This study investigated potential canine-specific markers for identifying MDSCs in the peripheral blood of dogs with cancer. We began our studies by first identifying potential phenotypic markers that could be used for identifying canine MDSCs. Prototypic MDSC phenotypic markers have been reported as CD11b+
in humans and CD11b+
in mice [1
]. In humans, CD15 (granulocytic) and CD14 (monocytic) identify subpopulations of these cells while, in mice, CD11b+
cells identify these cells as either granulocytic or monocytic, respectively. Unfortunately, the lack of readily available commercial canine antibodies has limited the identification of MDSCs and MDSC subsets in dogs.
CD11b, an integrin, is found on a variety of cells of myeloid origin and has been used as one phenotypic marker of MDSCs present in the mouse and man. In the dog, CD11b in known to be a marker of cells of myeloid origin, most specifically neutrophils [23
]. There is a commercially available anti-canine CD11b available which has been validated. In the mouse, Gr-1 is a granulocytic marker whose co-expression with CD11b is used to identify the immature myeloid cell population that represents MDSCs. Although CD11b is useful to identify myeloid (e.g. macrophages/DCs), neutrophils and mast cells in the dog, no Gr-1 equivalent for the dog has been described. Additionally, murine Gr-1 has demonstrated a variable ability to stain canine cells, and its individual components, Ly6G and Ly6C, are not reported to cross-react (unpublished observations) [21
In man, an equivalent antibody to Gr-1 does not exist. Rather, CD33, CD14 and HLA-DR (i.e. MHC class II) expression are used to specifically identify these cells. No canine specific, or cross-reacting, CD33 antibody is currently commercially available. Although human CD14 cross-reactive with canine cells, canine CD14 expression is variable on monocytes and likely may not truly mimic what is seen in humans [26
]. Additionally, standardized protocols and antibody concentrations using CD14 in the canine have not been established [27
]. However, a recent publication by Goulart et al. has demonstrated the use of negative labeling for CD14 expression to identify an MDSC population in the dog but no studies have investigated canine-specific granulocytic antibodies to identify an MDSCs in the dog [21
]. Considering the lack of anti-canine CD33 and human MDSC subpopulations express markers which are not commercially available (e.g. canine-specific CD15 or CD14), we chose to evaluate the levels of canine-specific CD11b and a granulocytic marker which models the phenotypic description for murine MDSCs.
Only two commercially available markers are reported to label canine granulocytes (i.e. CADO48A and DH59A) and our results show that CADO48A is able to distinguish separate populations that are not evident in DH59A stained populations from canine peripheral blood. While our focus was on canine-specific antibodies, we did evaluate the cross-reactivity of murine Gr-1 with canine peripheral blood samples. Murine Gr-1 (clone RB6-8C5) failed to stain canine peripheral blood cells, results which differ from the ability of this clone to stain canine cells by Goulart et al. [21
]. Our studies suggest that canine-specific CADO48A is an effective antibody to identify circulating myeloid cells in canine blood.
The ability of CADO48A to identify individual granulocytic populations of peripheral white blood cells (high versus low expression) as seen in Figure suggests that this antibody may be useful for identifying granulocytic versus monocytic subpopulation but purification and functional evaluation of such subpopulations were beyond the scope of this study. Our results demonstrate that CD11b and CADO48 staining is useful for detecting subpopulations of myeloid cells in the peripheral blood of dogs. From our optimization studies, we found that clinical samples can be stored up to 24 hours in EDTA, 48 hours in media and that cells that have been fixed following staining with flow antibodies can be analyzed up to 72 hours later without dramatically impacting the expression of CD11b+CADO48+ levels. From a diagnostic standpoint, it is important to know the relative expression levels of these markers depending on sample treatment and handling.
Following optimization, we next went on to determine whether tumor-bearing dogs demonstrated altered levels of myeloid cell populations. Our results show that a specific population of CD11blow
cells was increased in tumor-bearing dogs. The finding of an upregulated population of cells in tumor-bearing dogs expressing CD11b, as well as, the surrogate marker CADO48A, is suggestive of an MDSC phenotype in the canine. Classically, MDSC expression of CD11b is considered to be high. In our canine samples, we found that the expression in the canine patient to be low. While this finding is discordant from the literature on MDSCs in the mouse and human, CD11b expression is known to be variable both in healthy people at different time points and in animals with varying severity of inflammation [28
]. Additionally, it has been shown in human cell lines that during neutrophil maturation from a promyelocytic stage of development to a more terminally differentiated state CD11b expression starts low and eventually reaches high levels of expression once fully mature [30
]. If the cells in our tumor-bearing dogs represent an immature myeloid population, low CD11b expression may be representative of MDSCs. Additionally, work done by Furuhashi et al. has shown that DCs with high CD11b expression are more capable of eliciting a T-cell response in the pulmonary parenchyma of mice, indicating that increased CD11b expression is a relative marker of increased maturation and/or activation of DCs [31
]. Based on our results and previously published findings, low CD11b expression may in fact be indicative of an early myeloid derived cell population such as MDSC. Additionally, the cellular morphology of our purified CD11blow
cells is consistent with an immature and heterogeneous cellular morphology described for MDSC populations [1
] in mice which contain cells demonstrating monocytic, granulocytic and ring-shaped nuclear forms (Figure ). Thus, we postulated that CD11blow
represent an MDSC population.
In order to determine whether, indeed, CD11blowCADO48Alow were immunosuppressive, we utilized our in vitro model to generate canine MDSCs under the influence of a tumor environment and purified the CD11blowCADO48Alow population. As Figure demonstrates, CD11blowCADO48Alow were able to suppress the proliferation of responder canine lymphocytes. While cell sorting and functional evaluation of individual cancer patients were beyond the scope of this study, our data from purified CD11blowCADO48Alow (from control or tumor environment) verified that CD11blowCADO48Alow were immunosuppressive. Ongoing studies are evaluating the functional abilities and signaling pathways involved in the development of canine MDSCs using our in vitro MDSCs model.
While significant differences in R1 and P1 were found when tumor-bearing patients were grouped according to tumor type (Table ), it was somewhat surprising that P4 differences were not significant. The most likely reason for this is that subcategorization decreased the number of patients in each tumor type category which then resulted in a lack of statitistical significance. An additional prospective study with recruitment of more patients according to individual tumor types would likely address this possibility. One interesting aspect of our study is the mixed population of tumors evaluated. In the human literature, the predominance of tumors shown to upregulate MDSCs are carcinomas, including those arising from the pancreas, colon and lung [32
]. Unlike in humans, sarcomas are a relatively common tumor type seen in veterinary species and studies investigating MDSC populations in humans have found this cell type primarily in various carcinomas and melanoma [35
]. The fact that MDSCs were upregulated in our tumor-bearing population, regardless of tumor type suggests that common mechanisms exist between sarcomas and carcinomas for the induction of MDSCs in the canine population. Interestingly, melanomas appear to upregulate MDSCs regardless of species (Table ) [32-35]. Studies of MDSCs in dogs may be useful at dissecting the potential mechanisms by which carcinomas, sarcomas or melanomas differentially regulate MDSCs levels and why these may differ between species. Additionally, specific tumor types may uniquely upregulate specific MDSC subsets such as seen with the predominance of CD15+
MDSC or lineage negative MDSCs in patients with glioblastoma [36
]. At present, a detailed evaluated of canine MDSC subpopulations is not feasible but our data suggest that CADO48A may be a useful antibody for such distinctions.
cells have been identified as MDSCs in the peripheral blood of dogs (Goulart 2012) in both solitary and metastatic cancer and these cells were able to suppress T cell proliferation similar to what we found with CD11blow
cells . Our studies demonstrate that CADO48A is a useful marker to identify MDSCs in dogs and it is not known whether CD11blow
represent two distinct MDSC populations (e.g. granulocytic and monocytic, respectively) and, if so, whether these two populations have distinct functional profiles. The identification of MDSC subpopulations is likely to be useful for diagnostic and prognostic purposes but the characterization of these populations in canines is only in its infancy. Additional markers, such as S100A9, have been described to identify monocytic MDSCs in humans with colon cancer and cross-reacting S100A9 proteins are available [33
]. Ongoing studies evaluating various MDSC markers are needed to determine the phenotype and function of specific MDSC subpopulations (e.g. monocytic and granulocytic) in dogs. Utilization of both canine patients and an in vitro
model system as we provide here are likely to provide important data on MDSC phenotype and function and facilitate translational applications between canine and human MDSC studies.