Logical Development of the Cell Ontology
© Meehan et al; licensee BioMed Central Ltd. 2011
Received: 16 September 2010
Accepted: 5 January 2011
Published: 5 January 2011
The Cell Ontology (CL) is an ontology for the representation of in vivo cell types. As biological ontologies such as the CL grow in complexity, they become increasingly difficult to use and maintain. By making the information in the ontology computable, we can use automated reasoners to detect errors and assist with classification. Here we report on the generation of computable definitions for the hematopoietic cell types in the CL.
Computable definitions for over 340 CL classes have been created using a genus-differentia approach. These define cell types according to multiple axes of classification such as the protein complexes found on the surface of a cell type, the biological processes participated in by a cell type, or the phenotypic characteristics associated with a cell type. We employed automated reasoners to verify the ontology and to reveal mistakes in manual curation. The implementation of this process exposed areas in the ontology where new cell type classes were needed to accommodate species-specific expression of cellular markers. Our use of reasoners also inferred new relationships within the CL, and between the CL and the contributing ontologies. This restructured ontology can be used to identify immune cells by flow cytometry, supports sophisticated biological queries involving cells, and helps generate new hypotheses about cell function based on similarities to other cell types.
Use of computable definitions enhances the development of the CL and supports the interoperability of OBO ontologies.
Biological databases use ontologies to connect common elements between disparate data sets to facilitate searching and analysis. These ontologies are structured hierarchies of logically connected classes intended to represent biological entities and their relationships to each other. These ontologies are often conceived of as graph-theoretic structures, with classes as nodes and relationships as edges. Biological ontologies must evolve to capture the ever-increasing complexity of data being generated by the life sciences community. The widely used Gene Ontology (GO), for instance, has grown from about 4,000 classes at its inception to its current form that encompasses nearly 30,000 classes and over 50,000 relationships . Maintenance and development of such a large ontology is burdensome, a problem that is compounded by the creation of other ontologies with overlapping domains. In response to these concerns, the Open Biological and Biomedical Ontologies (OBO) Foundry was created to provide a set of principles to help standardize and coordinate a collection of ontologies to cover the biomedical domain . One of these principles is that an OBO Foundry candidate ontology must have clearly specified and delineated content that is orthogonal to other OBO ontologies. Within the OBO Foundry, the Cell Ontology (CL) is the designated reference ontology for the representation of in vivo cell types from all biology .
First developed in 2004 , the CL is composed of about 1000 classes connected via is_a and develops_from relationships. The CL classifies cell types according to multiple criteria, including function, histology, lineage and taxonomy. These different modes of classification are vital for the users of the CL; however, the initial classification was done manually and not in accordance with commonly accepted ontology engineering principles such as normalization and modularity . The resulting ontology contained numerous mistakes and omissions, and proved difficult to maintain and extend.
We have revised and extended the hematopoietic cell branch of the CL following a normalized approach in which computable definitions (sometimes called "logical definitions" or "cross-products") for classes are constructed in a modular fashion, using relationships to classes from other ontologies. These computable definitions can be expressed in ontology formats and languages such as OBO or OWL [2, 5], and are treated as equivalence relationships between the defined class and some conjunction of classes. For example, the class "nucleate erythrocyte" can be defined as equivalent to the class of things that are both erythrocytes and that have a nucleus as a part. In OWL this can be written as:
"nucleate erythrocyte" EquivalentTo erythrocyte and has_part some nucleus where "erythrocyte" is declared in the cell ontology and nucleus is declared in the GO cell component ontology.
These computable definitions not only provide an integration point between ontologies, but they also allow tools called reasoners to automatically construct the ontology hierarchy (for example, between "nucleate erythrocyte" and "nucleate cell"). Our computable definitions all follow a genus-differentia structure, in which a class is defined by refining an existing more general class (the genus - in the above example this would be "erythrocyte") and one or more differentiating characteristics (the differentiae - in the above example this is the has_part relationship to nucleus) [6, 7]. By adding disjointness constraints between classes in the ontology, reasoners can also detect errors in the graph structure. Two classes are disjoint if they cannot share any members. If during ontology construction a disjointness constraint is violated, the reasoner will flag the error. We also supply textual definitions that mirror the computable definitions, but are intended for human readers.
Other complex biomedical ontologies have employed computable definitions to support reasoning and help maintain the logical soundness of their structure including the Systematized Nomenclature of Medicine-Clinical Terms (SNOMED-CT), the Gene Ontology Next Generation Project (GONG), and the Ontology for Biomedical Investigations (OBI) [8–10]. The ontological structure of the GO has been improved in a similar manner by parsing class names into computable definitions; this approach originally added 223 missing relationships and is now routinely used in ongoing development . Masci et al confirmed this process is well suited for cell types by using computable definitions in the design of a dendritic cell ontology consisting of 29 classes . They defined dendritic cell types using cell surface expression of proteins as a major axis of differentia. Thus, a "CD8_alpha-negative, CD11b-negative dendritic cell" is a "conventional dendritic cell" (the genus) that "lacks_plasma_membrane_part CD8_alpha and CD11b" (the differentia). Masci et al found significant benefits in the amount and accuracy of information contained in the ontology compared to the information about dendritic cells available in the original implementation of the CL.
By integrating Masci et al.'s approach with our recent reorganization of the hematopoietic branch of the CL , we have created a new sub-ontology called "Hemo_CL". This ontology consists of over 330 hematopoietic cell classes (including 47 new additions) that have computable definitions generated using classes exclusively from ontologies in the OBO library. We demonstrate that computable definitions allow the use of automated reasoners to find mistakes in the ontology and to infer new relationships between classes. This approach also enriches other OBO ontologies by illuminating the need for new classes, and by highlighting ambiguous definitions of cell types in the scientific literature. Finally, an unexpected association between two classes highlights the potential use of computable definitions to generate new hypotheses regarding the nature of cell types.
Linking to OBO library ontologies by computable definitions
We make frequent use of the has_plasma_membrane_part relation as defined by Masci et al, in conjunction with classes from the Protein Ontology . For example, a "CD4-positive, alpha-beta T cell (CL:0000624)" is defined as " a mature alpha-beta T cell that has_plasma_membrane_part CD4 (PRO:000001004). 278 cell type classes defined according to this template have been generated, reflecting the common use of flow cytometry to distinguish hematopoietic cells based on cell surface markers. In some cases, cell types are distinguished not by the expression of single protein on the surface but rather by a protein complex, in which case we use classes from the GO-Cellular Component ontology (GO-CC). For example, a "gamma-delta T cell (CL:0000798)" is defined as "a T cell that has_plasma_membrane_part gamma-delta T cell receptor complex (GO:0042106)". Certain cell types are defined by the absence of a cell surface protein, or by difference in the quantity of cell surface protein relative to other cell types . In Figure 1b, the CD56-bright natural killer cell is partially defined as "mature natural killer" that lacks expression of one cell surface marker while expressing two other cell surface markers. One of these cell surface markers is expressed at higher levels than the geometric mean expression found for other leukocyte cell types and thus has a computable definition that uses the "has_high_plasma_membrane_amount" "has_low_plasma_membrane_amount" relationships as defined in Masci . This example also illustrates how using computable definitions with curated ontologies eliminates the confusion that may result from the use of a synonym ("CD56" in place of "neural cell adhesion molecule 1") in the common name of a cell type class.
Translated OBO relationships used in Hemo-CL
Simple OWL translation
Expanded OWL translation
has_plasma_membrane_part some ?Y
has_part some (GO:'plasma membrane' and has_part some ?Y)
lacks_plasma_membrane_part value ?Y
has_part exactly 0 (GO:'plasma membrane' and has_part some ?Y)
capable_of some ?Y
bearer_of some (realized_by only ?Y)
participates_in some ?Y
has_completed some ?Y
transformation_of some (participates_in some ?Y)
has_not_completed some ?Y
not transformation_of some (participates_in some ?Y)
Other OBO ontologies used in computable definitions are the GO-biological processes (GO-BP) , UBERON , and PATO . The majority of computable definitions that contain GO-BP classes have the structure "D = capable_of a GO-BP". In the example presented in Figure 1b, the natural killer cell type is partially defined from its parent cell type by being capable of interferon gamma production. Here, the capable _of relationship states that instances of this cell type have the potential to produce interferon gamma given the right biological context, but it does not necessitate that all instances at all times are producing interferon gamma. In the cases of those few cell types that are characterized by having all instances at all times actively engaged in a biological process, the participates_in relationship is used. Pro-B cells, for example, are described as being in the "early stages of recombination of B cell receptor genes", so a computable definition using "participates_in immunoglobulin V(D)J recombination (GO:0033152)" is appropriate. Other cell types are characterized as having executed a process that changed their cell type. In these cases, the has_completed relationship is used as in a "mature alpha-beta T cell" is an "alpha-beta T cell that has_completed T cell selection (GO:0045058)". The part_of relationship is used with the species-neutral UBERON anatomy ontology when anatomical location is critical to defining a cell . For example, a "Kupffer cell" is "a tissue-resident macrophage that is part_of hepatic sinusoid (UBERON:0001281)".
Clinical laboratories routinely identify granulocyte cell types both by flow cytometry and by staining characteristics. As we wanted to maximize the utility of the ontology, we decided to include both axes of differentia in our computable definitions. In the cases of granulocytes where cell types are differentiated based on nucleus or cytoplasm morphology, computable definitions are first generated between GO-CC and PATO  classes to create new cellular component classes. For example, "reniform nucleus" is defined as "nucleus (GO:0005634) that has_quality reniform (PATO:0001871)". Then basophilic metamyelocyte is defined as an "immature basophil that has_part reniform nucleus". The classes generated in this manner were given a separate namespace (cellular phenotype) and the prefix (CP) to distinguish them from the CL and are stored as part of the CL file. These classes may be incorporated into the GO or a phenotype ontology at a future date.
Addition of species-specific cell types
New cell types have been added to represent hematopoietic progenitor cell types that are currently an intense focus of research (Figure 2b). For Hemo_CL, progenitor cell classes are organized under multipotent progenitor, oligopotent progenitor, and lineage restricted progenitor cell types in a commonly used schema . Multipotent progenitors develop from stem cells and have the ability to develop into any hematopoietic cell type but have limited self-renewal capability. Oligopotent progenitors develop from multipotent progenitors, are unable to self renew, and are classified by which hematopoietic cell types they can differentiate into. As with stem cells, there are species-specific differences in the surface markers that necessitate the use of more granular cell type classes. These species-specific cell types are placed under species neutral cell types but are connected to each other through develops_from relationships (Figure 2b).
Ontology construction aided by computable definitions
Commonality between classes discovered by automated reasoners
Some ambiguities resulted from how a cell type is defined in the scientific literature. The automated reasoning for the "mature NK T cell" provided an interesting example. Reasoning over the ontology failed to infer an is_a relationship between "NK T cell" and the "natural killer cell" despite "NK T cell" having the synonym "natural killer T cell" (not shown). Investigation of the literature revealed that the name "natural killer T cell" is controversial; the origin of the name "NK T" cell refers to cell surface expression of the NK1.1 protein and was not intended to imply a direct link to natural killer cells . However some researchers feel there is enough functional overlap between "NK-" and "NK T-" cell types to justify the "natural killer T cell" moniker . In Hemo_CL, the synonym is left in place as it reflects common usage, but NK T cell types and natural killer cell types are kept in separate branches of the ontology as inferred by the reasoner.
One relationship the reasoner does infer for mature "NK T cell" is that it is a sub-type of "mucosal associated invariant T cells" (MAIT) (Figure 6c). This result comes from both classes being alpha-beta T cell types that are capable_of interacting with MHC-Ib protein complexes through invariant T cell receptors. Investigating whether these two cell types should be grouped together, we hypothesized that other commonalities might exist between the two cell types. Upon review of the literature, we discovered some researchers have renamed the MAIT cells as "mucosal-NKT" cells based on the expression of NK1.1 by MAIT cells, in addition to their association with MHC-Ib family members [22, 23]. Recently, two different research groups have demonstrated across species that MAIT cells are quickly activated and respond to bacterial antigens in a manner similar to NK T cells [24, 25]. Following from these findings, we created an "innate effector T cell" class defined as "effector T cell capable_of innate immune response (GO:0045087)" (Figure 6d). The computable definitions for NK T cells and MAIT were refined to reflect their capability of participating in an innate immune response. When the ontology is reasoned over, NK T cells and MAIT T cells are grouped under this term (Figure 6d). This serves as an example of how a well-structured ontology that has computable definitions and is updated to represent the most current research can be used to generate hypotheses.
This work examines the advantages of using computable definitions to define cell type classes. We demonstrate that the use of automated reasoners coupled with computable definitions assist curators by automatically inferring the ontology hierarchy and detecting inconsistent statements. Computable definitions also highlight areas that do not represent the current state of knowledge, leading in this case to the addition of 47 new cell classes. Describing hematopoietic cell types with computable definitions using multiple ontologies strengthens the central role that CL has among the other OBO ontologies. Biological processes, protein expression, anatomical location, and phenotypic qualities are all integrated with the CL through computable definitions. This approach enriches other OBO ontologies by suggesting new classes for other ontologies when a sufficiently granular class needed for a computable definition is lacking. Creation of more granular CL classes also provides points of reference for generating more specificity in other OBO Foundry ontologies. As part of the generation of computable definitions, class requests were made of several of the OBO ontologies including the GO, PRO, and PATO. Such interactions increase the interoperability of the OBO Foundry ontologies.
Computable definitions for hematopoietic cell types have importance beyond ontology maintenance and integration for the wider scientific community. Sophisticated queries that incorporate cell types, protein expression, anatomical location, and biological processes are now possible. From Hemo_CL a user can inquire, for example, which hematopoietic cell types express a FoxP3 protein family member and are capable of regulating the immune system. Another type of query that Hemo_CL is particularly suited for is flow cytometry as cell markers is one of the major axes of differentia used to distinguish cells. We are currently setting up several collaborations to develop this use of the ontology further.
One potential benefit of logically defining cell types is the unexpected association of cell types generally not considered to be in the same grouping. We noted, in this regard, that mature NK T cells were inferred by the reasoner to be associated with mucosal invariant T cells, a view held by some  although not part of our initial representation. From this inference, we hypothesize that MAIT and NK T cells would have similarities in biological function. Similar hypotheses about the two cell types were developed by experts in the field, recently tested, and found to be true [24, 25]. We believe hypothesis-generation represents a future use of well-structured biomedical ontologies that have computable definitions.
Our next focus for the CL is to examine and extend the representation of neuronal cell types including adding computable definitions. While the nervous system and the hematopoietic system are generally thought of as unrelated or even disjoint from each other (i.e. the immune privileged state of the nervous system) , there are intriguing commonalities among their cell types. The cell-to-cell contact that occurs between lymphocytes and antigen presenting cells has cellular structures similar to those in neuronal synapses (the "immune synapse") [28, 29] and there is also expression of numerous immune proteins by neurons including MHC-I protein complexes that are important in neuronal plasticity . By having computable definitions for these two branches, associations between cell types based upon biological function or protein expression may generate new hypotheses for testing neural-immune interactions.
We generated new hematopoietic cell types to represent entities described in current research and added computable definitions for all hematopoietic cell types using classes from other OBO library ontologies. These computable definitions allow the use of automated reasoners to infer relationships between the classes based on the inherent logic. Unexpected inferred relationships are found upon investigation to result from mistaken or incomplete computable definitions of classes, and in at least one case reflect the ambiguous use of the term in the biomedical literature. The inferred relationships also permit placement of cell classes into functional groups without the problems associated with multiple inheritance in an asserted hierarchy. Requesting new classes needed for a computable definition also enriched other ontologies. Finally, we believe inferred relationships between branches of the CL will lead to unexpected associations between disparate areas of biology and help in the generation of new hypotheses.
A sub-ontology of the CL called Hemo_CL was created by taking the is_a descendants of hematopoietic cell type and creating a new ontology. This new ontology is available at:
(note that this ontology has been integrated into the main CL ontology, but a record of the Hemo_CL subset as described in this paper will remain in the source control system. The revision number described in this paper is hemo_CL.obo revision 1.47).
Computable definition generation used a method modified from that described by Masci et al, where 29 dendritic cell types were redefined as cross-product classes . Computable definitions take the form "a C is_a G that D" where C is the hematopoietic cell class being defined, G is the more general class and D is the set of differentia that characterize instances of C from instances of D. The general class is the asserted is_a parent to the class being defined. Differentia (D) are defined as relationships to other class classes that reside in the OBO Foundry ontologies Gene Ontology (GO), Protein Ontology (PRO), Uber Anatomy Ontology (UBERON), Phenotypic Quality ontology (PATO), and CL. As cell types are considered independent continuants , only relationships that include this entity in their domain are used. Relationships used include those from the Relationship Ontology (RO) and its proposed extension, and the capable_of relationship described in our previous publication . The computable definitions for the 29 dendritic cell types described by Masci et al  containing the relationships has_disposition and has_function were converted to a definition containing the relationship capable_of.
For reasoning we use a combination of OBO-Edit and the OBO-Edit reasoner with Protégé 4 and OWL reasoners. We use a standard translation to OWL in which OBO format relationships are translated to SomeValuesFrom expressions in OWL by using the NCBO OBOinOWL mapping option that is part of OBO-Edit. In order to take advantage of OWL constructs such as negation and nested class expressions from within OBO format and OBO-Edit we use a macro expansion strategy. First we translate to OWL, then we selectively expand certain expressions as depicted in Table 1. The expansion of the capable_of relation does not affect reasoning but is required for compatibility with BFO. All the relations used in the expansions are defined in the new version of the RO ontology. Directly expressing the semantics of has_low_plasma_membrane_part and has_high_plasma_membrane_part is outside the scope of OWL. We adopt the formal textual definitions described in Masci et al  for has_low_plasma_membrane_amount and has_high_plasma_membrane_amount, but cannot provide an expansion rule that directly expresses the semantics in OWL.
Ontology editing was performed using OBO-Edit 2.1. Both the rule-based and the link piler reasoners were used to reason over the ontology. A MIREOT strategy  was used to bring in selected subsets of external ontologies into the hemo-CL ontology (this is implemented in OBO-Edit 2.1 using the save is_a closure option in the save menu). Figures are either screenshots from OBO-Edit 2.1 (Figures 3, 4, 5 and 6) or are Graphviz images exported from OBO-Edit and modified by Omnigraph Professional 5.2.1 (Figures 1 and 2).
OBO Cell Ontology
GO biological process
GO cellular component ontology
Open Biological and Biomedical Ontology
hematopoietic stem cells, OWL: Web Ontology Language
Phenotype and Trait ontology
TFM, CJM, AA, and ADD contributions were supported by an NHGRI-funded, ARRA administrative supplement grant HG002273-09Z to the parent grant, HG002273, to the Gene Ontology Consortium; Judith A. Blake, Michael Ashburner, Suzanna Lewis, J. Michael Cherry, PIs. LGC's contributions were supported by an NIAID-funded R01 (AI077706) and a Career Award at the Scientific Interface from the Burroughs-Wellcome Fund. This work was supported by the Director, Office of Science, Office of Basic Energy Sciences, of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231.
- The Gene Ontology in 2010: extensions and refinements Nucleic Acids Res 2010, 38: D331–335. 10.1093/nar/gkp1018
- Smith B, Ashburner M, Rosse C, Bard J, Bug W, Ceusters W, Goldberg LJ, Eilbeck K, Ireland A, Mungall CJ, Leontis N, Rocca-Serra P, Ruttenberg A, Sansone S, Scheuermann RH, Shah N, Whetzel PL, Lewis S: The OBO Foundry: coordinated evolution of ontologies to support biomedical data integration. Nat Biotechnol 2007, 25: 1251–1255. 10.1038/nbt1346PubMed CentralView ArticlePubMedGoogle Scholar
- Bard J, Rhee SY, Ashburner M: An ontology for cell types. Genome Biol 2005, 6: R21. 10.1186/gb-2005-6-2-r21PubMed CentralView ArticlePubMedGoogle Scholar
- Rector AL: Modularisation of domain ontologies implemented in description logics and related formalisms including OWL. Proceedings of the 2nd international conference on Knowledge capture Sanibel Island, FL, USA: ACM; 2003: 121–128. Accessed 19 August 2010 [http://portal.acm.org/citation.cfm?id=945664] Accessed 19 August 2010
- Rector A: Defaults, context, and knowledge: alternatives for OWL-indexed knowledge bases. Pac Symp Biocomput 2004, 226–237.Google Scholar
- Hill DP, Blake JA, Richardson JE, Ringwald M: Extension and integration of the gene ontology (GO): combining GO vocabularies with external vocabularies. Genome Res 2002, 12: 1982–1991. 10.1101/gr.580102PubMed CentralView ArticlePubMedGoogle Scholar
- Mungall CJ, Bada M, Berardini TZ, Deegan J, Ireland A, Harris MA, Hill DP, Lomax J: Cross-Product Extensions of the Gene Ontology. J Biomed Inform 2010. Accessed 18 February 2010 [http://www.ncbi.nlm.nih.gov/pubmed/20152934] Accessed 18 February 2010Google Scholar
- International Health Terminology Standards Development Organisation SNOMED Clinical Terms User Guide. Copenhagen 2010:
- Wroe CJ, Stevens R, Goble CA, Ashburner M: A methodology to migrate the gene ontology to a description logic environment using DAML+OIL. Pac Symp Biocomput 2003, 624–635.Google Scholar
- Brinkman RR, Courtot M, Derom D, Fostel JM, He Y, Lord P, Malone J, Parkinson H, Peters B, Rocca-Serra P, Ruttenberg A, Sansone S, Soldatova LN, Stoeckert CJ, Turner JA, Zheng J: Modeling biomedical experimental processes with OBI. J Biomed Semantics 2010, 1(Suppl 1):S7.PubMed CentralView ArticlePubMedGoogle Scholar
- Mungall CJ: Obol: integrating language and meaning in bio-ontologies. Comp Funct Genomics 2004, 5: 509–520. 10.1002/cfg.435PubMed CentralView ArticlePubMedGoogle Scholar
- Masci AM, Arighi CN, Diehl AD, Lieberman AE, Mungall C, Scheuermann RH, Smith B, Cowell LG: An improved ontological representation of dendritic cells as a paradigm for all cell types. BMC Bioinformatics 2009, 10: 70. 10.1186/1471-2105-10-70PubMed CentralView ArticlePubMedGoogle Scholar
- Diehl AD, Augustine AD, Blake JA, Cowell LG, Gold ES, Gondré-Lewis TA, Masci AM, Meehan TF, Morel PA, Nijnik A, Peters B, Pulendran B, Scheuermann RH, Yao QA, Zand MS, Mungall CJ: Hematopoietic cell types: Prototype for a revised cell ontology. J Biomed Inform 2010, in press. http://www.ncbi.nlm.nih.gov/pubmed/20123131 http://www.ncbi.nlm.nih.gov/pubmed/20123131Google Scholar
- Day-Richter J, Harris MA, Haendel M, Lewis S: OBO-Edit--an ontology editor for biologists. Bioinformatics 2007, 23: 2198–2200. 10.1093/bioinformatics/btm112View ArticlePubMedGoogle Scholar
- Natale DA, Arighi CN, Barker WC, Blake J, Chang T, Hu Z, Liu H, Smith B, Wu CH: Framework for a protein ontology. BMC Bioinformatics 2007, 8(Suppl 9):S1. 10.1186/1471-2105-8-S9-S1PubMed CentralView ArticlePubMedGoogle Scholar
- Mungall CJ, Ruttenberg A, Osumi-Sutherland D: Taking shortcuts with OWL using safe macros. Nature Precedings 2010. [http://precedings.nature.com./documents/3574/version/1]Google Scholar
- Washington NL, Haendel MA, Mungall CJ, Ashburner M, Westerfield M, Lewis SE: Linking human diseases to animal models using ontology-based phenotype annotation. PLoS Biol 2009, 7: e1000247. 10.1371/journal.pbio.1000247PubMed CentralView ArticlePubMedGoogle Scholar
- Gkoutos GV, Mungall C, Dolken S, Ashburner M, Lewis S, Hancock J, Schofield P, Kohler S, Robinson PN: Entity/quality-based logical definitions for the human skeletal phenome using PATO. Conf Proc IEEE Eng Med Biol Soc 2009, 1: 7069–7072.Google Scholar
- Chao MP, Seita J, Weissman IL: Establishment of a normal hematopoietic and leukemia stem cell hierarchy. Cold Spring Harb Symp Quant Biol 2008, 73: 439–449. 10.1101/sqb.2008.73.031View ArticlePubMedGoogle Scholar
- Curotto de Lafaille MA, Lafaille JJ: Natural and adaptive foxp3+ regulatory T cells: more of the same or a division of labor? Immunity 2009, 30: 626–635. 10.1016/j.immuni.2009.05.002View ArticlePubMedGoogle Scholar
- Godfrey DI, MacDonald HR, Kronenberg M, Smyth MJ, Van Kaer L: NKT cells: what's in a name? Nat Rev Immunol 2004, 4: 231–237. 10.1038/nri1309View ArticlePubMedGoogle Scholar
- Savage AK, Constantinides MG, Han J, Picard D, Martin E, Li B, Lantz O, Bendelac A: The transcription factor PLZF directs the effector program of the NKT cell lineage. Immunity 2008, 29: 391–403. 10.1016/j.immuni.2008.07.011PubMed CentralView ArticlePubMedGoogle Scholar
- Wingender G, Kronenberg M: Role of NKT cells in the digestive system. IV. The role of canonical natural killer T cells in mucosal immunity and inflammation. Am J Physiol Gastrointest Liver Physiol 2008, 294: G1–8. 10.1152/ajpgi.00437.2007View ArticlePubMedGoogle Scholar
- Le Bourhis L, Martin E, Péguillet I, Guihot A, Froux N, Coré M, Lévy E, Dusseaux M, Meyssonnier V, Premel V, Ngo C, Riteau B, Duban L, Robert D, Rottman M, Soudais C, Lantz O: Antimicrobial activity of mucosal-associated invariant T cells. Nat Immunol 2010, 11: 701–708. 10.1038/ni.1890View ArticlePubMedGoogle Scholar
- Gold MC, Cerri S, Smyk-Pearson S, Cansler ME, Vogt TM, Delepine J, Winata E, Swarbrick GM, Chua W, Yu YYL, Lantz O, Cook MS, Null MD, Jacoby DB, Harriff MJ, Lewinsohn DA, Hansen TH, Lewinsohn DM: Human mucosal associated invariant T cells detect bacterially infected cells. PLoS Biol 2010, 8: e1000407. 10.1371/journal.pbio.1000407PubMed CentralView ArticlePubMedGoogle Scholar
- Tilloy F, Treiner E, Park SH, Garcia C, Lemonnier F, de la Salle H, Bendelac A, Bonneville M, Lantz O: An invariant T cell receptor alpha chain defines a novel TAP-independent major histocompatibility complex class Ib-restricted alpha/beta T cell subpopulation in mammals. J Exp Med 1999, 189: 1907–1921. 10.1084/jem.189.12.1907PubMed CentralView ArticlePubMedGoogle Scholar
- Tian L, Rauvala H, Gahmberg CG: Neuronal regulation of immune responses in the central nervous system. Trends Immunol 2009, 30: 91–99. 10.1016/j.it.2008.11.002View ArticlePubMedGoogle Scholar
- Reichardt P, Dornbach B, Gunzer M: APC, T cells, and the immune synapse. Curr Top Microbiol Immunol 2010, 340: 229–249. full_textPubMedGoogle Scholar
- Carrasco YR: Molecular and cellular dynamics at the early stages of antigen encounter: the B-cell immunological synapse. Curr Top Microbiol Immunol 2010, 340: 51–62. full_textPubMedGoogle Scholar
- Boulanger LM: Immune proteins in brain development and synaptic plasticity. Neuron 2009, 64: 93–109. 10.1016/j.neuron.2009.09.001View ArticlePubMedGoogle Scholar
- Courtot Melanie, Gibson Frank, Lister AllysonL, Malone James, Schober Daniel, Brinkman RyanR, Ruttenberg Alan: MIREOT: the Minimum Information to Reference an External Ontology Term. Nature Precedings 2009. [http://precedings.nature.com./documents/5292/version/1]Google Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.