from Candida albicans stimulated epidermal growth factor receptor which caused rearrangement of the epithelial microfilaments and the formation of pseudopods taking up the fungus into the epithelial cells. In case of HIV, several pathways such as clathrin-, caveolin/lipid raft-associated endocytosis and micropinocytosis were involved in the epithelial uptake of the virus^38,39,40 The transferrin receptor (TfR) is another receptor relevant for endocytosis at the oral epithelium. TfR was used as a marker for oral epithelial cells and was reported to mediate the uptake of Porphyromonas gingivalis, a well-known periodontal pathogen. Some drug delivery strategies tried to exploit receptor-mediated endocytosis. For example, nanoparticles decorated with transferrin aimed at binding to the transferrin receptor and to increase the uptake of these nanoparticles via TfR-mediated endocytosis into epithelial cells.^41,42,43 In this context, endocytic transport processes have not been extensively studied across the BSBs and comprehensive investigations of endocytosis at the BSB are still missing.

In terms of barrier functionality, endocytic processes have been described to be involved in TJ regulation by AJC internalization, which results in the reversible opening of epithelial barriers or alteration of cell-cell adhesion properties.⁴⁴ Several pathogenic stimuli such as cytokines, growth factors, oxidative stress and bacterial or viral toxins have been shown to be strong inducers of cytosolic translocation of AJ as well as TJ proteins.⁴⁴ For example, proinflammatory IFN-γ was described to increase paracellular permeability in the T84 intestinal epithelial cell model by inducing endocytosis of the TJ proteins occludin, JAM-A and claudin-1, concretely via myosin II-dependent vacuolarization of the apical plasma membrane.⁴⁵ Studies performed by Harhaj et al. showed that PDGF altered permeability and TJ distribution in MDCK cells, most likely including the endosomal pathway.⁴⁶ Another example for endocytosis-inducing effects is Escherichia coIi’s toxin cytotoxic necrotizing factor-1 (CNF-1) that enhanced paracellular permeability across intestinal epithelial monolayers and arranged TJ protein redistribution essential for epithelial barrier functionality.⁴⁷ However, redistribution of several cadherins through distinct endocytic pathways not only leads to pathological processes, but also is a key mechanism during development and tissue patterning.⁴⁸

Mucus as an additional barrier

It is generally recognized that also extracellular components contribute to the barrier function of many epithelia. As an example, the lipid matrix composition plays an important role in the paracellular diffusion pathway, especially when the compounds such as peptides are hydrophilic or have high molecular weights. In addition, all mucosal surfaces throughout the body are protected by a superficial mucus layer, which forms the outermost physical barrier, primarily against infection as well as chemical, enzymatic and mechanical insults.⁴⁹ The composition of the mucus barrier varies between different tissues. Next to antimicrobial peptides such as defensins, cathelicidins, lysozyme, protegrins, collectins and histatins, the major macromolecular mucus components are mucin glycoproteins that are responsible for the viscous, gel-like properties of the mucus layer.⁵⁰ Depending on their site of location and gel-forming properties, up to 20 known mucins can be divided into three subfamilies, i.e. i) secreted, gel-forming mucins; ii) secreted, non-gel forming mucins; and iii) transmembrane, cell-surface mucins.^49,51 Typically, secreted monomeric mucins cross-link to form extended, homo-oligomeric and viscous networks, whereas their membrane-bound, monomeric counterparts include specific domains that enable their various functions as part of the glycocalyx.⁵¹ In the oral cavity, the mucosa is build up by squamous epithelial cells with underlying salivary glands, which continuously secrete saliva to protect oral and peri-oral tissues as well as to facilitate eating and speech.⁵² Saliva consists of 99% water, but also contains mucin glycoproteins, hormones, vitamins, urea, several ions and antimicrobial peptides, which form the first line of defense for ingested pathogens.^53,54

Measurement of Barrier Functionality in vitro

The function of a biological barrier in vitro could be determined by the paracellular flux of ions or hydrophilic molecules which are not actively transported transcellularly. This is measured by the transepithelial/transendothelial electrical resistance or by the usage of so-called paracellular marker molecules in tracer flux assays. Histological analysis visualizing tight junction structures using transmission electron microscopy (conventional or freeze fracture) or tight junction proteins via immunofluorescence microscopy are often used to support the functional data.

TEER

The measurement of transepithelial/transendothelial electrical resistance (TEER) is a widely used non-invasive, quantitative method to assess barrier integrity of filter-grown cell models.⁵⁵ TEER reflects the permeability of small ions and is commonly declared as measured ohmic resistance multiplied by growth area of filter-grown models (Ωcm²). An increased TEER value reflects a tighter paracellular barrier, since less ions can migrate across the cell layer. Changes of TEER were linked to changed expression and/or localization of tight junction proteins such as claudins. In this regard, it is important to mention that different claudins can be either important for the formation of the paracellular barrier or for the set-up of ions pores. For example, following the claudin-nomenclature of Mineta et al.³¹ claudin-1, -3, -4, -5, -6, -8, -9, -11, -14, -18-2 and -19 exhibited preferentially a sealing function and restricted permeability, whereas claudin-2, -7, -10a, -10b, -15, -16, -17 and -21 have been found to form channels in an anion, cation or water specific manner. In this regard, it has to be mentioned that the knowledge for several claudins (-6, -7, -8, -9, -12, -13, -20, -22 to -27) about their function is limited, their effects on epithelial barriers are partly inconsistent and further research has to be accomplished, especially to understand their role in the complex tight junction networks.^31,32,56,57

Thus, not only an increase of barrier-forming claudins, but also a reduction of pore-forming claudins might contribute to an increased TEER value. TEER data are dependent on several parameters such as temperature, used well format, pore-size and porosity of the membranes, membrane materials and the medium used for the measurement (serum content, viscosity, etc.). Consequently, these parameters should be considered when TEER data from different models are compared. In order to determine the TEER of a cell layer, the electrical resistance values of measured blank inserts without cells have to be substracted from the values obtained with the cell layer, and this difference is multiplied by the surface area. Several different TEER measurement techniques had been developed, their disadvantages and strengths for in vitro barrier models have been recently reviewed by Srinivasan et al.⁵⁸

Tracer flux assays

In order to assess the paracellular tightness and integrity of barrier-forming epithelia in vitro, another possibility is the study of fluorescently or radiolabeled tracer flux across filter-grown cell layers. Frequently used compounds include smaller molecules such as fluorescein, lucifer yellow, mannitol or sucrose, and bigger molecules such as fluorescein isothiocyanate (FITC)-dextrans of various sizes, albumin, IgGs, horseradish peroxidase or inulin. To characterize the transcellular, passive permeability of cell layers lipophilic compounds such as testosterone or diazepam could be used.⁵⁹ These substances should not be substrates of active transport systems and could be used to normalize for cell layers’s variabilities in co- or multidosage drug transport studies. Results of tracer flux assays are often given as the apparent permeability coefficient (P_app) that is defined as the rate of compound accumulation on the opposite of the membrane by time, considering the surface area of the porous membrane as well as the initial drug concentration and the volume of the acceptor chamber.⁶⁰ In contrast to Papp, the calculation of the permeability coefficients (PC or Pe) considers the influence of the membrane support for the transport processes and substracts the permeability across blank membranes without cells in order to obtain the permeability only across the cell layer. This seems to be very relevant for biological barriers with thin basal laminas in the nanometer scale, because the average membrane supports of in vitro transwell models are about 10 micrometers thick and thus represent an additional transport barrier.⁶¹

Barrier Studies of the Oral Mucosa

In addition to the culture conditions and the model set-up, the barrier properties of the in vitro models of the oral cavity depend on the region of origin and the cell type of the epithelial cells. Therefore, it is essential to know the origin and the type of cells in detail for a proper comparison. To account for this, this chapter begins with a brief overview of the anatomy of the oral cavity followed by the description of the in vitro models subdivided in tumor cell derived, primary and immortalized cell lines.

The Oral Cavity – Anatomy

The external anatomical borders of the oral cavity are lips and cheeks. The internal anatomical borders are (i) the anterior pillars of the fauces, (ii) the palate, (iv) the mylohyoid muscle, (iv) the cheeks and (v) the retromandibular region. The oral cavity is covered by three kinds of mucosa: lining, masticatory and specialized mucosa. Lining mucosa is red, consists of non-keratinized stratified squamous epithelium covering the loosely fibrous lamina propria and the submucosa containing fat deposits. This kind of mucosa covers the soft palate, the ventral surface of the tongue, the floor of the mouth, the internal surface of the lips, the cheeks and the alveolar process excluding the masticatory mucosa. Masticatory mucosa is keratinized or parakeratinized and located at the palate, the papilla free dorsal part of the tongue, and the upper part of the alveolar process. In the region of the upper part of the alveolar process and the raphe of the palate, the mucosa is firmly bund to the underlying bone and called gingival mucosa or gingivae, which appears pale pink. The specialized mucosa is the part where the tongue is dorsally covered by numerous papillae.¹¹

In order to analyze barrier properties of different oral mucosa epithelial subtypes, several authors have applied primary or immortalized keratinocytes and tumor-derived cell lines. Depending on the experimental set-up, epithelial monolayers, multilayers as well as 3D organotypic (co-) cultures have been described as important tools for barrier and permeability studies. Therefore, cells were mainly grown on commercially available filter inserts with varying pore sizes and of different materials like polyethylene terephthalate, polyester or polycarbonate.

Tumor-Derived Cell Lines of the Oral Mucosa

Frequently used cancer cell lines of human oral mucosa and their applications are listed in Table 1, whereas Table 2 provides an overview of immortalized cell lines of oral mucosa. Carcinoma cell lines, together with immortalized and primary cells of oral mucosa that have been used for the investigation of barrier properties are further concretized in Table 3.

Table 1. Human tumor-derived cell lines of oral mucosa.

Cell Line	Studies/Applications/Notes	References^1
BUCCAL CARCINOMA
TR146	epithelial barrier studies, model cell line for human buccal epithelium (i.a. drug delivery, trans–epithelial permeability) model cell line for OSCC (i.a. cytotoxicity studies), C. albicans infection studies, 3D organotypic culture (i.a. in commercially available systems of reconstituted oral mucosa, e.g. SkinEthic Laboratories)	^Citation62∏, Citation74#, Citation76, Citation78, Citation83, Citation94–Citation97 barrier/permeability studies see also Table 3 >50 publications (“TR146”)
SqCC/Y1	3D organotypic culture, epigenetic studies, model cell line for malignant buccal carcinoma (i.a. in comparison to normal or immortalized keratinocytes such as SVpgC2a, see Table 3)	^Citation98∏, Citation99–Citation102, Citation103# >40 publications (“SqCC/Y1”)
HO-1-N-1 (Nakata-1; JCRB0831)	wound healing assays/migration, metastasis studies, investigation of retinoic acid (receptors) in OSCC	^Citation104∏, Citation105–Citation107#, Citation108,Citation109 >10 publications (“HO-1-N-1”)
H413 (06092007 SIGMA)	P. gingivalis infection studies, epithelial barrier studies, epigenetic studies, TGFβ studies	^Citation110∏, Citation111–Citation114#, Citation115 barrier/permeability studies see also Table 3 >10 publications (“H413 oral”)
YD-9	1 out of 8 YD oral cancer cell lines (see ^Citation116), chemoresistant SCC model cell line, sensitivity to anticancer drugs, cancer-related over-expression studies	^Citation116∏, Citation117–Citation120# >10 publications (“YD-9” or “YD9)
OC3	areca nut-chewing related buccal carcinoma cell line, cytotoxicity studies/tumor suppression	^Citation121Л, Citation122,Citation123#, Citation124,Citation125 >10 publications (“OC3 oral”)
TW2.6	areca nut-chewing and tobacco smoke-related buccal carcinoma cell line, tumor aggressiveness/migration/invasion studies, epigenetic studies	^Citation126∏, Citation127#, Citation128–Citation130 >10 publications (“TW2.6”)
TONGUE CARCINOMA
CAL27	OSCC model cell line, cancer invasion/migration/metastasis/drug sensitivity studies, …	^Citation131Л; recent publications: ^Citation132#, Citation133,Citation134 >100 publications (“Cal27”)
CAL33	OSCC model cell line, cancer invasion/migration/metastasis/drug sensitivity studies, …	^Citation131Л, recent publications: ^Citation135–Citation137 >40 publications (“Cal33”)
WSU-HN6	cancer cell migration/motility/overexpression studies, drug sensitivity	^Citation138Л, Citation139–Citation141 >10 publications (“WSU-HN6”)
H357 (06092004 SIGMA)	OSCC model cell line for several cancer–related studies: drug sensitivity, cell migration, invasion, aggressiveness, …	^Citation110Л; recent publications: ^Citation142–Citation144 >40 publications (“H357 cancer”)
SCC25 (ATCC® CRL-1628™)	1 out of 6 “SCC” cell lines, weakly invasive (see ^Citation145), OSCC model cell line for several cancer–related studies: drug sensitivity, signaling, migration, invasion, …	^Citation145Л; recent publications: ^Citation133,Citation146,Citation147 ~>300 publications (“SCC25” or “SCC-25”)
D20	moderate dysplastic cell line, 1 out of 16 dysplasias of the oral cavity (see ^Citation148), 3D organotypic culture of oral dysplasia	^Citation148Л, Citation149–Citation152# > 5 publications (“D20 dysplasia”)
DOK “dysplastic oral keratinocyte“	early and mild/moderate dysplastic cell line, tumor progression/migration studies, 3D organotypic culture of oral dysplasia, analysis of stemcell–associated markers	^Citation153Л, Citation150,Citation154–Citation157# >15 publications (“DOK dysplastic”)
OSCC-BD “oral squamous cell carcinoma–B(a)P/DMBA”	established from DOK by B(a)P/DMBA mixture (key carcinogens of tobacco smoke), identification of malignant transformation–related cancer proteins	^Citation158∏ 1 publication (“OSCC-BD”)
HSC-3 (JCRB0623)	OSCC model cell line for several cancer–related studies: drug sensitivity, cell migration, invasion, aggressiveness, cell signaling, stem cell markers …	^Citation159Л, recent publications: ^Citation160–Citation162 ~>300 publications (“HSC-3 oral” or “HSC3 oral”)
HSC-4 (JCRB0624)	OSCC model cell line for several cancer–related studies: drug sensitivity, cell migration, invasion, aggressiveness, cell signaling, stem cell markers …	^Citation163Л, recent publications: ^Citation160,Citation164,Citation165 ~>150 publications (“HSC-4 oral” or “HSC4 oral”)
OSC-19	low-grade invasive cell model of epithelial phenotype, metastatic-lesion derived, metastasis studies, epigenetic studies, cytotoxicity studies, drug sensitivity studies, …	^Citation166Л, Citation167–Citation170# >50 publications (“OSC-19 oral” or “OSC19 oral”)
OSC-20	low-grade invasive cell model of epithelial phenotype, metastatic-lesion derived, metastasis studies, cancer stem cell studies, …	^Citation171Л, Citation168,Citation170#,Citation172 >20 publications (“OSC-20 oral” or “OSC20 oral”)
SCCKN	mouse xenograft cell model, cytotoxicity/drug sensitivity studies, …	^Citation173Л, Citation174–Citation176# >20 publications (“SCCKN”)
PE⁄CA-PJ15	cytotoxicity/drug sensitivity studies, 3D organotypic culture, …	^Citation177Л, Citation156,Citation178–Citation180# >15 publications (“SCCKN”)
YD-8	1 out of 8 YD oral cancer cell lines (see ^Citation116), SCC, cytotoxicity/drug sensitivity studies, …	^Citation116Л, Citation181–Citation183# >5 publications (“YD-8” or “YD8)
YD-10B	1 out of 8 YD oral cancer cell lines (see ^Citation116), SCC, p53-mutant OSCC cell line, cytotoxicity/drug sensitivity studies, …	^Citation116Л, Citation182, Citation183#, Citation184, Citation185 >20 publications (“YD-10B” or “YD10B)
YD-15	1 out of 8 YD oral cancer cell lines (see ^Citation116), mucoepidermoid carcinoma, cytotoxicity/drug sensitivity studies, …	^Citation116Л, Citation186–Citation188# >10 publications (“YD-15” or “YD15”)
SCC-4 (ATCC® CRL-1624)	OSCC model cell line, metastasis/migration studies, cytotoxicity/drug sensitivity studies, …	^Citation189Л, recent publications: ^Citation190, Citation191# >100 publications (“SCC-4 oral” or “SCC4 oral”)
SCC-9 (ATCC® CRL-1629)	OSCC model cell line, metastasis/migration/invasion studies, overexpression studies, cytotoxicity/drug sensitivity studies, …	^Citation145Л, recent publications: ^Citation192,Citation193 >100 publications (“SCC-9 oral” or “SCC9 oral”)
SCC-15 (ATCC® CRL-1623)	OSCC model cell line, metastasis/migration/invasion studies, overexpression studies, cytotoxicity/drug sensitivity studies, …	^Citation189Л, recent publications: ^Citation194,Citation195# >100 publications (“SCC-15 oral” or “SCC15 oral”)
UMSCC2	metastatic SCC cell line, cytotoxicity/drug sensitivity studies, mouse xenograft cell model	^Citation196–Citation198# >10 publications (“UMSCC2”)
SAS	cytotoxicity/drug sensitivity studies, cancer stem cell properties, migration/invasion studies, …	recent publications: ^Citation191Л, Citation199,Citation200 ~>200 publications (“SAS oral carcinoma”)
GINGIVAL CARCINOMA
Ca9-22	OSCC model cell line, cytotoxicity/drug sensitivity studies, epigenetic studies, cancer-related expression studies, …	^Citation201Л, recent publications: ^Citation200,Citation202 >200 publications (“Ca9-22”)
MSCC-1	metastastic-lesion derived, metastasis/migration/invasion studies,	^Citation203Л, Citation204 2 publications (“MSCC-1”)
KOSC-3	invasion studies, mutational analysis	^Citation205Л, Citation206–Citation208# >5 publications (“KOSC-3” or “KOSC3”)
YD-38	1 out of 8 YD oral cancer cell lines (see ^Citation116), cytotoxicity/drug sensitivity studies, studies on tumor-associated macrophages	^Citation116Л, Citation117,Citation209,Citation210# >10 publications (“YD-38” or “YD38”)
OEC-M1	Metastasis/migration/invasion studies, cytotoxicity/drug sensitivity studies, cancer-, metastasis- related (over-)expression studies	^Citation211Л, Citation213–Citation215# >40 publications (“OEC-M1 oral” or “OECM1 oral”)
CARCINOMA OF THE FLOOR OF THE MOUTH
KOSC-2	invasion studies, mutational analysis	^Citation205Л, Citation206,Citation207,Citation215# >5 publications (“KOSC-2” or “KOSC2”)
POE-9n	severe dysplastic oral keratinocytes, not immortal but extended culture lifespan, 3D organotypic culture/tissue engineering,	^Citation216Л, Citation150,Citation156,Citation217# >5 publications (“POE9n” or “POE-9n”)
HO-1-U-1 (Ueda 1; JCRB0828)	metastasis/invasion studies, epithelial barrier studies	^Citation90Л, Citation91,Citation93,Citation165#, Citation218 >15 publications (“HO-1-U-1” or “HO1U1”)
H376 (06092005 SIGMA)	nanoparticle transport studies, cancer-related expression studies	^Citation110Л, Citation219–Citation221# >10 publications (“H376 oral”)
H314 (06092003 SIGMA)	cancer-related expression studies, TGFβ–related studies	^Citation110Л, Citation222–Citation224# >5 publications (“H314 oral”)
OTHERS (or not specified)
KB (ATCC CCL-17) epidermoid carcinoma ! cross - contaminated with HeLa cells! continued misrepresentation of KB cells as oral cancerphenotype (see ^Citation225)	cancer-related (over-)expression studies, cytotoxicity/drug sensitivity studies, migration/metastasis/invasion studies …	^Citation226Л, recent publications: ^Citation120,Citation200,Citation227 >300 publications (“KB epidermoid carcinoma oral”)
H400 (06092006 Sigma) alveolar process squamous cell carcinoma	cytotoxicity/drug sensitivity studies, 3D organotypic culture	^Citation110Л, Citation228–Citation230# >10 publications (“H400 oral”)
MEMO oral mucosal melanoma cell line	established from unusual case of oral mucosal melanoma on hard palate and maxillary alveolar ridge	^Citation231∏ 1 publication, i.e. establishment (“MEMO melanoma”)
1483/SCC1483 carcinoma of retromolar trigone region of the oropharynx some stocks genetically identical to UM-SCC1 (see ^Citation233)	cancer-related expression studies, cytotoxicity/drug sensitivity studies	^Citation233Л, Citation234–Citation236 >10 publications (“1483 retromolar”)
FaDu hypopharyngeal squamous cell carcinoma	migration/invasion studies, cytotoxicity/drug sensitivity studies	^Citation237∏; recent publications: ^Citation238–Citation240 >80 publications (“FaDu oral”)
BHY squamous cell carcinoma of lower alveolus	cytotoxicity/drug sensitivity studies, cancer-related expression studies	^Citation241Л, Citation242–Citation244# >20 publications (“BHY oral”)
HSC-2 (JCRB0622) SCC of oral cavity	OSCC model cell line, cytotoxicity/drug sensitivity studies, cancer-related (over-) expression studies, metastasis/migration/invasion studies, …	^Citation163Л, recent publications: ^Citation165,Citation199,Citation245# >200 publications (“HSC-2 oral” or “HSC2 oral”)
HOC313 OSCC involving the mandibular gingiva and oral floor	high-grade invasive cell model of mesenchymal phenotype, metastatic-lesion derived, cancer progression/metastasis/EMT studies	^Citation246Л, Citation170#,Citation247, Citation248# >20 publications (“HOC313”)
SCC Cell Line Collections
PCI cell lines	21 HNSCC cell lines	^Citation249Л
JHU cell lines (“John Hopkins University”)	5 SCC cell lines of oral cavity: JHU-011 (larynx), −012 (oral cavity), −013 (oral cavity), −019 (base of tongue), −028 (unknown)	^Citation250
73 UM-SCC (“University of Michigan Squamous Cell Carcinoma”) cell lines	73 SCC cell lines of oral cavity	genotyping for all cell lines see ^Citation251

# … latest publication found using respective cell line

∏… publication describing establishment of respective cell line

1) References do not represent a complete list of publications, but list exemplary and/or latest studies. Approximate number of publications should give an impression of how often respective cell lines have been used previously. Search terms are given in parentheses.

abbreviations: HNSCC, head and neck squamous cell carcinoma; OSCC, oral squamous cell carcinoma; SCC, squamous cell carcinoma;

Table 2. Immortalized Cell Lines of Oral Mucosa.

Cell Line	Immortalization	Studies/Applications	References^Citation1
BUCCAL MUCOSA
SVpgC2a human	SV40T	3D organotypic culture, model cell line for immortalized/transformed buccal keratinocytes, head and neck cancer biomarker study	^Citation28∏, Citation99, Citation253, Citation100, Citation282#
NOM9-CT human	Cdk4 + hTERT	effect of curcumin treatment	^Citation283∏#⌂
IMOK mouse buccal and palatal mucosa	spontaneous	first cell line to recapitulate mouse oral epithelium differentiation	^Citation284)∏, Citation285# ⌂
foec (clonal cell lines 1–8) mouse	p53–deficient mouse	foec-6, −8: 3D organotypic culture, foec-2, −3, -5, −6, −7, and −8: preparation of bioengineered tooth germ	^Citation286∏# ⌂
GINGIVA
HMK human	spontaneous	3D organotypic culture, HPV/HSV infection studies	^Citation287∏, Citation288, Citation289, Citation290, Citation291#
IHGK human	HPV16 E6/E7	normal keratinocytes control cell line for cancer studies, 3D organotypic culture, nicotine studies	^Citation292∏ Citation293, Citation294, Citation295, Citation296#
MOE (1a and 1b) human	MOE1a: mutant Cdk4, Cyclin D1, hTERT, dominant–negative p53mutant MOE1b: mutant Cdk4, Cyclin D1, hTERT	immortalization process without viral oncogenes	^Citation297∏#⌂
GE1 mouse	SV40T	epithelial barrier studies, gingival gene expression studies	^Citation257∏, Citation260, Citation258, Citation259, Citation298, Citation299)#
TIGK human	bmi1/hTERT	bacterial infection studies (mainly P.gingivalis)	^Citation300∏, Citation301, Citation302, Citation303, Citation304#
HGEK-16 human	HPV16 E6/E7	3D organotypic culture	^Citation305∏#⌂
NDUSD	SV40T	cytological and cytogenetic characterization	^Citation306∏#⌂
human
OKG4/bmi1/TERT	bmi1/TERT	3D organotypic culture	^Citation216∏, Citation296#
human
MCNR 3B11 =  Gie-No3B11 human	HPV16 E6/E7	epithelial barrier studies, bacterial infection studies (P.gingivalis)	^Citation254∏, Citation255, Citation256#⌂
OBA-9 human	SV40T	bacterial infection studies	^Citation307∏, Citation308, Citation309, Citation310, Citation311#
NOK-SI human	spontaneous	epithelial stem cell functions, wound healing/migration, normal keratinocyte control cell line in cancer and overexpression studies, epigenetics …	^Citation312∏, Citation313, Citation314, Citation315, Citation316#
FLOOR OF THE MOUTH
OKF-6/TERT2 human	hTERT	3D organotypic culture (oral mucosa equivalents, periodontal biofilm model, bone–oral mucosa model), stemcell–associated marker expression, infection studies (bacteria, C.albicans), toxicological studies …	^Citation216∏, Citation281, Citation317, Citation318, Citation319, Citation320, Citation156, Citation321, Citation322, Citation323#
LABIAL VESTIBULE
GMSM-K human	SV40T	bacterial infection studies, model cell line for normal keratinocytes	^Citation324∏, Citation325, Citation326, Citation327#⌂
TONGUE
ROE2 rat	SV40T	toxicological studies of fluoride	^Citation328∏, Citation329#⌂

# … latest publication found using respective cell line

⌂ … complete list of publications using respective cell line is given

∏… publication describing establishment of respective cell line

1 references do not represent a complete list of publications, but list exemplary and/or latest studies; unless marked with ⌂

abbreviations: Cdk4, cyclin – dependent kinase 4; HPV16 E6/E7, human papillomavirus 16 E6/E7 proteins; hTERT, human telomerase reverse transcriptase; SV40T, simian virus 40 T antigen

Table 3. Oral Mucosa Cell Models used for In vitro Barrier Studies.

Cell Line	Cell Culture Medium	Model Set – Up (permeability assays)	Studied Barrier Properties	References
tumor-derived
TR146 human buccal carcinoma	DMEM + 10% FCS + 50 µg/ml gentamicin + 0.2 µg/ml p-hydroxybenzoic acid n-butyl ester DMEM + 10% FCS + 3.7 mg/ml NaHCO3 + 50 µg/ml gentamicin + 0.2 µg/ml p-hydroxybenzoic acid n-butyl DMEM + 10% FCS + 100 IU/ml penicillin + 100 µg/ml streptomycin DMEM + 10% FBS + 200 µM L-glutamine + 100 IU/ml penicillin + 100 µg/ml streptomycin	polyethylene terephthalate filters, Becton Dickinson, 1.6x10^6 pores/cm^2, 0.45 µm pore size, 4.6 cm^2 growth area, 2.4x10^4 cells/cm^2; media change: 3x/week 12-well polyethylene terephthalate filters, Becton Dickinson, 0.4 µm pore size, 0.9 cm^2 growth area, 2.1x10^4 cells/filter polyethylene terephthalate filters, Becton Dickinson, 1.6x10^6 pores/cm^2, 0.40 µm pore size, 4.2 cm^2 growth area, 2.4x10^4 cells/cm^2 polyethylene terephthalate filters, Becton Dickinson, 0.40 µm pore size, 4.2 cm^2 growth area, 2.4x10^4 cells/cm^2 12 well polyethylene terephthalate filters, Becton Dickinson, 0.4 µm pore size, 0.9 cm^2 growth area, 2.2x10^4 cells/cm^2	-TEER^Δ: max. TEER at d30: 55–120 Ωcm^2 (inter-passage variation) -tracer flux: mannitol (hydrophilic), min. Papp at d30: ~ 4x10^−6 cm/s; testosterone (lipophilic), Papp independent of d: ~ 2x10^−5 cm/s -TEM: intermediate filaments, microvilli-like processes, no TJs, membrane coating granules, absence of complete keratinization -TEER^Δ: 339 ± 89 Ωcm^2 (d28-d30) -tracer flux: ^3H mannitol, FITC-dextran (4 kDa, 10 kDa, 20 kDa); enhancement of permeability by chitosan glutamate; e.g. 4 kDa FITC-dextran: Pc = 7.85 ± 2.22 (x10^7) cm/s increased by 20 µg/ml chitosan to Pc = 22.52 ± 6.60 (x10^7) cm/s; 10 kDa FITC-dextran: Pc = 2.95 ± 1.18 (x10^7) cm/s increased by 20 µg/ml chitosan to Pc = 5.32 ± 0.44 (x10^7) cm/s (more values see Table II in ^284) -TEER^Δ: max. TEER at d23, submerged culture: 102 ± 5 Ωcm^2 -tracer flux: ^14C mannitol, min. Papp at d23, submerged culture: 4.08 ± 0.15x10^−6 cm/s -TEER^Δ: 154 ± 13 Ωcm^2 (~ d21), decreased by 10 mM GC treatment: 92 ± 49 Ωcm^2 -tracer flux: FITC-dextran (4–40 kDa): decreasing Papp with increasing Mw 0.65 ± 0.055x10^−8 cm/s to 44 ± 7.5x10^−8 cm/s; -TEER^Δ: 151 ± 38 Ωcm^2 (d27-d29), decreased by chitosan treatment (up to 20 mg/ml) -tracer flux: ^14C mannitol, Papp: 3.79 ± 0.68x10^−6 cm/s, increased by chitosan treatment (up to 20 mg/ml) test of different polymers (chitosan, LM-pectin, HM-pectin, AM-pectin, Eudragit, p(NIPAAM-co-MAA), HM-HEC, HM-EHEC) on cell permeability: -TEER^Δ: significant decrease only by chitosan (d27-d29), but reduced cell viability -tracer flux: ^14C mannitol: significant increase of Papp only by chitosan (d27-d29), but reduced cell viability	^Citation63 ^Citation73 ^Citation64 ^Citation65 ^Citation72 ^Citation74
HO-1-u-1 (Ueda-1) human SCC of the floor of the mouth	DMEM + 10% FCS + 90 IU/penicillin + 90 µg/ml streptomycin Ham’s F-12 + 10% FCS + 2 mM glutamine + 100 IU penicillin/streptomycin + 10 μg gentamicin DMEM-F12 (1:1) + 10% FCS + 50 µg/ml gentamicin + 100 IU/ml penicillin + 100 µg/ml streptomycin	12 well PC filters, Corning Costar, 3.0 µm pore size, 1.131 cm^2 growth area, 2.4x10^4 cells/cm^2 polyethylene terephthalate filters, Becton Dickinson, 0.4 µm pore size, 0.9 cm^2 growth area 12 well PET filters, Appletonwoods, 2.4x10^4 cells/cm^2; media change: every 2–3 days 6 well polyethylene terephthalate inserts, Becton Dickinson, 0.45 µm pore size; media change: 3-4x/week 6 well polyethylene terephthalate inserts, Becton Dickinson, 0.45 µm pore size, 1.5x10^5 cells/well; media change: 3-4x/week	-TEER^Δ: 50.02 ± 2.87 Ωcm^2 (d27-d28) -transport studies of polystyrene nanoparticles using an advanced TR146 model including a mucus layer -TEER^Δ: 36.4 ± 7.6 Ωcm^2 (d28-d30); 50% decrease by ~ 12 mM GC treatment -tracer flux: ^3H mannitol, Papp: 3.3 ± 0.4x10^−6 cm/s, increased to 6.4 ± 0.1x10^−6 cm/s by ~ 12 mM GC treatment -TEER^Δ: ~ 200 Ωcm^2 (d30); no significant changes by lysine (10 µg/ml), histidine (10 µg/ml), glutamic acid (100 µg/ml, 200 µg/ml) and aspartic acid (200 µg/ml) treatment -transport studies: insulin, max. permeation: 4.07 ± 0.82% (4h); significant enhancement of insulin permeation by lysine (10 µg/ml), histidine (10 µg/ml), glutamic acid (100 µg/ml, 200 µg/ml), aspartic acid (200 µg/ml) treatment -TEER^Δ: time-course (d8–d30): continuous increase from 71.4 ± 9.8 Ωcm^2 to a maximum of 326.3 ± 41.3 Ωcm^2 (d17), then slowly decrease to 219.1 ± 28.8 Ωcm^2 (d30) -tracer flux: ^14C mannitol (hydrophilic), Papp: decrease from 3.64 ± 0.28 × 10^−6 cm/s (d15) to 2.04 ± 0.17 × 10^−6 cm/s (d26), then slightly increase to 2.42 ± 0.29 × 10^−6 cm/s (d29); ^3H testosterone (lipophilic), Papp no significant change until d29, ~ 5x10^−6 cm/s; 8 different β-blockers (compounds involving passive diffusion), Papp: range from 2.89 ± 0.17 × 10^−6 cm/s (atenolol) to 6.37 ± 0.37 × 10^−6 cm/s (alprenolol) (d2) -TEM: polarized structure, 2–3 cell layers, short microvilli-like processes, desmosomes, no TJs (d23) -TEER^Δ: no significant differences at various pH (3.5–10), e.g. 175.4 ± 2.5 Ωcm^2 at pH 3.5, 167.5 ± 9.2 Ωcm^2 at pH 10 -tracer flux: ^14C mannitol, ^3H testosterone (neutral markers), Papp: independent of solution pH, e.g. ^14C mannitol with 1.92 ± 0.18x10^−6 cm/s at pH 3.5, 1.63 ± 0.14x10^−6 cm/s at pH 10; β-blockers [aterolol (hydrophilic), metoprolol (lipophilic), propranolol (lipophilic)], several Papp at different osmolarities and GDC concentrations	^Citation87 ^Citation330 ^Citation75 ^Citation91 ^Citation93
H413 human buccal carcinoma	MEM + 5% FCS (low calcium medium, 0.2 mM calcium contributed from FCS)	24 mm Transwell PET filters, 0.4 µm pore size, Corning, 2 × 10^5 cells/cm^2	-tracer flux: monolayers impermeable to tetramethylrhodamine dextran (2 MDa), permeable to Dextran Alexa Fluor 647 (10 kDa); + anti-CD24: reduced permeability -TJ/AJ expression: PCR: + anti-CD24: upregulation: ZO-1, ZO-2, Occl WB: + anti-CD24: upregulation: ZO-1, ZO-2, Occl -tracer flux: Dextran Alexa Fluor 647 (10 kDa); + 0.25 ng/ml TGFβ3: decreased permeability (3h); increased permeability from 4h treatment -TJ/AJ expression: gene expression/protein analysis of 28 TJ components in response to 0.25 ng/ml TGFβ3: PCR: 1) marked up-regulation: Cldn4 2) moderate/low up-regulation: Cldn20, Cldn18 A2.1, Cldn 14, Cldn 15, Cldn 10a, ZO-2;/JAM-B, Cldn11 3) marked down-regulation: Cldn1 4) moderate/low down-regulation: Cldn2, Cldn12, Cldn8, Cldn6, JAM-C 5) no regulation: JAM-A, ZO-1, ZO-3, Occ, Cldn7, Cldn9, Cldn16, Cldn17, Cldn22 6) no expression: Cldn10b, Cldn3, Cldn5, Cldn19, Cldn23, Cldn18A1.1	^Citation331 ^Citation113
immortalized MCNr3B11 (Gie-No3B11) human gingival keratinocytes, HPV–16 E6/E7 immortalization	1. DMEM–Ham’s F12 (4:1) induction of differentiation: 2. DMEM–Ham’s F12 (4:1) + 10% FCS + 1.8 mM calcium + 10 mM HEPES buffer + 200 mM L-glutamine 1. DMEM–Ham’s F12 (4:1) + HEPES induction of differentiation: 2. DMEM–Ham’s F12 (4:1) + 10% FCS + 1.8 mM calcium	24 well Transwell-Col® filter inserts, Corning Costar, 3.25x10^5 cells/insert, 24 well ThinCert^TM filter inserts, Greiner Bio One, 3.25x10^5 cells/insert	WB: 1) significant decrease: Cldn2, Cldn4, Cldn10 2) significant increase: Cldn20, Occ- TEER^Δ: ~ 160 Ωcm^2 (d1); stable until d3, decrease from d4 -TJ/AJ expression: IF: Cldn1^pos (cellular walls), Cldn2^pos, Occ^pos -TEER^Δ: ~ 110 Ωcm^2 (d1) + 15 µM RA: 127 Ωcm^2 (after 1h), 150 Ωcm^2 (after 4h), 189 Ωcm^2 (after 6h) -TJ/AJ expression: WB: Cldn1, Cldn2, Cldn4, Occ, ZO-1 + 15 µM RA: increased expression of Cldn4, Occ; decreased expression of ZO-1; no effect on expression of Cldn1, Cldn2;	^Citation254, Citation255 ^Citation256
GE-1 mouse gingival keratinocytes, temperature -sensitive SV40T immortalization	SFM101 + 1% FCS + 10 ng/ml EGF	-	.-TJ/AJ expression: IF: Cldn1^pos (cellular walls), Cldn4^pos (cellular walls of medial layers), Cldn5^neg, ZO–1^mod + RA: Cldn1^pos (cytoplasm), Cldn4^pos (cellular walls), Cldn5^pos, ZO–1^pos PCR: Cldn1, Cldn4, Cldn5, Occ, ZO-1 + RA: increased expression of Cldn4, Occ, ZO–1 (n.s.); decreased expression of Cldn1, Cldn5; -gap junction expression: IF: Cx26^pos, Cx31.1^pos, Cx32^pos, Cx43^pos PCR: Cx26, Cx30.3, Cx31.1, Cx32, Cx43 + RA treatment: reduced expression of Cx31.1 and Cx32 -TEM: stratified epithelium, desmosomes, hemidesmosomes, anchoring tonofilaments; + RA: thinner cell layer, loss of tonofibrils, less and shorter desmosomes, complete loss of hemidesmosomes -anchoring junctional molecules expression: IF: desmosomes: K13^pos, K4^pos, desmoglein 1^pos + RA: All decreased PCR: desmoglein 1, desmocollin 1, K13, K14, BPAG1, BPAG2, plectin, laminin 5 + RA: decreased expression of desmoglein 1, desmocollin 1, K13, K14, BPAG1; increased expression of plectin; no changes for BPAG2 and laminin 5	^Citation258 ^Citation259 ^Citation260
primary human gingival keratinocytes	DMEM + 15% FCS KGM	24 well cellagen® filter inserts, Costar, 4 × 10^5 cells/insert 12 mm Transwell® PC filter inserts, Corning, 0.4 µm pore size, 5 × 10^4 cells/insert	-TEER^Δ: 374 Ωcm^2 (d5), decrease to 120 Ωcm^2 (d13) -TEM: tight junctions, some desmosomes, no gap junctions -TEER^Δ: ~ 100 Ωcm^2 (~ d3) -TJ/AJ expression: IF: poor detection of Cldn1 (data not shown), no formation of TJs PCR: increased expression of E-cad, β-cat and β-actin by bacterial toxin Cdt (in majority of examined donors)	^Citation265 ^Citation261
human gingival keratinocytes	KGM	0.40 µM polyethylenetetraphthalate track-etched membranes, BD Biosystems	-TEER^Δ: 176–208 Ωcm^2 (independent biopsie samples) -TJ/AJ expression: no expression of ZO proteins (data not shown)	^Citation294
human soft palate keratinocytes	CnT-30 corneal epithelium medium + 1mM CaCl2 after 1 day of submerged culture, 6 days airlift in: 1) CnT-30 corneal epithelium medium 2) CnT-30 corneal epithelium medium + 10 ng/ml EGF 3) CnT-30 corneal epithelium medium + conditioned medium from human foreskin fibroblasts 4) CnT-30 corneal epithelium medium + conditioned medium from human foreskin fibroblasts + 10 ng/ml EGF	mouse collagen IV–coated filter inserts, BD Bioscience, 1 µm pores; media change: every day or every 2–3 days	-TEER^Δ: medium 1) 2379 ± 1167 Ωcm^2 medium 2) 2205 ± 1200 Ωcm^2 medium 3) 2007 ± 1119 Ωcm^2 medium 4) 1610 ± 439 Ωcm^2 (all d13-d17) -TJ/AJ expression: indirect IF: ZO–1^pos, Occ^pos (medium 1)	^Citation267,332
human gingival keratinocytes (EpiGingival^TM 3D model, MatTek)	KGM + 0.05 mM CaCl2 + 200 mM L-glutamine	24 well plate format	-TJ/AJ expression: PCR: mRNA expression of 62 genes (TJs, AJs, gap junctions; see Table 1 of publication ^297)	^Citation266

d … days in culture

Δ … TEER readings from filters without cells were subtracted from values obtained with filters and seeded cells

abbreviations: AJ, adherens junction; AM – pectin, amidated pectin; β-cat, β-catenin; CK, cytokeratin; Cldn, claudin; DMEM, Dulbecco’s Modified Eagle Medium; E-cad, E – cadherin; EGF, epidermal growth factor; FCS, fetal calf serum; GC, sodium glycocholate; GDC, sodium glycodeoxycholate; HM-EHE, hydrophobically modified ethyl hydroxyethyl cellulose; HM-HEC, hydrophobically modified hydroxyethyl cellulose;HM-pectin, highmethoxylated pectin; hTERT, human telomerase reverse transcriptase; IF, immunofluorescence; JAM-A, junctional adhesion molecule A; LM-pectin, low-methoxylated pectin; MUC, mucin; Occ, occludin; p(NIPAAM-co-MAA, poly(N-isopropylacrylamide-co-methacrylic acid); P_app, apparent permeability coefficient; P_c, cellular permability coefficient; PC, polycarbonate; PET, polyester; SCC, squamous cell carcinoma; TEER, transepithelial electrical resistance; TEM, transmission electron microscopy; TJ, tight junction; WB, Western Blot; ZO, zonula occludens.

For the investigation of oral barrier properties, the human cell line TR146, which originates from a neck node metastasis of buccal carcinoma,⁶² has become a standard model. Thus, filter-grown TR146 cells are stated as a model of human buccal epithelium for more than 20 years.⁶³ The suitability of TR146 cells as an in vitro culture model for mechanistic buccal drug delivery, including ionized drugs, and permeability studies has been extensively reported by Jacobsen et al. and Nielsen et al.^63–68 TR146 cells are able to form confluent monolayers as well as multilayered, squamous stratified epithelia of approximately 4–7 cell layers and an average thickness of 40 µm after 3 weeks of culture.⁶⁵ TEER and permeation studies of tracers via TR146 epithelial sheets like FITC-dextran with different molecular weights, mannitol or insulin have been extensively studied and are summarized in Table 3. Comparing submerged cultures and cells additionally grown at the air-liquid interface after 10 days of submerged culture, “tightest” permeability barriers with most distinct stratification were formed at day 23, when continuously cultured submerged.⁶⁴ Compared to other in vitro barrier models of e.g. the gastrointestinal tract or airway epithelium,⁵⁸ TR146 cells form “loose” epithelia with lower TEER values. As summarized in Table 3, TEER is often close to the threshold of 150–200 Ωcm², which are the lowest measurements accepted for drug permeability studies using endothelial cell systems of BBB in vitro models.^69,70 Nonetheless, TR146 cells are a widely accepted model to study oral transmucosal drug delivery, which is considered as an attractive alternative to drug absorption via the gastrointestinal tract.⁷¹ Concretely, TR146 cells are frequently applied to study the effect of permeability enhancers like chitosan,^72,74 bile salts like sodium glycocholate⁶⁵ or amino acids⁷⁵ on drug transport. TR146 cells were recently used for the evaluation of nanosystems^{76, 77} and for cytotoxicity studies of bioadhesive hydrogels for buccal drug delivery.⁷⁸ Approaches that have shown promising effects in enhancing transport of macromolecules in several in vitro and in vivo buccal model systems have been extensively reviewed previously.^79–81 TR146 cells grown at the air-liquid interface have also been used in commercially available 3D models of human oral mucosa. For example, the reconstituted human oral epithelium from Episkin (Lyon, France) consists of multilayered TR146 cells without submucosal compartments. They have been widely used for Candida albicans (C. albicans) infection studies.^82–84

Another important aspect for epithelial barrier formation are cellular contacts like adherens and tight junctions (TJs). For stratified buccal mucosa it is reported that the paracellular permeability barrier is rather based on membrane coating granules (MCGs) than on TJs.^{85, 86} TEM analysis revealed that TR146 cell sheets do not form TJs.⁶³ In line with this, Teubl et al. showed that zona occludens stainings were rare in TR146 cultures.⁸⁷ To our knowledge, the expression of claudin family members has not been studied for the TR146 model.

It also has to be considered that the barrier functionality of oral mucosa in vivo is not only made up of integrated epithelial cell sheets. The oral cavity is continuously moistened with saliva, which contains enzymes and the glycoproteins mucins that form an important “external” acellular barrier to mucosal pathogens or nanoparticles.^49,88 In order to include mucus as a penetration barrier, an advanced buccal in vitro model combining TR146 cells with an adherent mucus layer has been developed.⁸⁷

Due to the carcinogenic nature of TR146, the usage of those cells has shown some limitations. In comparison to normal oral keratinocytes, TR146 based 3D models differ from fully differentiated oral epithelium in terms of histology and differentiation marker expression.⁸⁹ Furthermore, mannitol and testosterone passed the TR146 model ten time faster than human mucosa.⁶⁸ This highlights the fact that a single cell line cannot mimic the complexity of oral tissue in vitro.

Another oral mucosa tumor cell line that has been studied for its barrier properties is HO-1-u-1. In more details, HO-1-u-1 (also referred to as “Ueda 1”) is a human tumor cell line derived from a squamous cell carcinoma of the floor of the mouth.⁹⁰ In comparison to buccal mucosa, which is 500–800 µm thick, sublingual mucosa is 190 µm thin, highly vascularized and therefore an attractive target for drug delivery.⁹¹ A well-known example for sublingual drug administration is the treatment of acute angina attacks with nitroglycerin, which is pharmacologically active within 1–2 minutes.⁹² Another advantage of drug delivery via the oral mucosa in general is the circumvention of enzymatic drug degradation in the gut and the liver. In order to analyze sublingual drug delivery, Wang et al. presented the HO-1-u-1 cell line as an in vitro model for screening of sublingual drug permeation involving passive diffusion.^91,93 Filter-grown HO-1-u-1 formed stratified, polarized and epithelia-like structures, but were devoid of TJs as characterized by TEM analysis. P_app values of β-blockers using the HO-1-u-1 cell model correlated well to those of porcine sublingual mucosa, but measured values were much lower in the cell model, which the authors explained by the loose intercellular structure of HO-1-u-1 layers compared to sublingual tissue.⁹¹ Further evaluation of the HO-1-u-1 model revealed that pH changes are an effective approach to enhance the permeation of β-blockers.⁹³ However, in summary the currently developed in vitro models based on tumor derived cell lines poorly recapitulate barrier properties of the oral mucosa epithelia in vivo.

Primary Cells and Immortalized Cell Lines of Oral Mucosa

In contrast to tumor-derived cell lines, the use of primary and immortalized keratinocytes for barrier studies more closely mimics native oral mucosa in vivo. In order to circumvent senescence of oral primary keratinocytes, they have been immortalized using different approaches, most frequently including a) silencing of tumor suppressor genes like p53 or KRAS; b) infection with viral oncogenes like HPV-16 E6/E7 or SV40T; or c) telomerase reverse transcriptase (TERT) expression.²⁵² Another possibility is a spontaneous immortalization process where often unknown genetic alterations lead to circumvention of senescence. However, all approaches introduce several molecular changes to the cells, including alterations in DNA, mRNA and miRNA profiles, as characterized recently by Dickman et al. for a panel of normal and dysplastic oral cell lines.²⁵² Consequently, immortalized cells can acquire cancer – like features and resemble dysplastic cell lines as described for SV40T – immortalized buccal SVpgC2a cells, which differ from primary keratinocytes in terms of keratin expression, proliferation, apoptosis or responsiveness to serum.²⁵³

Immortalized Cell Lines

A list of immortalized cell lines that have been established from different compartments of the oral mucosa and their primary applications are given in Table 2. Analysis of monocultures revealed that immortalized keratinocytes can show functional characteristics of the epithelial barrier. Gröger et al. established immortalized human gingival keratinocyte cell lines with cytokeratin expression patterns comparable to that of primary gingival keratinocytes. Cells formed multi-layered structures, were able to develop TEER and showed expression of claudin-1, claudin-2 and occludin.^254–256 One of those cell lines, namely Gie-No3B11, was used to study the influence of retinoic acid (RA) on human gingival epithelial barriers, showing increased TEER, increased claudin-4 and occludin expression, while ZO-1 was downregulated by RA treatment.²⁵⁶ The effect of RA on cell junctions of the gingival epithelium was also studied using the immortalized gingival cell line GE1, which was established from transgenic mice harboring a temperature-sensitive SV40T gene.²⁵⁷ GE1 cells form multilayered structures that are connected by desmosomes.²⁵⁷ Studies by Hatakeyama et al. revealed that RA treatment a) altered TJ expression of claudin-1, claudin-4, occludin and ZO-1²⁵⁸; b) decreased the expression of connexin gap junction Cx31.1²⁵⁹ and c) induced downregulation of desmosomes and loss of hemidesmosomes.²⁶⁰

Primary Cells

Formation of oral epithelial barriers by primary cells has also been analyzed. Bacterial infections and their effects on barrier integrity are main foci of several studies, mainly using primary gingival keratinocytes.^{255, 261–263} P. gingivalis has been previously described to degrade epithelial cell-cell junctions in Madin-Darby canine kidney (MDCK) cells.²⁶⁴ The occurrence of TJs in gingival tissue has been reported contrarily. Meyle et al. identified TJs for cultured primary keratinocytes and gingival biopsies by TEM analysis.²⁶⁵ On the other hand, Damek-Poprawa et al. could not detect claudin-1 expression in vitro, but did not test for other TJ protein members.²⁶¹ The same study also revealed that the bacterial toxin Cdt induces changes in the distribution and expression of adherens junction components, highlighting that barrier function could also be a consequence of adherens junction stability. The mRNA expression of 62 genes encoding for TJs, gap junctions and adherens junctions was concretized by Belibasakis et al. investigated using multilayered gingival epithelial cultures, concretely the commercially available EpiGinginval^TM model from MatTek.²⁶⁶ Further studies that analyzed oral mucosa epithelial barriers using primary cells include a multilayered, serum- and feeder-free oral mucosa model of primary cells originating from human soft palate that express TJ proteins ZO-1 and occludin. Furthermore, Ilmarinen and co-workers reported high TEER values for this model, ranging from 1600–2400 Ωcm², depending on culture medium supplements.²⁶⁷ Compared to other studies with TEER values ranging from approximately 25 Ωcm² for human cultivated oral mucosal epithelial sheets²⁶⁸ to 140 Ωcm² for human primary gingival primary cells,²⁶⁹ very high TEER values presented by Ilmarinen et al. are probably a result of strong stratification.

Tissue-Engingeering/Organotypic Cultures

Since cultures of one cell type do not mimic the anatomical complexity of oral mucosa with different infiltrated cell types, blood vessels and submucosal compartments, organotypic co-cultures have been developed. Although a huge variety of set-ups exists, the “basic” components of most 3D in vitro oral mucosal systems are stratified, epithelial cell sheets, which are formed onto scaffold-embedded fibroblasts, mimicking submucosal compartments. In oral and maxillofacial surgery, the main objective is the development of biocompatible oral mucosa equivalents as suitable graft materials. Extensive research is done on models for clinical applications such as burn treatment,²⁷⁰ ocular surface reconstruction^{271, 272} or substitution urethroplasty.^{273, 274} Oral mucosa engineering in terms of advantages and disadvantages of different experimental set-ups, cell sources, scaffolds and applications for oral mucosa engineering has been extensively reviewed previously.^275–278

To our knowledge, complex full – thickness oral mucosa 3D models, such as approaches including Langerhans cells²⁷⁹ or endothelial cells,²⁸⁰ have not been characterized for their barrier functionality in vitro. As described previously, multilayered TR146 cultures are frequently applied for transbuccal drug delivery analyses, although they do not represent a fully differentiated oral epithelium.⁸⁹ As TR146 are derived from buccal carcinoma, 3D models using immortalized oral keratinocytes could be applied in the future to circumvent the limited life-span of primary cells and to enhance the reproducibility of the test model. A widely used immortalized human oral keratinocyte cell line is OKF6/TERT2, which originates from the floor of the mouth. A standardized protocol of OKF6/TERT2 cells cultured on fibroblasts-populated collagen gels, grown at the air-liquid interface, has been published previously.²⁸¹

Barrier Studies of Salivary Glands

In order to be able to assess the data of in vitro models of salivary gland epithelia in a proper manner, it is important to know about the diversity, structure, anatomy and cell types of salivary glands. Moreover, to evaluate and develop models which are functionally similar to the in vivo epithelium it is crucial to become acquainted with the production and composition of saliva. Therefore, this chapter starts with a brief introduction of the anatomy of salivary glands and saliva’s production and composition, before barrier studies using models of the salivary gland epithelia are described in detail.

Salivary Glands – Anatomy

Salivary glands are tubuloacinar exocrine glands with ducts opening into the oral cavity. The glands secrete saliva, which moisten the mucosa, lubricate food for deglutition and works as a solvent for taste and has many more functions. Saliva also contains digestive enzymes such as amylase and antimicrobial agents such as immunoglobulin A, lysozyme and lactoferrin which are secreted by the salivary glands. Saliva seems to have disease preventive properties, as a significant decrease in saliva production can lead to diseases such as periodontal inflammation and dental caries.¹¹

The parotid, submandibular and sublingual glands are the three major paired salivary glands. Numerous other minor salivary glands open into the mouth and are scattered throughout the oral cavity. The parotid gland, the largest of the saliva glands, is situated in front of the external ear and is almost entirely serous. The palpable parotid duct runs superficial of the buccinator muscle and through the cheek to drain into the mouth opposite of the second permanent maxillary molar. The submandibular gland is the size of a walnut and irregular in shape but generally spheroid and is located at the posterior and lower part of the mylohyoid muscle and mostly serous. The sublingual gland is the smallest of the three major paired salivary glands; it is flat and shaped like an almond. The location is cranial of the mylohyoid muscle and beneath the mouth floor mucosa. The sublingual gland is seromucous but most cells are mucous.¹¹

Salivary glands are separable in numerous lobes, which are composed of smaller lobules containing excretory ducts, blood vessels, lymph vessels, nerve fibers and small ganglia. Each duct has branches beginning with dilated secretory “endpieces”, with which the flow of saliva starts. The endpieces can be either tubular or acinar in shape. Different types endpieces can be dominant in the different salivary glands. In the parotid and submandibular gland, the secretory units are mostly serous acini with a few mucous tubules and acini. In the submandibular glands, the dominant cells are the mucous tubules and acini; a minority of serous cells occur as acini or demilunes.¹¹

Myoepithelial cells are associated with secretory endpieces and have an abundance of actin microfilaments which facilitate contraction; these contractions accelerate the outflow of saliva and contribute to the secretory pressure. Lining cells of the ducts are flat near the secretory endpiece and become more cuboidal in excretion direction. Lining cells function as a conduit for saliva. Together with striated ducts, lining cells may modify the salivary content with regard to electrolytes and immunoglobulin A.¹¹ The striated ducts cells build a low columnar epithelium with eponymous basal striations. The striations are a highly infolded region of the basal plasma membrane. Mitochondria being abundant in these infolded regions, are typical for epithelial cells that actively transport electrolytes. Collecting ducts run between the lobes and transport the saliva to the main duct. In the main duct, the lining cells may vary, they can be pseudostatified columnar, stratified cuboidal or columnar, with a distinct basal layer, and a stratified squamous epithelium near the buccal orification.¹¹

Saliva production and composition

In a lifetime a human produces about 25,000 liters of saliva. The unstimulated saliva secretion is about 15 mL/h, of this the parotid glands secrets 25%, the submandibular glands 60% and the sub lingual glands 7–8%. Stimulated by usual food the salivary secretion rises to an estimated 60 mL/h; in this case the parotid glands secrets 50%, the submandibular glands 35% and the sub lingual glands 7–8%. Saliva consists of 994 g/L of water and 6 g/L dry substances, 80% of which are solved and 20% are suspended. The density amounts to 1.01 – 1.02 g/mL and a pH of 5.5 – 6.5 unstimulated and 7.7 stimulated.³³³

Saliva contains approximately 200 mg protein per 100 mL, including α-amylase 25–120 mg/100mL, IgA 20mg/100mL, IgG 1.5 mg/100mL and IgM 0.2mg/100mL. Antibacterial proteins in the saliva are lysozyme, lactoferrin, and sialoperoxidase and glycoproteins, muco glycoprotein 1, 2 and proline-rich glycoproteins. There are also polypeptides such as statherin and sialin and even free amino acids in saliva. The latter is in such a low concentration that extensive bacterial growth is not supported; the same goes for the glucose concentration but glucose concentration rises during food intake. The urea in saliva can be hydrolysed by bacteria leading to a higher pH. Inorganic contents are: potassium, sodium, chloride, phosphorus, bicarbonate, calcium, thiocyanate and fluoride. The composition can be different in different salivary glands and vary with the salivary flow rate.³³⁴

Barrier Studies of Salivary Gland Cell Models

Significant less in vitro models for salivary gland epithelia have been developed in comparison to the models for oral mucosa epithelia. A summary of salivary gland cell models as well as their applications in bioengineering is provided by Nelson et al.³³⁵ In Table 4 (tumor-or immortalized cells) and Table 5 (primary salivary gland epithelial cells) barrier studies with according cellular models of the salivary glands are presented in detail. Following paragraphs summarize different barrier parameters of the salivary gland epithelium in vitro models and their changes after specific treatments.

Table 4. SALIVARY GLAND EPITHELIAL CELL LINES, This table summarizes cell models that have been used for the study of epithelial barrier properties of salivary glands.

Cell Line	Source	Cell Culture Medium	Characterization (ductal/acinar)	Model Set-Up (permeability assays)	Studied Barrier Properties	References
tumor-derived
HSG^# human	neoplastic submandibular gland intercalated duct epithelium	DMEM/F-12 (1:1)^1 + 10% FCS + 100 U/ml penicillin + 100 µg/ml streptomycin + 2.5 µg/ml amphotericin B^2	n.a. -non-coated: IF (d5): CK^neg, VIM^mod(88%), αSMA^neg, αAMY^neg, AQP5^neg, MUC1^neg -MG-coated: IF (d5): CK^pos, VIM^mod(61%), αSMA^neg, αAMY^pos, AQP5^pos, MUC1^mod(30%)	24 mm^3 Transwell PC filter, Corning Costar, 1 × 10^6 cells/well 24 mm Transwell-Clear PET filter, non-coated or MG-coated (f.c. 2 mg/ml, BD Biosciences), 5 × 10^4 cells/cm^2	-TEER^Δ: Ø50 Ωcm^2 (time-course, d0-d5)^4 -TJ/AJ expression: IF: ZO-1^neg, Occ^neg, Cldn1^neg, Cldn2^neg, β-cat^pos -TEER^Δ: -non-coated: 2,1,3 Ωcm^2 (d1, d3, d7) -MG-coated: 401, 417, 332 Ωcm^2 (d1, d3, d7) -TJ/AJ expression: IF (d5): -non-coated: Cldn1^neg, Cldn2^neg, Cldn3^neg, Cldn4^neg, Occ^neg, JAM-A^neg, ZO-1^neg -MG-coated: Cldn1^pos, Cldn2^pos, Cldn3^pos, Cldn4^pos, Occ^pos, JAM-A^pos, ZO-1^pos PCR (d3): -non-coated: Cldn1, Cldn2 -MG-coated: Cldn1, Cldn2 -TEM: -non-coated: small secretory granules, no Golgi apparatus, no rER, no microvilli -MG-coated: secretory-like granules, Golgi saccules, rER, TJs, no basal lamina	^Citation346 ^Citation347
CSG 120/7 mouse	submandibular gland carcinoma	DMEM + 10% FCS + 200 U/ml penicillin + 200 U/ml streptomycin	n.a. n.a.	12 mm Transwell filters, 0.4 µm pore size, Costar 12 mm Transwell PC filters, 0.4 µm pore size, Millipore	-TEER^Δ: e.g. ~ 250 Ωcm^2 (d2), 2350 ± 290 Ωcm^2 (d14) (time-course, d0-d14, measured every 2 days) -TJ/AJ expression: IF (d7): Occ^pos, ZO-1^pos, ZO-2^pos, JAM-A^pos, E-cad^pos, β-cat^pos -tracer flux: 4 kDa FITC-dextran (0.62 ± 0.39 ng/ml, d14) -tracer flux: [^3H]inulin (100%^5)	^Citation357 ^Citation358
immortalized
HSDEC human	-submandibular gland ductal epithelium -hTERT immortalization	BEBM + BEGM Single-Quots (Lonza) + 100 U/ml penicillin + 100 µg/ml streptomycin + 2.5 µg/ml amphotericin B	IF^6: CK7^pos, VIM^neg	12 mm Transwell filters, 0.4 µm pore size, Corning	-TEER^Δ: ~ 1700 Ωcm^2 (d5); + 100 ng/ml TNFγ: ~ 4000 Ωcm^2; + 100 ng/ml TGFβ: ~ 100 Ωcm^2 -TJ/AJ expression: IF: Cldn4^pos, Cldn7^pos, Occ^pos, JAM-A^pos + 100 ng/ml TNFγ: increased Cldn7 expression (WB, PCR), strong expression of Occ (IF); + 100 ng/ml TGFβ: lower expression of Occ at cell borders (IF)	^7
SMIE rat	-submandibular gland -adenovirus 12S E1A gene product immortalization	DMEM + 10% FCS + 100 U/ml penicillin + 100 μg/ml streptomycin	n.a. n.a. n.a.	24 mm Transwell filters, 0.4 μm pore size, Corning Costar, 10^5 cells/cm^2 24 mm Transwell types I and III collagen-coated PC inserts, Corning, 0.4 µm pore size, 3 × 10^6 cells/well 24.5 mm Transwell types I and III collagen-coated PC filter, Costar, 3 × 10^6 cells/well	-TEER^Δ: Ø58 Ωcm^2 (time-course, d4-d7); overexpression of GFP-Cldn4: ~ 40 Ωcm^2 (3d after plating, i.e. 2d after transfection); ~ 80 Ωcm^2 (4d after plating, i.e. 3 days after transfection) -TJ/AJ expression: IF (d4): ZO-1^pos, Occ^pos, Cldn3^pos, Cldn4^neg WB (d n.i.): E-cad, Occ, Cldn3, Cldn2^n.d., Cldn4^n.d., Cldn6-8^n.d., Cldn10^n.d., Cldn12^n.d. PCR (d n.i.): E-cad, Occ, Cldn3^7, Cldn12, Cldn1^n.d., Cldn2^n.d., Cldn4^n.d.- Cldn^11, Cldn14^n.d.–Cldn20^n.d., Cldn22^n.d., Cldn23^n.d. -tracer flux^8: 4 kDa FITC-dextran (~ 18% paracellular flux of input), 70 kDa FITC-dextran (~ 8% paracellular flux of input); overexpression of GFP-Cldn4: decreased flux of 4 kDa and 70 kDa FITC-dextran (4d after plating, i.e. 3 days after transfection) -TEER^Δ: ~ 40 Ωcm^2 (d10), serum-free: < 10 Ωcm^2 (d10) -TJ/AJ expression: IF (d10): ZO-1^pos, Occ^pos WB (d10): Cldn3, Occ, Cldn4^n.d. -tracer flux: 4 kDa FITC-dextran (~ 15% paracellular flux of empty filter control; serum-free: ~ 75% paracellular flux of control)^9 -TEER^Δ: ~ 30 Ωcm^2 (d n.i.) -TJ/AJ expression: IF (d7): ZO-1^pos	^Citation340 ^Citation359 ^Citation360
SMG-C6 rat	-submandibular gland acinar cells -replication defective SV40 immortalization	DMEM/F-12 (1:1) + 2.5% FCS + 5 μg/ml transferrin + 1.1 μmol/l hydrocortisone + 0.1 μmol/l retinoic acid + 2 nmol/l thyronine T3 + 5 μg/ml insulin + 80 ng/ml EGF + 50 μg/ml gentamicin sulfate + 5 mmol/l glutamine + 100 U/ml penicillin + 100 μg/ml streptomycin	WB (d n.i.): CK19^neg n.a.	12 mm Snapwell PC filters, 0.4 µm pore size, Costar, coated with 1.0 µg/cm^2 human collagen type I (Becton Dickinson), 2.5x10^5 cells/cm^2 6.5 mm Transwell filters, Corning, 0.4 μm pore size, 3.8x10^4 cells/cm^2	-TEER^Δ: 956 ± 84 Ωcm^2 (d n.i.) -TEER^Δ: 463.69 ± 13.35 Ωcm^2 (d6); + TNFα: concentration- + time-dependent decrease, e.g. to ~ 200 Ωcm^2 (50 ng/ml, 48h) -TJ/AJ expression: PCR ± WB (d n.i.): Cldn1, Cldn3, Cldn4, ZO-1, ZO-2, β-cat; + TNFα: concentration- + time-dependent disruption of Cldn3 distribution (WB, IF) -tracer flux:4 kDa + 40 kDa FITC-dextran (time course 0-48h, start d6)^10; + TNFα: 4 kDa and 40 kDa FITC-dextran flux significantly elevated (24h, 48h) ® overexpression, depletion or re-expression of Cldn3 altered the response of TNFα induced paracellular permeability and tight junction contents	^Citation345 ^Citation8
			n.a. n.a.	24 well Transwell chambers (0.33 cm^2 growth area), 0.4 µm pore size, Costar 24 well Transwell chambers (0.33 cm^2 growth area), PC filter, Costar, confluent density	-TEERΔ: ~ 600 Ωcm2 (d6/7)^11; no altered TEER by ZO-1, ZO-2, ZO-1/ZO-2 KD; significant drop in TEER by CAP (4-60min), but effect reversed in ZO-1, ZO-2, ZO-1/ZO-2 KD -TJ/AJ expression: PCR ± IF ± WB (d n.i.): ZO-1, ZO-2, ZO-3, Occ, β-cat -tracer flux: 4 kDa FITC-dextran^11; 4 kDa FITC-dextran flux significantly increased in ZO-1, ZO-2, ZO-1/ZO-2 KD, CAP significantly increased 4 kDa FITC-dextran flux, no effect in KD cells -TEER^Δ: 516.2 ± 98.62 Ωcm^2 (d5-d7); + CAP: concentration- + time-dependent decrease + Occ KD or ERK1/2 KD: significant decrease in TEER -TJ/AJ expression: IF (d n.i.): E-cad^pos, β-cat^pos, Cldn3^pos PCR (d n.i.): ZO-1, Occ, Cldn1, Cldn3, Cldn4, Cldn2^n.d., Cldn5^n.d. WB: Occ	^Citation361 ^Citation362
			n.a. n.a. n.a.	24 well Transwell chambers, PC filter (0.33 cm^2 growth area), 0.4 µm pore size, Costar 24 well Transwell chambers, PC filter, 0.33 cm^2 growth area, 0.4 µm pore size, Costar 24 well Transwell chambers, PC filter 0.33 cm^2 growth area, 0.4 µm pore size, Costar	-tracer flux:4 kDa FITC-dextran, trypan blue^12 + CAP: increased permeability for 4 kDa FITC-dextran and trypan blue -TEER^Δ: 632.8 ± 107.1 Ωcm^2 (d7); + Cch: concentration- + time-dependent decrease -TJ/AJ expression: WB (d n.i.): E-cad, Occ, ZO-1, Cldn1, Cldn4 IF (d n.i.): Cldn4^pos, Occ^pos, ZO-1^pos; + Cch: altered distribution of Cldn4 (IF) -tracer flux: 4 kDa + 40 kDa FITC-dextran^12; + Cch: concentration- + time-dependent increase of 4 kDa flux -TEER^Δ: 592.67 + 29.8 Ωcm^2 (d n.i.); + AICAR: decreased TEER values by ~ 46% (60min); Ca2+ free medium decreased TEER values by ~ 60% (60min); -TJ/AJ expression: WB ± IF (d n.i.): Occ, ZO-1, Cldn1, Cldn3, Cldn4 + AICAR: altered distribution of Cldn4 (IF) -tracer flux: 4 kDa FITC-dextran^12; + AICAR: increased permeability of 4 kDa FITC-dextran by 81.59% (60min) -TEER^Δ:635 ± 36.64 Ωcm^2 (d5/6/7); time-dependent decrease of TEER: + fAd (~ 60% decrease, 30min), + gAd (~ 60% decrease, 60min), + AICAR (~ 40% decrease, 60min)	^Citation342 ^Citation341 ^Citation363
Par-C10 rat	-parotid gland acinar cells -replication defective SV40 immortalization	DMEM-F12 (1:1) + 2.5% FCS + 0.1 μM retinoic acid + 80 ng/ml EGF + 2 nM thyronine T3 + 5 mM glutamine + 0.4 μg/ml hydrocortisone + 5 μg/ml insulin + 5 μg/ml transferrin + 5 ng/ml sodium selenite + 50 μg/ml gentamicin + 8.4 ng/ml cholera toxin	n.a. n.a.	n.i. 12 mm Falcon permeable supports, 0.4 µm pore size, Becton Dickinson, 5 × 10^5 cells/well	-TEM: secretory granules, TJ, intermediate junctions, desmosomes, microvilli -TEER^Δ: ~ 3500 Ωcm^2 (d4); + TNFα/INFγ: concentration- + time-dependent decrease, e.g. + 10 ng/ml TNFα: ~ 1200 Ωcm^2 (48h), + 10 ng/ml INFγ: ~ 2000 Ωcm^2 (48h), + 10 ng/ml TNFα/INFγ: ~ 400 Ωcm^2 (48h) -TJ/AJ expression: WB (d n.i.): Cldn1, Cldn3, Cldn4, Cldn10, Occ, ZO-1, JAM-A -tracer flux: 1.9 kDa/20Å microperoxidase MP-11, 44 kDa/50–60 Å horseradish peroxidase type VI, 80 kDa/120Å bovine milk lactoperoxidase^13; + TNFα, + INFγ, + TNFα/INFγ: increased permeability of microperoxidase MP-11 and horseradish peroxidase type VI	^Citation364 ^Citation9
Par-C5  = Pa-4 rat	-parotid gland acinar cells -replication defective SV40 immortalization	DMEM F-12 (1:1) + 2.5% FCS + 5 µg/ml insulin + 5 µg/ml transferrin + 25 ng/ml EGF + 1.1 µM hydrocortisone + 5 mM glutamate + 60 µg/ml kanamycin monosulfate^14	n.a. n.a. n.a.	-Transwell, collagen-coated clear PC filters, 1cm^2 growth area, 0.4 µm pore size, Costar, 10^6 cells/cm^2 Clearwell, Costar Corning	-TEM: secretory granules, TJ, intermediate junctions, desmosomes, microvilli -TEER^Δ: 2000 Ωcm^2 (d7) -TJ/AJ expression: IF (d n.i.): Occ^pos, ZO-1^pos, E-cad^pos WB (d n.i.): Occ, ZO-1, E-cad, Cldn1 -TEER^Δ: ~ 2000 Ωcm^2 (d5) -TJ/AJ expression: IF (d n.i.): β-cat^pos, Occ^pos	^Citation364 ^Citation343 ^Citation344

d … days in culture, n.a. … not analyzed, n.i. … not indicated, n.d. … not detected, f.c. … final concentration

# … listed as cross-contaminated cell lines (HeLa derivates by ATCC, ECACC

Δ … TEER readings from filters without cells were subtracted from values obtained with filters and seeded cells

1) establishment and characterization in different cell culture medium: MEM, 10% FCS, 2 mmol/l L-glutamine ^365,366)

2) different antibiotic/antimycotic substances used: 10 mg/ml gentamicin instead of 2.5 µg/ml amphotericin B³⁴⁷

3) for Transwell inserts, membrane diameters are given in mm

4) collagen I/III coating of filters had no effect on TEER

5) CSG WT monolayers used as control cell line, i.e. inulin flux set to 100%

6) for all experiments, cells were used at the second passage, 5 – 7 days after plating

7) Cldn3 main claudin expressed in SMIE cells

8) for FITC-dextran flux, medium was changed to HEPES-buffered saline (10 mM HEPES/NaOH pH 7.4, 1 mM sodium pyruvate, 10 mM glucose, 3 mM CaCl2, 145 mM NaCl)

9) for FITC-dextran flux, medium was changed to DMEM containing 10 mM HEPES buffer (pH 7.4)

10) WT SMG-C6 monolayers given as “relative paracellular flux” = 1

11) TEER and paracellular flux given for scrambled siRNA SMG-C6 control cells, i.e. relative paracellular flux of FITC-dextran set to 1

12) apparent permeability coefficients given as “% of initial value”, WT SMG-C6 set to 100%

13) peroxidase activity in cell supernatant given as “absorbance values at 405nm”

14) Li et al.³⁴³ used 600 µg/ml G418 instead of kanamycin monosulfate; Quissel et al.³⁶⁴ additionally used 8.4 ng/ml cholera toxin

abbreviations: αAMY, α amylase; αSMA, α smooth muscle actin; AICAR, 5-aminoimidazole-4-carboxamideribonucleoside (AMPK activator); AJ, adherens junction; AMPK, AMP-activated protein kinase; AQP, aquaporin; β-cat, β-catenin; BEBM, bronchial epithelial basal medium; CAP, capsaicin; Cch, carbachol; CK; cytokeratin (panel); Cldn, claudin; DMEM, Dulbecco’s Modified Eagle Medium; E-cad, E – cadherin; EGF, epidermal growth factor; fAd, full length adiponectin; FCS, fetal calf serum; gAd, globular adiponectin; hTERT, human telomerase reverse transcriptase; IF, immunofluorescence; JAM-A, junctional adhesion molecule A; MEM, minimum essential media; MG, Matrigel; MUC, mucin; Occ, occludin; PC, polycarbonate; PET, polyester; TEM, transmission electron microscopy; TEER, transepithelial electrical resistance; TJ, tight junction; VIM, vimentin; WB, Western Blot; ZO, zonula occludens;

Table 5. Salivary gland primary epithelial cells: This table summarizes primary cell culture techniques that have been used for the study of epithelial barrier properties of salivary glands.

Primary Cells	Cell Culture Medium	Max. Passage used	Characterization (ductal/acinar)	Model Set-Up (permeability assays)	Studied Barrier Properties	References
phmSG human minor salivary gland	KGM-high (0.8 nM Ca^2+) + BPEs + hEGF + insulin + hydrocortisone + gentamicin + epinephrine + transferrin^1	8–10 (in KGM-low, 0.05 nM Ca^2+)	-PCR (d3)^2: ZO-1^high, CLDN1^high, AQP5^rel.high, SLC12A2 ^rel.high, CST3^rel.high, ORAI1^rel.high, αAMY^modest, ANO1^modest KRT19^low, KLK1^low -IF (d3): SLC12A2^pos, CSTe^pos, AQP5^pos -agonist-induced αAMY secretion and activity	Transwell-COL, collagen-coated PET filter, 0.4 µm pore size, Corning, 20 × 10^4 cells/well	-TER^Δ3: ~ 90–180 Ωcm^2 (time-course, d3-d10) -TJ/AJ expression: IF (d3): ZO-1^pos PCR (d3):E-cad, ZO-1, Cldn1	^Citation367
huSMG human submandibular gland	serum-free Hepato-STIM + 1% glutamine + 500 U/ml penicillin + 500 µg/ml streptomycin + 12.5 µg/ml fungizone	n.i. 3	-IF (d n.i.): E-cad^pos, Vim^neg -PCR (d n.i.): αAMY, AQP5, NKCC1, ENaC, Cldn1 - αAMY secretion: strong secretion on Transwell vs plastic (5022–1682 U/l vs 742–214 U/l)	24 mm Transwell-Clear, PET filter, Corning Transwell-Clear, PET filters, FCS-coated, 0.4 µm pore size, 1.12 cm^2 growth area, Corning Costar	-TER^Δ: ~ 250–300 Ωcm^2 (d n.i.) -TJ/AJ expression: IF (d n.i.): ZO-1^pos, Cldn1^pos, E-cad^pos -TEM: tight junctions, microvilli, no secretory granules -TER^Δ: 622 ± 117 Ωcm^2 (d7-d10)^4 -TJ/AJ expression: PCR (d n.i.): Cldn1	^Citation368 ^Citation369 5
huSG human submandibular or parotid gland	serum-free Hepato-STIM + 1% glutamine + 500 U/ml penicillin + 500 µg/ml streptomycin + 12.5 µg/ml fungizone	3	-IF (d n.i.): non-coated: CK^pos(85%), VIM^mod(40%), αSMA^neg, αAMY^neg, AQP5^neg, MUC1^neg MG-coated: CK^pos, VIM^pos, αSMA^neg, αAMY^pos, AQP5^pos, MUC1^mod(50%) -enhanced αAMY secretion in MG-coated set-up	24 mm Transwell-Clear, non-coated or MG-coated (f.c. 2 mg/mL BD Biosciences) PET filters, Corning, 5 × 10^4 cells/cm^2	-TER^Δ: non-coated: 381, 409, 415 Ωcm^2 (d3, d5, d7); MG-coated: 409, 407, 415 Ωcm^2 (d3, d5, d7) -TJ/AJ expression: IF (d n.i.): non-coated± MG-coated: Cldn1^pos, Cldn2^pos, Cldn3^pos, Cldn4^pos, Occ^pos, JAM-A^pos, ZO-1^pos -TEM: non-coated: TJs, no Golgi apparatus, no rER, no secretory granules, no microvilli MG-coated: TJs, many Golgi saccules, rER, secretory granules	^Citation370
PTHSG^6 human submandibular gland	Hepato-STIM + 10% FCS + 1% glutamine + 500 U/ml penicillin + 500 µg/ml streptomycin + 12.5 µg/ml fungizone	4	IF (p0): non-coated or BME-coated: αAMY^pos and αAMY^neg clusters, non-coated: some Vim^pos clusters, BME-coated: Vim^neg	-	-TJ/AJ expression: PCR: describes differential expression of αAMY, Cldn1, Cldn3, KLK and Vim in 13 patients	^Citation371
		3	-PCR (d n.i.): αAMY, AQP5, NKCC1, ENaC, Cldn1 - αAMY secretion: strong secretion on Transwell vs plastic (3043–882 U/l vs 568–193 U/l)	Transwell-Clear, PET filters, FCS-coated, 0.4 µm pore size, 1.12 cm^2 growth area, Corning Costar	-TER^Δ: 595 ± 93 Ωcm^2 (d7-d10)^4 -TJ/AJ expression: PCR (d n.i.): Cldn1	^Citation369
RPG rhesus monkey parotid gland	serum-free Hepato-STIM + 1% glutamine + 500 U/ml penicillin + 500 µg/ml streptomycin + 12.5 µg/ml fungizone	4	IF (d n.i.): E-cad^pos, Na/K ATPase^pos	24 mm Transwell-Clear, PET filter, Corning	-TER^Δ: ~ 135 Ωcm^2 (d3) -TJ/AJ expression: IF (d n.i.): E-cad^pos, ZO-1^pos, Cldn1^pos -restricted paracellular fluid movement (2.0 µl/cm^2 vs. 6.5 µl/cm^2 in HSG)	^Citation372
ASGE BALB/c mouse submandibular gland	DMEM–F12 + 2 mM glutamine + 10 ng/ml EGF + 5 µg/ml insulin + 5 µg/ml transferrin + 100 U/ml penicillin + 100 U/ml streptomycin + 0.25 µg/ml amphotericin + 2 nM triiodothyronine T3 + 0.4 µg/ml hydrocortisone + 0.18 mM adenine + 100 nM cholera toxin	5	IF (d n.i.): CK^pos, VIM^neg (ASGEs on feeder layer of irradiated, VIM^pos NIH fibroblasts)	24 mm Transwell, PC filters, Corning Costar, 1 × 10^6 cells/well	-TER^Δ: ~ 150 Ωcm^2 (d5) -TEM: desmosomes, TJs, microvilli, secretory granules (ASGEs on feeder layer of irradiated NIH 3T3 fibroblasts)	^Citation373

d … days in culture, n.a. … not analyzed, n.i. … not indicated, n.d. … not detected, f.c. … final concentration

Δ … TEER readings from filters without cells were subtracted from values obtained with filters and seeded cells

1 Jang et al.³⁶⁷ tested 6 differently supplemented growth media (no concrete supplement concentrations given). In this table, growth conditions that achieve the maintenance of phmSG in an acinar-like phenotype are given.

2) Expression levels of respective genes (high, relatively high, modest, low; in comparison to expression levels in other growth media) were adopted from Jang et al..³⁶⁷

3) Cells grown in KGM-H (0.8 nM Ca²⁺) showed fastest and highest rates of TEER development; after 7 days, TEER showed no significant difference between cells grown in KGM-low and KGM-high.

4) In MEM medium, TEER values were below 200 Ωcm².

5) Preparation of tissue biopsies and cell isolation as previously described by Tran et al. (2005), with some modifications.

6) The isolation and culture method for PTHSGs described by Szlavik et al.³⁷¹ and Hegyesi et al.³⁶⁹ is based on the protocol of Tran et al.,³⁶⁸ with some modifications.

abbreviations: αAMY, alpha amylase; AQP, aquaporin; BPEs, bovine pituitary extracts; β-cat, β-catenin; CK; cytokeratin (panel); Cldn, claudin; d, day; DMEM, Dulbecco’s Modified Eagle Medium; E-cad, E – cadherin; EGF, epidermal growth factor; FCS, fetal calf serum; FH, fibrin hydrogel; GFR, growth factor reduced; hTERT, human telomerase reverse transcriptase; JAM-A, junctional adhesion molecule A; KGM, keratinocyte growth medium; MG, Matrigel; MUC, mucin; n.a., not analyzed; Occ, occludin; PC, polycarbonate; PET, polyester; P_app, apparent permeability coefficient; rER, rough endoplasmic reticulum; VIM, vimentin; ZO, zonula occludens;

Tight junctions (TJs)

Regarding epithelial barrier studies of salivary glands, TJs represent the most extensively studied structure. TJs build the primary barrier against paracelluar fluid and ion movement and are thus responsible for the maintenance of cell polarity and selective transepithelial ion gradients required for saliva secretion.³³⁶ Several authors reported previously about TJ organization and expression in salivary epithelium cell models and salivary gland tissue.^{336, 337, 338, 339} A major topic of several studies is the relationship between the expression of different claudins and the corresponding barrier functionality, especially in immortalized rat salivary gland epithelial cell lines such as SMIE, SMG-C6, Par-C10 or Par-C5 (Table 4). In more detail, SMIE cells were reported to have a limited claudin expression profile, which was in line with their low barrier function, indicated by low TEER and high dextran flux. However, overexpression of claudin-4 in SMIE cells approximately doubled TEER values and decreased 70 kDa dextran flux.³⁴⁰ A barrier – regulating role of claudin-4 was also described for other salivary gland epithelial cell lines. Xiang et al. reported that knockdown of claudin-4 in SMG-C6 cells reversed an AMPK-induced decrease of TEER values and increase of 4 kDa FITC-dextran flux.³⁴¹ Similarly, claudin-4 is essential for modulation of paracellular permeability by muscarinic acetylcholine receptors in SMG-C6 cells.³⁴²

TEER

TEER values strongly vary between different salivary gland epithelial cell models. Highest TEER have been reported for immortalized salivary gland cell lines such as human HSDEC⁷ or rat Par-C10⁹ and Par-C5,^343,344 the latter also referred to as Pa-4. SMG-C6, a rat submandibular gland acinar cell line, was described by several authors to develop TEER values ranging from ~464 Ωcm^{2 8} to a maximum of ~956 Ωcm^{2 345} when seeded on collagen – coated inserts (see Table 4). In contrast, the human neoplastic cell line HSG exhibits no or low resistances (e.g. ~50 Ωcm^{2 346} ~1-3 Ωcm^{2 347}), but can be increased to 332–417 Ωcm² by appropriate insert coating such as Matrigel.³⁴⁷ However, results from studies using HSG cells should be critically questioned as it has been identified as a HeLa cross – contaminated cell line.³⁴⁸ Another tumor – derived cell line that has been studied for TEER development is CSG 120/7, originating from murine submandibular gland carcinoma, showing an increase from day 2 with ~250 Ωcm² to day 14 with an average of 2350 Ωcm².

Active Transport – Membrane transporters

Physiological expression of ATP-binding cassette (ABC) transporters in human salivary duct tissue has been described previously.³⁴⁹ Nishimura et al. analyzed mRNA expression profiles of 46 human ABC transporters and 108 human solute carrier (SLC) transports in several tissues, including salivary gland.¹⁹ These studies reported expression at the protein level of major ABC transporters such as ABCB1 (P-glycoprotein) or ABCC1 and ABCC2 (MRP1, MRP2) in salivary gland ductus epithelial cells of healthy tissue, and a huge number of mRNA transcripts of almost all investigated ABC and SLC transporters. Although these data suggest a physiological role of these transporters, functionality studies of these transporters are missing in salivary glands.

Endocytic Processes

As mentioned before, either receptor – or adsorption mediated endocytosis contributes to the uptake and transport of substances across biological barriers. Several publications reported endocytic activity of salivary gland epithelial cells functioning as reuptake mechanism from primary saliva or as uptake/transcytosis from blood into saliva. For example, ferritin or BSA injected in the main excretory duct of the rat submandibular gland was uptaken by light (types I and II) and dark cells or the intercalated or striated duct,^350,351 clathrin-mediated uptake of muscarin-3 receptor and clathrin independent uptake of flotilins was proven in HSG cells,³⁵² the endocytosis of E-cadherin in salivary epithelial cells controlled by Pak1 in Drosophila was essential for the formation of multitubes in the development of salivary glands,³⁵³ the uptake of autoantibodies against Ro and La –relevant in the pathology of the Sjören’s syndrome – via Fc-gamma receptors in human salivary gland cell line A-253 induced apoptosis³⁵⁴ and the presence of LRPs such as LRP1B in salivary glands suggested also LRP endocytic/transcytotic activity at the salivary gland epithelium.³⁵⁵ However, although expression and functional data in vitro as well as in vivo revealed evidence for endocytic processes at the salivary gland epithelia, only a limited number of studies and no comprehensive characterization exist.

Conclusions

The aim of this review was to provide an overview of BSB in vitro models and their barrier properties. In summary, many models of the oral mucosa and a smaller number of salivary gland epithelia models exist. They have been used for several different applications, cultivated under diverse conditions, but were hardly comprehensively characterized for their barrier properties. No ultimate in vitro model exists for any part of the BSB. Since the knowledge of the barrier properties and the corresponding cultivation conditions is the essential basis for each in vitro study with a model of a biological barrier, we hope that this review will help researchers to gain a valuable overview and to choose the most suitable model for their purpose. The comprehensive selection of information and data within this review also provides a starting point for future studies and the development of improved and thoroughly characterized models. In addition to the needed qualification and correlation to in vivo data of the in vitro models for the paracellular, the transport and the metabolic barrier at the expression, localization and most importantly the functionality level, future model development and validation will include the microenvironment and its influence on barrier properties. For example, Burghartz et al. showed recently in a 3D model the beneficial effects of the co-cultivation of salivary gland epithelial cells with microvascular endothelial cells on the barrier properties.³⁵⁶ Especially, for future 3D models such as organ-on-chips or spheroids based on e.g. hiPSC-differentiated epithelial cells co-cultivated with surrounding cells (endothelial cells, fibroblasts, etc.) a comprehensive characterization of the barrier properties will be indispensable in order to understand how close the models will be to the human in vivo situation.

Disclosure statement

No potential conflict of interest was reported by the authors.