Background
Certain myeloid cell populations are involved in the pathogenesis of both chronic inflammation and cancer [
1‐
3]. These myeloid cells are sensitive to environmental cues and display a high degree of plasticity. Thus, cells with a similar phenotype can either promote or suppress immune responses depending on the local microenvironment [
4,
5].
To survive and expand, tumors must evade the host’s immune system [
6]. Interestingly, myeloid cells located within tumor tissue are highly immunosuppressive [
7‐
9], as a result of various signaling molecules provided by tumors and stromal cells. The most prominent myeloid cell populations located in tumors are tumor-associated macrophages (TAM) and Gr1
+ cells. The Gr1
+ population can be subdivided into Ly6C
hi monocytes and Ly6G
hi granulocytes [
10]. The Ly6C
hi cells can within the tumor give rise to different populations of TAM [
11,
12]. Additionally, Ly6C
hi and Ly6G
hi cells present within tumor tissue are non-proliferative, have a short half-life, and thus need to be constantly replenished [
11,
13‐
16]. This may, however, not apply to the F4/80
hi TAM whose numbers can also be maintained by local proliferation [
12,
17,
18].
Tumors promote the expansion of myeloid cells by producing various pro-inflammatory molecules such as GM-CSF, G-CSF, IL-1β and IL-6 [
8,
9,
19,
20]. Depending on the tumor model used, these cells may expand either in the bone marrow or the spleen [
9]. The expansion in the spleen creates a myeloid cell reservoir and a source of Ly6C
hi cells that can be recruited into tissues [
21,
22]. A role of this splenic myeloid cell reservoir during tumor growth was identified, as it was shown that splenectomized mice displayed reduced tumor progression, which was associated with a decrease in Ly6C
hi and Ly6G
hi cells in the tumor area [
14,
23]. In this setting, an increase in granulocyte-macrophage progenitors (GMP), that were able to generate the myeloid cells recruited to the tumor, was identified within the Lin
− c-kit
+ Sca1
− cells in the spleen [
14,
23]. A similar phenomenon was also observed in other models of inflammatory disease such as atherosclerosis and colitis [
24,
25].
Quinoline-3-carboxamides (Q compounds) are small molecule immunomodulators. One such Q compound, laquinimod, is currently in a phase III cinical trial for multiple sclerosis (NCT01707992). Another Q compound, tasquinimod (ABR-215050), has shown proof of concept in castration-resistant prostate cancer [
26,
27]. In pre-clinical settings, tasquinimod has been shown to potently reduce the growth of several transplantable mouse and human xenograft tumors [
28‐
34]. The reduced tumor growth was, in some of these studies, associated with anti-angiogenic effects [
28,
29,
34]. Further, this compound was also shown to modulate the function of TAM and reduce immune suppression [
33,
34]. Previous work has identified S100A9 as a target for the Q compounds, which prevent S100A9 interaction with its receptors Toll-like receptor 4 (TLR4) and receptor for glycation end-products (RAGE) [
35]. By binding to these receptors, S100A9, a well-known alarmin, induces the transcription of various proinflammatory genes and thus promotes inflammatory responses [
36‐
38]. S100A9 has, however, also been implicated in tumor development as it was shown to be important for the accumulation of suppressive myeloid cells and the negative regulation of the maturation of these cells to dendritic cells [
39‐
41].
Our previous work demonstrated that the Q compound paquinimod, which is structurally similar to tasquinimod, reduced the accumulation of Ly6C
hi cells and SiglecF
+ eosinophils in a model of sterile acute inflammation [
42]. Further, we could show that the ameliorating effect of this compound in acute EAE, a mouse model for multiple sclerosis, operated early during the induction phase of the autoimmune response [
43]. In this model, paquinimod also reduced the immunization-induced splenic myelopoiesis. In the current study, we test the hypothesis that the anti-tumor effects of tasquinimod could in part be mediated through an impact on alterations in myelopoiesis and myeloid cell recruitment to tumors, as these cells are known to promote tumor growth by their protumorigenic properties. Herein, we provide evidence in support of this hypothesis.
Methods
Mice and treatment
Wild type female BALB/c and C57Bl/6 mice were purchased from Taconic Europe (Ry, Denmark). All animal experiments were performed with the permit of the local committee on the ethics of animal experiments of Malmö and Lund (permit M12-13). To study the effects of the Q compound tasquinimod, female mice at the age of 7–10 weeks were treated with tasquinimod dissolved in drinking water corresponding to a daily dose of about 25 mg/kg body weight/day. Tasquinimod was provided by Active Biotech, Lund, Sweden.
Tumor cell lines
The 4 T1 mammary carcinoma and the B16-F10 melanoma cell lines were initially obtained from ATCC and provided to us by Active Biotech. The EG7 cell line (OVA-transfected EL4 lymphoma cell line) [
44] was obtained from Dr Clotilde Thery, Institute Curie, INSERM U932, Paris, France. The cell lines were expanded, frozen in aliquots and new aliquots regularly used for the
in vivo experiments. The cells were cultured in RPMI medium (RPMI-1640 supplemented with 10 % fetal calf serum, 10 mM HEPES, 1 mM sodium pyruvate, 100 U/ml penicillin-streptomycin and 50 μM β-mercaptoethanol; all supplements from Invitrogen Life Technologies, Paisley, UK) at 37 °C, 5 % CO
2. For trypsinization of 4 T1 cells, trypsin-EDTA (Sigma-Aldrich, St. Louis, MO) was briefly added to cells at approx. 80 % confluence and the cells were washed with RPMI medium.
In vivo tumor growth
Tumor cells were harvested, washed twice in PBS (Invitrogen Life Technologies) and resuspended on ice in growth factor-reduced matrigel (BD Biosceinces, San Jose, CA) at a concentration of 106 cells/ml. Mice were injected s.c. in the right flank with 105 cells in 100 μl matrigel and tumors were allowed to grow for up to 15 days.
In experiments where cell recruitment was studied, tumor-bearing mice were injected i.p. with a total of three injections of 2 mg 5-bromo-2’-deoxyuridine (BrdU; Sigma-Aldrich) starting at day 5 post-inoculation. The injections were given with 14 h intervals and mice were sacrificed 14 h following the last injection. In this setting, tasquinimod treatment was started 24 h before the first BrdU injection and continued until the end of the study. Seven mice were included in each group.
In experiments where tumor growth was studied, tasquinimod treatment was started at the day of tumor cell inoculation and continued either until day 7 post-inoculation or throughout the study. In some experiments, tasquinimod treatment was started at day 3 or 7 post-inoculation and continued until the end of the study. Tumors were measured with a caliper every second day starting on day 6–7 post-inoculation, when tumors were palpable. The tumor volume was calculated using the following formula: length x width2 × 0.4. At the end of each experiment, tumors and spleens were carefully excised and weighed. Six to ten mice were included in each group.
Antibody-mediated depletion
Gr1+ or Ly6G+ cells were depleted by i.p. injection of 500 μg anti-Gr1 (clone RB6-8C5) or anti-Ly6G (clone 1A8) antibody (BioXCell, West Lebanon, NH), respectively. Control mice were injected with the equal amount of an isotype control antibody (clone MPC-11) (BioXCell). In experiments where tumor growth was studied in conjunction with cell depletion, tumor cells were inoculated 24 h after antibody injection. Six to seven mice were included in each group.
Cell preparation
The dissected spleens were mashed in 70 μm cell strainers, which were washed with Hank’s balanced salt solution (HBSS) (Invitrogen Life Technologies). Tibias were crushed in a mortar and the recovered cells washed with HBSS. Tumors were cut into small pieces with a scalpel and treated with 2 mg/ml collagenase IV (Worthington, Lakewood, NJ) and 0.1 % DNase (Sigma-Aldrich) for 40 min at 37 °C. Following the enzymatic treatment, the pieces were mashed in 70 μm cell strainers. Cells were quantified using AccuCount beads (Spherotech, Lake Forest, IL).
Antibodies and flow cytometry
The following antibodies were purchased from Biolegend (Nordic Biosite, Täby, Sweden): B220-PerCP-Cy5.5 (RA3-6B2), c-kit-APC-Cy7 (2B8), CD3ε-PerCP-Cy5.5 (145-2C11), CD11b-Alexa700 (M1/70), CD11c-APC-Cy7 (N418), CD16/32-PE (93), CD45.2-PerCP-Cy5.5 (104), CD105-PE-Cy7 (MJ7/18), CD115-APC (AFS98), CD150-APC (TC15-12 F12.2), F4/80-PE-Cy7 (BM8), Ly6G-Brilliant Violet 421 (1A8), Sca1-PacificBlue (D7) and streptavidin-Brilliant Violet 605. The following antibodies were purchased from BD Biosciences: BrdU-FITC, CD19-PerCP-Cy5.5 (1D3), Ly6C-biotin (AL-21) and SiglecF-PE (E50-2440). Cells were stained with the above antibodies in FACS buffer (PBS supplemented with 5 % fetal calf serum and 0.05 % NaN3 (Sigma-Aldrich)). Fixable Viability Dye-eFluor506 purchased from eBioscience (Nordic Biosite) was used to detect dead cells. For BrdU staining, the FITC BrdU Flow Kit (BD Biosciences) was used according to the manufacturer’s protocol. Analysis of stained cells was performed using the LSRII flow cytometer (BD Biosciences).
Statistical analyses
All statistical analyses were performed using the Mann–Whitney U test.
Discussion
The focus of this study was to investigate the effects of the Q compound tasquinimod on myeloid cells during tumor development. In previous reports, we have evaluated the impact of the Q compound paquinimod on myeloid cells during complete Freund’s adjuvant (CFA)-induced inflammation and could show that the expansion of these cells in the spleen was reduced [
43,
50]. In particular, Ly6C
hi and SiglecF
+ cells were affected [
43]. In a model of necrotic cell-induced peritonitis, paquinimod also reduced the number of the same cell populations at the inflammatory site. This was not simply a result of compound toxicity, as it had no effect on these cells during steady-state conditions [
42]. Previous work by others has indicated a potential effect of Q compounds on cell recruitment. One such compound, linomide, was demonstrated to impair leukocyte-endothelium interactions in a rat model of TNF-α-induced hepatic injury [
45]. Another Q compound, laquinimod, was suggested to reduce the transmigration of lipopolysaccharide (LPS)-stimulated monocytes
in vitro [
46]. More recently, paquinimod was also shown to increase the rolling velocity of leukocytes on inflamed endothelium
in vivo [
54]. The exact mechanism of action of the Q compounds, and whether the target cells in our experiments are myeloid cells or endothelial cells, is still unknown. The human S100A9 protein was identified as one target molecule of paquinimod and this compound was shown to inhibit the binding of S100A9 to both of the pro-inflammatory receptors TLR4 and RAGE [
35]. In a previous study, we proposed that the S100A9-TLR4 interaction may promote tumor growth [
31], and as discussed therein and in more recent publications from our group [
33,
34], one mode of action of tasquinimod may be to interfere with that interaction. Both the myeloid and the endothelial cells could potentially be targets for such blockade, as they both express TLR4.
Taking these findings into consideration, and the previous knowledge that Q compounds are able to reduce the growth of various tumors [
28‐
34], we have here addressed the impact of tasquinimod on recruitment of myeloid cells to a transplantable tumor. Further, we have also investigated its impact on the accumulation of these cells in the spleen, as the spleen acts as an important reservoir of myeloid cells during tumor growth in certain tumor models [
14,
16,
23]. The reason why our attention was turned to myeloid cells was due to the fact that specific myeloid cell populations, Ly6C
hi cells in particular, have been implicated in promoting tumor development because of their immunosuppressive and pro-tumorigenic properties [
47,
48].
To analyze cell recruitment to tumors, we used an approach based on BrdU pulse labeling. It has been demonstrated not only in the 4 T1 tumor model [
16], but also in other spontaneous as well as transplantable tumor models [
11,
13,
14], that Ly6C
hi and Ly6G
hi cells within tumors are in a non-proliferative state and that their maintenance requires a constant input from external reservoirs [
21]. For this reason, a BrdU pulse of tumor-bearing mice is likely to result in labeling of these cells only in peripheral compartments such as the spleen. Indeed, a previous study that evaluated myeloid cell proliferation in various organs of 4 T1 tumor-bearing mice, identified the spleen as the main site for this event [
16]. The number of proliferating myeloid cells in the bone marrow, however, was very low and for this reason we focused our attention on BrdU
+ cells in the spleen. We show here that short-term treatment with tasquinimod reduced the number of BrdU
+ Ly6C
hi cells in the tumor, but did not affect BrdU
+ myeloid cell populations in the spleen. These two observations, together with the previously published results [
11,
13,
14], led us to conclude that tasquinimod interferes with recruitment of Ly6C
hi cells to the tumor rather than decreases the number of recruitable cells. Because of the differential effect of tasquinimod on Ly6C
hi and SiglecF
+ cells in the spleen and tumor in these experiments, we find it unlikely that tasquinimod would have a cytotoxic effect on these cells. A caveat of this approach, however, is that not all myeloid cell populations incorporate BrdU equally efficiently. Although a major fraction of Ly6C
hi cells and a minor fraction of Ly6G
hi cells within the tumor were BrdU
+ following the pulse, SiglecF
+ eosinophils displayed undetectable levels of BrdU incorporation. Thus, even though the proportion of eosinophils was reduced in tumors of the treated mice, we could not formally address whether the reduction was caused by reduced recruitment of these cells.
The finding that very few F4/80
+ cells within the tumors had detectable BrdU labeling is somewhat surprising, considering that previous reports have suggested that Ly6C
hi cells within tumors have the potential to differentiate to various types of TAM [
11]. Furthermore, it has been suggested that differentiated TAM can proliferate [
12,
17,
18]. Our data would indicate that the TAM identified in 4 T1 tumors grown for 7 days do not proliferate extensively and that very few BrdU-labeled Ly6C
hi cells differentiate to TAM within the time frame of the study.
To assess whether tasquinimod-mediated reduction in tumor-infiltrating Ly6Chi cells could in part underlie its anti-tumor effects, we decided to temporarily deplete these cells using a specific antibody. Using this approach, we confirmed that these cells promote 4 T1 tumor growth, thereby indicating that reduction of these cells in tumors is also one anti-tumor effect of tasquinimod. Additionally, these experiments revealed that the absence of these cells in the spleen, bone marrow and tumor during the first 3 days of tumor development was sufficient to reduce tumor growth. Similarly, tasquinimod treatment for 7 days was also sufficient to reduce tumor growth. However, treatment throughout all 14 days of tumor growth did not result in an additional anti-tumor effect despite the fact that the frequency of Ly6Chi cells within the tumors normalized once the treatment was terminated at day 7. Taken together, these observations indicate that the anti-tumor effect of tasquinimod is important during an early phase of tumor development.
In contrast, antibody-mediated depletion of Ly6Ghi cells did not significantly influence growth of the primary tumor in our experiments, suggesting that in this tumor model, these cells may be less important for initiation of tumor cell growth than Ly6Chi cells. Further, long-term treatment of tumor-bearing mice with tasquinimod did not affect the frequency of splenic Ly6Ghi cells, but reduced the frequency of Ly6Chi and SiglecF+ cells. Thus, tasquinimod treatment targeted the same myeloid cell populations in tumor and spleen. At present we do not know whether this observation means that the compound targets a common mechanism or independent mechanisms at these two sites. We speculate that vascular extravasation of cells in the tumor and retention of cells in spleen might both involve adhesion to endothelium and that tasquinimod could potentially interfere with certain cell-endothelial interactions at these two sites.
As previous reports have demonstrated that Ly6C
hi cells contribute to immune suppression, angiogenesis and metastasis in tumor-bearing mice [
47,
48], it seems reasonable to assume that the reduced recruitment of Ly6C
hi cells during early tumor development results in an environment less hospitable for tumor growth. Indeed, tasquinimod is known to affect the tumor microenvironment in several ways [
55]. Thus, tasquinimod possesses anti-angiogenic properties [
28,
29,
34] and, interestingly, the tumor vasculature was significantly reduced following 1 week of treatment [
34]. The anti-angiogenic effects of tasquinimod correlated with a skewing of the functional phenotype of TAM from CD206
+ MHCII
low M2 macrophages to CD206
− MHCII
hi M1 macrophages [
33,
34]. It is well established that M2 macrophages promote angiogenesis [
56,
57]. Recent papers also demonstrated that tasquinimod treatment altered the immunosuppressive properties of CD11b
+ cells within tumors [
33,
34], which in turn may be related to the switch from M2 to M1 macrophages. We speculate that the reduced recruitment of Ly6C
hi cells (and potentially of eosinophils) to the tumor might disrupt the balance between M2 and M1 macrophages. Possibly, the M2 phenotype of TAM can only be maintained provided that newly arrived Ly6C
hi cells can continuously be polarized to this phenotype and the M1 macrophages may dominate functionally once replenishment of the M2 population is reduced. These factors might impact on the initiation of tumor cell growth or seeding, resulting in reduced growth of the tumor itself. Also, when plotting the 4 T1 and B16 tumor growth curves using a logarithmic scale, tasquinimod treatment did not influence the slope of the growth curves (not shown). This further supports that the treatment may affect the initiation of tumor cell growth or seeding rather than tumor cell proliferation
per se. In contrast, tasquinimod treatment started at day 7, when tumors are well established, did not result in reduced tumor growth despite a reduced frequency of Ly6C
hi cells within the tumors. It might be argued, however, that due to the aggressive nature of 4 T1 tumors, treatment started at this stage is futile. When treatment instead was started at day 3, the anti-tumor effect of tasquinimod was maintained, indicating that there is a critical time frame between days 3–7 within which the anti-tumor effect of tasquinimod is lost.
While 4 T1 tumors in their early phase of development induced an expansion of myeloid cells in the spleen, there was no detectable splenomegaly. At a later phase, however, when tumors were more developed, the spleens were significantly enlarged and contained increased numbers of Lin
− c-kit
+ Sca1
− hematopoietic precursor cells. This is likely the result of an increased production of myelogenic cytokines by tumor and stromal cells as well as infiltrating immune cells [
8,
47]. Indeed, one previous study showed that the accumulation of Ly6G
hi cells in 4 T1 tumor-bearing mice is driven by tumor-produced G-CSF [
20]. Recently, it was also demonstrated that G-CSF treatment could mimic the effects of the 4 T1 tumor on the spleen, as it induced splenomegaly and an increased erythropoiesis in this compartment [
53]. We detected an increased frequency of cells with the Lin
− c-kit
+ Sca1
+ phenotype in spleens from both tasquinimod-treated mice and anti-Gr1-treated mice, as well as in the bone marrow of tasquinimod-treated mice. At present we do not know the nature of these cells, but based on their phenotype, they may contain hematopoietic precursors [
58]. Since both these treatment regimes involve reduction of myeloid cells, we speculate that the increased frequency of these cells might be the result of some compensatory mechanism.
We further noted that the numbers of the various myeloerythroid progenitors that we analyzed in the spleen were all increased in tumor-bearing mice. In addition, the composition of the Lin− c-kit+ Sca1− population was altered in these mice, such that the frequencies of Pre MegE and Pre CFU-E cells increased at the expense of GMP and Pre GM cells. Tasquinimod treatment restored the composition of this cell population to a naïve-like state. This observation rules out the possibility that the reduction in the frequency of Ly6Chi and SiglecF+ cells in the spleen following long-term tasquinimod treatment would be due to reduction of the number of GMP or Pre GM cells. We detected similar effects of tasquinimod on splenic myeloid cells and hematopoietic precursor cell populations when treatment of tumor-bearing mice was initiated 7 days post-inoculation, despite the lack of anti-tumor effect. Thus, the effects of tasquinimod on splenic myeloid cells and their precursors are not a result of reduced tumor burden. However, while these effects of the compound on splenic cells did not reduce the growth of already established rapidly growing tumors, it remains possible that such effects might have an impact on other, more slowly developing tumors.
In the bone marrow, tumor growth changed the composition of the Lin
− c-kit
+ Sca1
− population such that GMP and Pre GM cells were increased at the expense of Pre MegE and Pre CFU-E cells. The absolute numbers of these cell populations, however, were not altered which correlates with previous findings that have proposed the spleen as the main site of progenitor cell expansion during 4 T1 tumor growth [
16]. The effects of tasquinimod in this compartment differed from the spleen such that treatment did not completely normalize the composition of the Lin
− c-kit
+ Sca-1
− population, but rather decreased the frequency of Pre GM cells. Further studies are required to elucidate the compartment-specific effects of tasquinimod. Recently, a subset of F4/80
hi VCAM-1
+ CD169
+ macrophages was demonstrated to support splenic myelopoiesis as well as erythropoiesis, similar to what has previously been shown for the bone marrow [
53,
59,
60]. However, tasquinimod displayed no effect on number or frequency of this particular subset of macrophages in the spleen. Thus, the effects of the compound on the myeloerythroid precursors cannot be explained by an impact on numbers of these cells, but it remains possible that it may impact on their function.