Background
Natural killer cells were initially defined as a heterogeneous population of large granular lymphocytes that were able to spontaneously kill target cells. Human natural killer cells were later phenotypically defined as CD3
-/CD56
+ lymphocytes [
1], and can be divided into distinct functional subsets, the CD56
bright being mainly cytokine producers and CD56
dim being more effective killers [
2]. Mice and rats do not express CD56 on haematopoietic cells; thus NK cells in these species are instead defined as non-T lymphocytes bearing NKR-P1 molecules [
3], although the use of NKp46 as the standard NK cell identifier in the rat has been suggested [
4]. There is less evidence for a clear functional division of NK cells in rodents [
5]. Criteria for the definition of NK cells have not been clarified in most farm animal species [
6]. In the cow, NK-like cells have been described as CD3
-/CD2
+ lymphocytes [
7‐
10], and NK-like cells have been isolated on the basis of markers not commonly expressed by NK cells [
11]. Recently, the characterization of bovine NKp46 (CD335) enabled a more precise identification of bovine NK cells [
12]. The NKp46 receptor has been proposed as the most accurate marker for human NK cells [
13], as it is highly NK-restricted and uniformly expressed by cultured as well as resting NK cells [
14‐
16].
The CD2 molecule is known as an adhesion molecule as well as a receptor for activation on both NK cells and T-cells, reviewed in [
17]. In the cow, CD2 is frequently used as a cell marker in cellular immunology, and several anti-bovine CD2 monoclonal antibodies (mAbs) have been described [
18,
19]. The functional role of CD2 in bovine NK cells has not been previously studied, but NKp46
+ cells in cattle peripheral blood can be divided into distinct CD2
+ and CD2
- phenotypes, the latter typically comprising around 20% of the cells [
12]. Similar proportions have been reported in human blood [
20,
21], but only a few studies have been published that functionally describe CD2
- NK cells from blood in healthy humans [
22‐
24]. There does not seem to be a strong difference in expression of CD2 between the human CD56
bright and CD56
dim NK cell subsets [
2]. In the mouse, CD2
- cells has been associated with immature cell stages [
5], but blocking and knock-out experiments have shown that CD2 is not necessary for well-developed NK cell function [
25,
26]. Accordingly, CD2 has been classified as a co-receptor [
27], although the use of this term has been questioned, since apparently no known activating NK cell receptor stand out as dominating, unlike in T-cells [
16].
The scope of the present study was to investigate bovine NK cell subsets, defined by CD2 expression. Our results show that both subsets are fully capable of NK cell activity, and blocking experiments confirm a redundancy of CD2 for NK cell function. Furthermore, the CD2- subset is more activated in circulation and show a stronger proliferative and interferon-gamma (IFN-γ) producing capacity.
Discussion
The recent characterization of bovine NK cells [
12] allows this important field of innate immunity to be studied in the cow. Human NK cells can be divided into functional subsets depending on the level of CD56 expression, but like in rats and mice, this antigen is probably not present on NK cells in the cow [
11], thus not allowing a direct comparison with the human subsets. However, CD2 expression divides bovine NK cells into two distinct populations [
12], and the scope of the present study was to examine CD2
+ and CD2
- bovine NK cell subsets for functional properties and status of activation.
Around 20% of NK cells in peripheral blood of cattle did not express CD2, similar to findings in man [
20,
21]. During two weeks of primary culture with IL-2, the NK cell population changed towards a dominance of the CD2
- phenotype. This change tended to be more pronounced in young animals, but was also present in adults. No intermediate forms were seen at any point, indicating that the two subsets were discretely preserved. However, under these unfractioned culturing conditions, the initial CD2
- population appeared to acquire a low density of CD2 receptors. In contrast, no such CD2 acquisition was seen when CD2
- NK cells were cultured separately. More detailed studies would be required to explain this phenomenon, but a conceivable hypothesis could be intercellular exchange of receptors during cell culture, as recently demonstrated [
32]. We have consequently chosen to name these subsets CD2
-/low and CD2
+ when cultured in the presence of IL-2. IL-2 is reported to upregulate CD2 on human NK cells; partly as a minor intensity increase like observed here, but also as a proportional increase of the CD2
+ population [
16,
22], unlike our findings. Species differences may explain these conflicting proliferation patterns.
Interestingly, the CD2
- subset contained the highest relative numbers of IFN-γ
+ NK cells following cytokine stimulation, and IFN-γ secretion to the supernatant was also stronger following IL-2 stimulation. To our knowledge, experiments showing IFN-γ responses from corresponding subsets have not been reported in other species. Both subsets were cytotoxic against conventional target cells, the CD2
-/low most often, but not always, being the most effective killers. These findings are in agreement with reports from human NK cells where both subsets were comparably cytotoxic [
22]. We conclude that both subsets had fully developed natural killer cell functions, the CD2
-/low NK cells being the more effective IFN-γ producers.
To identify appropriate activation markers in bovine NK cells we analyzed antigens known to be upregulated on other activated bovine lymphocytes [
8,
33]. We observed IL-2-induced upregulation of CD25, CD44 and CD45RO, in accordance with reports from human NK cells [
34‐
37]. The unchanged CD11b expression and slight but transient CD11c upregulation in IL-2 culture was also similar to human NK cells [
20].
In peripheral blood NK cells, two of the investigated markers (CD44 and CD25) were highest on CD2
- subset. Expression of CD25 on NK cells has been associated with higher proliferative potential [
38], providing support for the increased proliferation of the CD2
- subset seen here. CD2, CD44 as well as the integrins are associated with cell adhesion, and notably, CD44 has been associated with binding to high endothelial venules [
36]. Hence, the differential expression of these as well as CD11c on NK cell subsets could possibly relate to adhesive and/or migratory properties, although it has been warned against inferring
in vivo migratory properties based on the expression of adhesion molecules alone [
39].
In contrast to CD44 and CD25, CD45RO was moderately brighter on the CD2
+ subset. Increased expression of CD45RO has been reported to be a transient phenomena in human NK cells upon activation [
37,
40]. , but the role of this molecule in NK cells remains unclear [
41,
42]. It is noteworthy that in parallel to our findings with NK cells, bovine gamma delta (γδ) T-cells have been reported to express more CD44 and less CD45 in a CD2
- subset, although it was not stated which isoform of CD45 that was targeted [
33]. Furthermore, human peripheral CD2
-/low T-cells were recently reported to show greater proliferative capacity as well as CD45RO expression, unlike CD2
+ T-cells; these were interpreted to be immature T-cells [
43]. They were by definition CD3
+, and therefore of a different lineage, but also in NK cells the lack of CD2 expression has been linked to immaturity [
5,
22]. The knowledge of NK cell development is still limited, but definition criteria for differentiation stages have recently been suggested [
44]. According to those, the CD2
- subset presented here would qualify as mature NK cells since they show full cytotoxicity and IFN-γ production. The findings that they comprise up to 30% of circulating NK cells and express all NK receptors investigated on the mRNA level would appear paradoxical if they represent immature stages. Further studies would be required to target such aspects.
To complement these studies on NK cell subsets defined by CD2 expression, we examined the role of the CD2 molecule itself on bovine NK cells, by employing two bovine anti-CD2 mAbs in functional NK cell assays. Cross-linking by one of the mAbs induced a clear IFN-γ response, similar to studies in human NK cells [
45]. Neither of the tested mAbs blocked cytotoxicity against NK-susceptible targets, and they did not induce redirected lysis of an FcR-bearing target. In human NK cells, significant blocking of cytotoxicity by CD2 mAbs usually does not occur [
46], and no redirected lysis effect was observed [
47], although cross-linking by conjugated mAbs against different epitopes could trigger cytotoxicity [
48‐
50]. Similar observations have been done in mice [
25]. The human CD2 molecule has later been shown to contain a continuum of overlapping epitopes, to which different mAbs may bind with various degrees of functional interference [
51]. On this background, our experiments cannot exclude a CD2 function in cytotoxicity. Nevertheless, one CD2 mAb that apparently bound to a reactive epitope, since it could trigger IFN-γ production, did not affect target lysis. Together with the observation that CD2
- NK cells were fully cytotoxic and produced IFN-γ, we conclude that, similar to other species, the CD2 molecule is not necessary for bovine NK cell function.
Bovine NK or NK-like cells have sometimes been defined as CD3
-/CD2
+ lymphocytes [
7‐
9]. This may partly have been founded on reports that bovine PBMCs expressing CD2 were more cytotoxic than CD2
- PBMCs [
10]. However, due to lack of NK-specific markers, those results could not be adjusted for the relative content of NK cells within each population. Our studies show that most NK cells in blood are CD2
+. Considering that the majority of PBMCs do not express CD2, the proportion of CD2
- NK cells would constitute only a small fraction of this population, hence providing a likely explanation for the cytotoxicity pattern shown by Campos
et al. [
10]. Furthermore, in a report where CD3
-/CD2
+ cells were classified as NK cells, CD3
-/CD2
-lymphocytes were reported as a not previously described "null cell" population [
8]. Our present and previous studies [
12] should provide evidence that these were most likely NK cells. In conclusion, we find that NK cell definitions based on a CD2
+ phenotype exclude an important subset of bovine NK cells.
Authors' contributions
PB designed the study and carried out cell isolation, separation and culture, flow cytometry, cytotoxicity assays, participated in IFN-γ assays, CFSE and gene expression analysis, performed literature searches and drafted the manuscript. IO and SK carried out CFSE proliferation analysis and intracellular IFN-γ assays. IB carried out rbIL-2 expression and gene expression analyses. GMJ carried out gene expression analyses, IFN-γ ELISAs and participated in cell isolation and culture. AKS conceived, coordinated and supervised the study, and participated in all experiments as well as manuscript preparation.