Under the conditions described here, single-cell suspensions from head kidney, anterior portion of trunk kidney, and spleen of rainbow trout give rise to populations of non-adherent cells with low buoyant density that are highly motile in culture. As with dendritic cells in mammals, when mixed with allogeneic spleen cells, tDCs induce potent proliferative responses compared to macrophages and B cells. As this is the functional definition of DCs in mammals, we conclude that tDCs serve a specialized APC function in fish analogous to the role of DCs in mammals. To our knowledge this is the first functional demonstration of DCs in fish. Further analyses suggest tDCs share other characteristics with mDCs as well. tDCs extend slender cytoplasmic dendrites, express surface MHCII, express TLR, costimulatory molecule, CD83, CD209, CXCR-4, CCR-7, IL-12p40, and MHC class II mRNAs, are phagocytic, and are activated by TLR-ligands. The kinetics of the activation of tDCs by TLR-ligands shows delayed kinetics when compared to mammals. However, several measures of immune responses in fish suggest that in vivo
immune responses occur more slowly than in mammals 
The rationale behind the approach taken here is that DCs could be generated from fish tissues using the same methods employed for the production of bone-marrow derived DCs in mammals. Typically, this involves the addition of growth factors to bone marrow cells. Initially, we added recombinant mammalian GM-CSF to trout cell cultures but found that we could generate tDCs even in its absence. This is similar to long-term stromal cell cultures from mouse spleen that support the production of immature DCs without the requirement for exogenous sources of GM-CSF 
. We therefore presume tDCs emerge from progenitor cells under the influence of endogenous growth factors present in cultures. As noted in the text, tDCs appear to develop from non-adherent monocytic cells in culture. However, when these cells were removed and the media replenished, tDCs continued to emerge for up to four months indicating that they may arise from adherent precursors in addition to non-adherent monocytic progenitors.
Although the studies reported here describe cultures from head kidney, anterior portion of trunk kidney, and spleen tissues combined, tDCs could also be generated from each of those tissues individually, as well as from the posterior portion of the trunk kidney. Larger numbers of melanomacrophages (along with unidentified debris) appeared in cultures containing posterior portions of the trunk kidney, and therefore these areas of the kidney were avoided. Interestingly, similar to mDCs, cells with the same morphology as tDCs could also be generated from peripheral blood mononuclear cells of rainbow trout or isolated from spleen using methods from the mammalian literature. These cells expressed MHCII, but functional studies to determine whether these cells are also capable of stimulating the MLR have yet to be performed.
tDCs possess distinctive multi-lobed nuclei that superficially resemble the nuclei of polymorphonuclear cells (PMNs). PMN nuclear lobes however, are circular in arrangement, while tDCs most often display a clover leaf-type arrangement with multiple lobes linked at a central point by thin extensions of nuclear material. Early publications describing mammalian DC cultures show cells with remarkably similar nuclear morphology to tDC cultures that do not stain with granulocyte antibodies, suggesting that these cells are a maturation stage of DCs in culture 
. Additionally, tDCs are not highly phagocytic, as would be expected of PMNs. tDCs do phagocytose beads, but the cells do not fill to capacity as is typically observed for PMNs and macrophages. The scarcity of granules in tDCs also supports our conclusion that these cells are not PMNs, and respresent a distinct subset of cells.
While only a limited number of specific antibodies are available for delineating cell types in fish, our data indicate that tDCs fail to bind either IgM or thrombocyte-specific antibodies. In contrast, rabbit antisera against rainbow trout MHC class II bound uniformly to tDCs. In addition to the surface expression of MHC class II, we show that CD83 and MHC class II genes are expressed in these cells using real-time RT-PCR. Surface protein expression of CD83, a specific marker for mature DCs in mammals, has recently received a great deal of scrutiny as its function is unraveled. While the full story has not been elucidated it does appear that its expression is correlated with MHC class II surface expression 
. The pattern of CD83 mRNA tissue expression is reported to correlate with that of MHC class II mRNA in rainbow trout, suggesting that, like mammals, CD83 and MHC class II expression may be correlated 
. CD83 gene expression has been reported in fish leukocytes including trout macrophages (where it is reportedly upregulated in response to LPS) 
, the RTS11 macrophage-like cell line 
, a melanin-producing leukocyte cell line 
, and the Atlantic salmon phagocytic TO cell line 
, as well as endothelial cells 
. CD83 mRNA expression is reported in a range of mammalian cells as well, including: T cells 
, B cells 
, neutrophils 
, monocytes, and macrophages, but stable protein surface expression is seen only in DCs 
. It is possible that CD83 mRNA expression is not a specific marker for DCs in fish; however, it should be noted that mRNA levels do not necessarily reflect surface expression of translated protein, and CD83 could undergo post-translational regulation in fish as it does in mammals 
. Despite mRNA expression in cells that are not DCs in fish, surface expression of CD83 protein may still prove to be a specific marker for DCs in teleosts. Generation of a CD83 specific antibody to test this has yet to be achieved. It should also be noted that along with expression of CD83 mRNA, apparent expression of transcripts for other DC markers (TLR, co-stimulatory molecule, CD83, CD209, CXCR-4, CCR-7, and IL-12p40) in tDCs based on PCR is still somewhat tentative given the nature of the samples and the lack of data regarding the expression of these markers in other fish cell types.
Irrespective of the role of CD83 in antigen-presentation, we found that CD83 mRNA transcripts, along with surface levels of MHC class II were upregulated in tDCs in response to addition of TLR-ligands to culture media. Upregulation of the expression of surface MHC class II and CD83 mRNA transcripts are consistent with the activation pattern seen in TLR-ligand treated mDCs and has not been shown in fish cells before. To examine this further, we investigated whether the phagocytic capacity of tDCs was also altered in response to PAMPs. In particular, we were curious as to whether tDCs would become less phagocytic, as mDCs do, following activation via toll-like receptor signaling. Inconclusively, we found that phagocytosis in tDCs after TLR-ligand treatment increased in some experiments and decreased in others (data not shown). We hypothesize that differences in the maturation state of tDCs from culture to culture could account for the inconsistent response. Relatively immature cells could become more phagocytic with the addition of TLR ligands, taking longer to reach a mature state where they would lose their phagocytic capacity. On the other hand, cells that are relatively mature may be phagocytic and become less so with the addition of TLR-ligands. The maturation state of any given culture could depend on the relative levels of endogenous cytokines present, which in turn may vary from culture to culture. Additionally, it is important to bear in mind that trout used in these studies were from out-bred populations and were not housed in specific pathogen free conditions. In fact, spleens of euthanized fish were frequently enlarged, suggesting an on-going immune response. These factors are likely to contribute significant variation in the maturation status of leukocytes from individual fish. As with mDCs, the relative maturation state of tDCs could potentially be normalized by the addition of exogenous cytokines to cell cultures, or by treatment with TLR-ligands such as CpG 
. In effect, our treatment of tDCs with TLR-ligands for four days resulting in surface MHCII upregulation in the ensuing population of cells suggests that tDCs can be matured by TLR-stimulation. Further studies on four-day stimulated tDCs are planned to characterize these cells, including assessment of changes in phagocytic capacity, as they may represent more uniformly activated tDCs.
MLRs have previously been demonstrated in several species of fish including rainbow trout 
. The seminal studies on APC function in fish were conducted in channel catfish. These studies established a role for “accessory cells” or APCs in fish by determining that Ig-negative lymphocyte (presumably T cell) responses to LPS or Con A required the presence of monocytes/macrophages, defined as peripheral blood leukocytes that adhered to baby hamster kidney cell microexudate-coated surfaces (presumed to be fibronectin) or Sephadex G-10 
. It was also observed that monocytes/macrophages, B cells, and to a lesser extent, Ig-negative lymphocytes, could stimulate the MLR, whereas Ig-negative lymphocytes were the only cells capable of acting as responders 
. The number of monocytes/macrophages that could be isolated in these studies was limited, preventing a direct comparison with B cells, although monocytes/macrophages appeared to be better stimulators. The monocyte/macrophage fraction of peripheral blood leukocytes used in these studies may be or contain the channel catfish-equivalent of rainbow trout DCs. Studies to determine the relevance of tDC culture methods to other species of fish, such as catfish, are an important next step.
When compared, fish and mammalian immune systems have notable similarities although anatomically they clearly differ. Fish do not have bone marrow. The head kidney, and in some species (such as rainbow trout), both the head kidney and spleen are the bone marrow equivalents. Additionally, fish do not have lymph nodes, the sites of antigen presentation and germinal center formation in mammals. It is thought that antigen presentation occurs in the spleen and possibly head kidney in fish, but this has yet to be demonstrated. Formation of germinal centers has not been observed in fish. The splenic architecture of fish shows organized B cell and T cell areas, similar to mammals, suggesting that in the absence of lymph nodes, the spleen functions as an important secondary lymphoid tissue, as it does in mammals. Immune responses in mice that lack lymph nodes and have disrupted splenic architecture are surprisingly robust, although delayed 
. Thus the lack of lymph nodes in fish may not represent a significant difference between fish and mammals in the way immune responses are mounted. In fact, the delayed immune responses seen in mice lacking functional lymph nodes invites conjecture that perhaps the comparatively slower immune responses in fish may also be partly attributable to their lack of lymph nodes.
This paper is an important advance in our understanding of how adaptive responses are initiated in fish, which is essential for rational vaccine design for aquaculture species. In mammals, direct targeting of antigen peptides to DCs has proven to elicit T cell responses that are superior to traditional antigen and adjuvant vaccines 
. Such approaches can be applied to fish vaccines, but rely on the identification of this cell type, as demonstrated here for rainbow trout.
Evolutionarily distant species have not traditionally been used as basic immunological models, however, information central to our current understanding of mammalian immunology has resulted from the study of such species. For example: the Toll gene was discovered in the invertebrate Drosophila
in 1985, and was subsequently shown to have a dual role in development and the innate immune response 
. Toll-like receptors were consequently identified in humans in 1997 and have since been demonstrated to be one of the main modes of sensing pathogens by the mammalian immune system 
. This important discovery, crucial to our current understanding of immunology, was based on work done in an invertebrate. Clearly study of our ancient vertebrate ancestors is relevant to our understanding of mammalian immunology. As with most things, it behooves us to start at the beginning.