In this manuscript, we showed that the anti-tumor effectiveness of hu14.18K322A in vivo is enhanced by αCD40+ CpG therapy. The benefit of combining hu14.18K322A with αCD40+ CpG was particularly apparent in mice with small tumors. Notably, therapeutic efficacy depended more on tumor amount than time of establishment (), indicating this regimen may be effective in minimal residual disease. Combined therapy (with hu14.18K322A added to αCD40+ CpG), was effective at eradicating small tumors and led to long-term survival in some animals; this was not seen in the experiments when mice received only hu14.18K322A or only αCD40+ CpG. Long-term survival occurred in immunocompetent mice without inducing immunological memory, and also could be induced by combined therapy in mice lacking T cells.
Secondary induction of adaptive immunity after monoclonal antibody-based therapy has been observed in various tumor models [24
], but the mechanism and the dominant effector cell population depends on the tumor type. Recently, Abes et al observed the induction of a persistent, T-cell dependent, anti-tumor response elicited by αCD20 mAb, CAT-13 [24
]. However, T cell involvement is likely tumor dependent as our group has previously shown that immunocytokine therapy with anti-EpCAM-IL-2 can induce a T cell response against tumor with high MHC class I expression, but not against low MHC class I variants of the same tumor [27
]. In the latter case with poor MHC expression, NK cells are the dominant effector cell that responds to immunocytokine and persistent T cell immunity is not achieved [27
]. In this manuscript, we are using B78D14 tumor cells, a subclone of B16-F10. B16-F10 is well documented to have low expression of MHC class I, which may be one reason we did not observe a T cell response in this model.
We investigated the potential role of NK cells and macrophages, as both are capable of mediating ADCC. We have previously reported that αCD40 can activate both NK cells and macrophages in tumor bearing mice [28
]. Our data presented here indicate that NK cells are required for therapeutic efficacy in vivo
(); however, studies using SCID/beige mice suggested that lytic function of NK cells might not be a requirement () in vivo
. Recently, a model of allogeneic cell rejection was used to demonstrate that macrophage effector functions cannot be well deciphered in vivo unless NK cell function is impaired [29
]. Previous studies from our laboratory [9
] and others [30
] have shown that interferon-γ (IFNγ) is required for macrophage activation in certain mouse models of tumor immunotherapy. NK cells are the primary source of IFNγ after in vivo administration of CpG + antibody-coated tumor cells [31
]. Furthermore, co-administration of interleukin-12 alongside Trastuzumab therapy produces greater antitumor responses and is dependent on NK cell production of IFNγ[32
]. Both anti-CD40 [10
] and CpG 1826 [33
] promote secretion of IFNγ when administered independently. Therefore, while it may be difficult to separate the independent contributions of macrophages and NK cells in vivo
, NK cell production of IFNγ in SCID/beige mice that would be absent in NK cell depleted mice may explain the difference in results between these two settings ( and ).
Enhancement of ADCC by CpG-activation of effector cells was initially shown in the late 1990’s in a mouse model of B cell lymphoma [34
]. Much of this was assumed to be the direct activation NK cells, which has been described for both human and mouse NK cells [35
]. Subsequently, two classes of CpG-containing oligoneucleotides, class-A and -B, were described to activate different effector populations involved in ADCC [36
]. In vivo depletion studies have been used to show that while CpG-A-enhanced ADCC requires only NK cells, CpG-B-enhanced ADCC is mediated by both NK cell and granulocytes [36
]. CpG therapy was combined with rituximab in a phase II trial in patients with follicular lymphoma, in which responses were similar to those previously reported for rituximab alone [37
]. However, the patients in the combination trial had all undergone prior rituximab treatment, and therefore comparison of the two reported response rates may be unfair [37
]. The route of CpG administration may affect its ability to enhance ADCC. Intratumoral CpG rather than systemic CpG was shown to be better at increasing the anti-tumor activity of an αCD20 antibody [38
]. In our model system, we have observed that priming with αCD40 increases the expression of TLR9 by macrophages peaking at three days after αCD40. Therefore, we used this information to devise our treatment regimen. We believe that in addition to activating macrophages and other cells of the innate immune system, αCD40 is sensitizing cells to TLR9 activation resulting in an augmented response to CpG.
Studies depleting macrophages in vivo
with clodronate liposomes [12
] were inconclusive in our hands; while the anti-tumor effect of the combined therapy was substantially reduced by treatment with the clodronate liposomes, treatment of tumor bearing mice with control, PBS-containing, liposomes also slightly but significantly reduced the anti-tumor effect of therapy. However, a recent publication by Tsai and colleagues suggested that phagocytosis of gold particles larger than 4nm in size reduced the cellular response to TLR9 [39
]. Tsai et al. concluded that the effect was due to a gold-specific interference of high mobility groupbox 1 that regulates TLR9 signaling. As they did not compare gold nanoparticles to a non-gold control, they could not conclusively say if the inhibition of TLR9 responses was restricted to gold particles, or possibly a function of phagocytosis of large particles [39
]. Since we observed a decrease in therapeutic efficacy when mice were treated with PBS-liposomes, we were concerned that the ingestion of liposome particles may have decreased the functional response to TLR9 stimulation. Therefore, we used an in vitro system to evaluate macrophage-mediated anti-tumor activity.
αCD40 -primed macrophages can act as effector cells against numerous mouse and human tumors [7
]. αCD40 activated human macrophages mediate anti-tumor responses to human pancreatic cancer [43
]. Few reports have evaluated CD40 and/or CpG activation of macrophages for increasing antibody responses. The data presented here demonstrate at least a partial role for αCD40+ CPG primed macrophages in the response to tumor-targeted hu14.18K322A.
Using a co-culture system, we found that enriched αCD40+ CpG primed macrophages could inhibit tumor cell proliferation in an antibody-dependent manner (). These in vitro data suggest that macrophages are involved in the anti-tumor response. Clinical data from our laboratory have recently suggested that myeloid cells may play a role in the therapeutic effect of mAb in neuroblastoma patients [20
]. In a small phase II study of hu14.18-IL2 immunocytokine, we observed a clear association between the expression of a high affinity Fc receptor on myeloid cells (CD32) and clinical response [20
]. Furthermore, we participated in recently demonstrating that αGD2 mAb ch14.18, when combined with GM-CSF and interleukin-2 (IL2) is better than the standard of care for high-risk neuroblastoma patients [44
]. Although not compared directly, ch14.18 + GMCSF + IL2 appeared to have a bigger impact on survival than ch14.18-IL2 mAb alone, without GMCSF [45
]. While more data are needed, these two studies may suggest an important role for myeloid cells in the response to αGD2 mAbs in patients with neuroblastoma.