Modulation of anti-tumor immunity by the brain’s reward system

Regulating immunity is a leading target for cancer therapy. Here, we show that the anti-tumor immune response can be modulated by the brain’s reward system, a key circuitry in emotional processes. Activation of the reward system in tumor-bearing mice (Lewis lung carcinoma (LLC) and B16 melanoma) using chemogenetics (DREADDs), resulted in reduced tumor weight. This effect was mediated via the sympathetic nervous system (SNS), manifested by an attenuated noradrenergic input to a major immunological site, the bone marrow. Myeloid derived suppressor cells (MDSCs), which develop in the bone marrow, became less immunosuppressive following reward system activation. By depleting or adoptively transferring the MDSCs, we demonstrated that these cells are both necessary and sufficient to mediate reward system effects on tumor growth. Given the central role of the reward system in positive emotions, these findings introduce a physiological mechanism whereby the patient’s psychological state can impact anti-tumor immunity and cancer progression.

(1) It would be helpful in the evaluation of their findings to know whether VTA activation influences the efficiency of cellular adoptive transfer. For instance, does their treatment conditions promote a more robust peritoneal immune response that results in an initial rejection of tumor cells and/or MDSCs immediately following injection?
(2) Control images and quantification are needed to validate equal neuronal infection between control and VTA virus injections in Figure 1B-C and Supplementary Figures 1-2. (3) Control groups should be more extensively described in all of the figure legends. For instance, are the control groups AAV8-hSyn-DIO-mCherry infection plus CNO treatment? Or just one or the other? This should be stated in each of the figure legends.
(4) Confirmation is needed to validate that DREADD-induced activation of cells in the VTA is specifically responsible for orchestrating the observed changes in anti-tumor immune responses. It could be that TH+ cells in other regions of the brain are also infected. To demonstrate that the reported effects are due to DREADD-mediated activation only in the VTA, the authors should additionally evaluate mCherry staining in other brain regions that have also been proposed to wire the brain's reward system, including the nucleus accumbens, lateral hypothalamus frontal cortex, etc. (5) Utilization of 6OHDA to ablate the SNS could potentially have off-target effects that may influence the interpretation of their data. Therefore, a secondary approach to ablate SNS function would help to strengthen their conclusion that the SNS is mechanistically involved in mediating the effects of VTA activation on tumor eradication. Furthermore, the authors should also confirm that 6OHDA does not cross the BBB in their model. They cite that 6OHDA does not normally cross the BBB under homeostatic conditions. However, they are now introducing a number of variables into their system including tumor-induced inflammation and CNO treatment, both of which could potentially influence the integrity of the BBB separately or in tandem. (6) The authors rely on Gr-1+CD11b+ staining to identify MDSCs in their studies. The possibility exists that some of their differences that they ascribe to MDSCs in VTA-activated vs. control mice are actually due to the effects of VTA stimulation on neutrophil responses.
Gr-1 stains both Ly6C and Ly6G, thus their gating strategy could include both neutrophils (CD11b+Ly6G+) and MDSCs (CD11b+Ly6C/G+). (7) The induction of p-CREB MFI levels following beta-agonist treatment is not all that striking. Representative histogram plots should be provided to accompany the MFI data plotted in Figure 3C to properly evaluate these differences. (8) The technical aspects of their MDSC:T cell suppression experiments are not adequately described in the methods, figure legend, or actual text. It is unclear if they are measuring percentage of total T cells that have proliferated, accumulation of proliferating T cells, or something else. Moreover, with in vitro suppression assays it is usually good practice to provide a dose curve with decreasing ratios of the suppressive cell population to account for culture related issues and other variables. (9) Only evaluating CD69 expression by tumor-infiltrating T cells is not sufficient to report an effect of VTA activation on in vivo anti-tumor T cell responses. CD69 does not really indicate much in terms of anti-tumor effector T cell function as it can be up regulated by antigen exposure or type I IFNs. To corroborate that VTA activation attenuates the immunosuppressive activities of MDSCs on in vivo anti-tumor T cell responses, the authors should evaluate T cell-mediated cytokine production and/or cytolytic function. (10) The results presented in Figure 4 are very compelling. However, these findings also raise a number of important questions about MDSC plasticity and longevity. For instance, it is unclear if the MDSCs are completely reprogrammed by the VTA activation and are no longer able to revert back to a pro-tumor phenotype. To have such a profound effect over 14 days would suggest that MDSCs survive a long period of time in vivo and do not turnover. How long do the adoptively transferred MDSCs survive in vivo? It is unclear if long-term survival of MDSCs is required for their effect or if only a short period of time is needed for them to influence the anti-tumor T cell response. It could also be that they are transferring over more neutrophils in the VTA-activated group (see above comment # 6). Or conversely, it could be that VTA-activated MDSCs do not survive as well following adoptive transfer. Understanding what contributes to this effect would be valuable to know in order to better appreciate what biologically underlies these results. Moreover, it would also be important to know whether these differences in tumor load are associated with altered antitumor T cell responses or if this occurs independently.
Minor comments (1) It is unclear how NA levels were measured in Figure 2? qRT-PCR?

Reviewer #2 (Remarks to the Author):
In this manuscript, Ben-Shaanan et al expand on their interesting observation (published in Nat. Med) that activation of the reward system can lead to the stimulation of systemic immunity. Here, they demonstrate that DREADD expression in VTA can abrogate the growth of two subcutaneous tumor models (LLC lung cancer and B16 melanoma). Finding reported here are interesting and consistent with their previous observations. However, several issues have to be addressed before the manuscript is accepted for publication. The exact mechanism by which VTA activation inhibited tumor growth still remains unclear.
Although, they suggest modulation of the bone marrow MDSCs by sympathetic nervous system (SNS) to be important, other potential mechanisms (such as activation of other innate immune cells such as NK cells) have not been explored. Furthermore, other microenvironmental changes (such as angiogenesis) were not examined. Additional mechanistic studies (highlighted below) would have strengthen this paper.z Fig. 1: Durability of VTA activation (as measured by c-Fos exression) is only presented at 14 days post tumor implantation. Because some experiments extended to 21 and 28 days ( Fig.  1g and 1h), persistence of VTA activation should be confirmed in these groups as well. Additionally, have they tested the impact of DREADD expression in other CNS locations besides VTA? Could their observations be due to generalized CNS DREADD expression or is it unique to VTA. Fig. 2: Tumors in 6OHDA-treated mice (irrespective of VTA activation) appear to be twice as large as the untreated mice (Fig. 2b  3h) yet NA levels were not changed (Fig. 1e). This observation appears to be inconsistent with the proposed mechanism of SNS inactivation of MDSCs? Have the authors examined the activity of NK cells (or other CD3 cells) in the spleen or in the tumors in their model? Characterization of MDSC activity also appears to be superficial. Additional techniques, such as flow (to examine the MDSC or TAM phenotype markers like Arg, IDO, etc) or RNA-seq (nanotstring) would have provided more comprehensive information about MDSC polarization in VTA-activated mice. Fig. 4: Have the authors performed depletion studies (Gr1 Ab) to confirm the role of MDSCs in this model?

Reviewer #3 (Remarks to the Author):
This is a very exciting set of experimental results regarding the CNS pathways that regulate the effects of peripheral neurobiology on tumor growth. The core model supported by these studies is that VTA activation suppresses peripheral SNS signaling (particularly in bone marrow) and thereby reduces the production of MDSCs that would otherwise subsequently inhibit cellular immune responses against cancer. The authors present a promising set of studies to support this hypothesis, but several additional highly feasible studies would be required to more definitively support it. Please consider the remarks below as encouraging suggestions to enhance the impact of this already promising work.

Abstract:
Needs to include a description of tumor model/s used as a read-out (i.e., LLC and B16).
Abstract lacked any allusion to studies blocking MDSCs in the VTA-activation model in order to fully establish causation. Mimicry of VTA-activation effects in naive animals by adoptive transfer of MDSCs from VTA-activated mice shows that MDSCs are sufficient to facilitate tumor growth, but does not establish whether they are necessary. For that MDSC inhibition (e.g., by inhibiting myeloid cell growth differentiation in general, or MDSC differentiation in particular) is required.

Introduction:
First 5 refs listed are not particularly comprehensive, and some are controversial/discredited. Perhaps cite more comprehensive review articles (e.g., 2 from Paige Green and colleagues in Nature Reviews Cancer and/or Andrew Steptoe et al in Nat Clin Pract Onc)?
It is not accurate to claim that mechanisms of psychobiological effects on cancer remain "unknown"; quite a lot is now known about peripheral neural, cellular, and molecular mechanisms of such effects (see the 2 review articles referenced above in Nat Rev Cancer). More accurate would be to claim that little is known about the CNS mechanisms involved, where this paper makes a very significant contribution (but see also Cao et al, Cell. 2010 Jul 9;142(1):52-64) Need to clarify the nature of LLC cells when initially referenced in the text (what kind of cells are these, and how was their identity verified), along with the locale of sc injection.
6OHDA is not a clean abrogation of SNS (it induces massive catecholamine release prior to nerve fiber death). Results would be more persuasive if data show that pharmacologic abrogation of beta-adrenergic signaling (preferably with an agent that does not cross BBB, e.g., nadolol) also blocks effects of VTA activation on tumor growth.
The experiment in Fig 2 (6OHDA) lacks a positive control. Needs to be repeated in parallel with a sham SNS intervention (e.g., saline) that continues to show VTA activation of tumor growth in the same experiment as it is abrogated by 6OHDA (i.e., a 2 x 2 design instead of the current 2 condition experiment that attempts to compare with other separate experiments).
"Stress responses" can occur in the bone marrow, so it is inaccurate to equate stress with plasma catecholamine levels (as the ms appears to do now). More accurate to simply state no VTA activation-induced difference observed in plasma but significant difference observed in bone marrow.
SNS directly innervates some tumors, but certainly not all. Unless authors directly document SNS innervation of all/most tumors in the particular model examined here (e.g., by histological detection of nerve fibers in tumor tissue as in Fig 2h for bone marrow), it would be more accurate to say SNS innervates some tumors, or that SNS can innervate tumors, or something similarly less general. The intratumoral catecholamines assessed here may come predominately from blood or perivascular nerve fibers rather than from true innervation of the tumor parenchyma per se.

The results in Fig 3 are interpreted as showing that VTA activation suppresses SNS output
to the bone marrow and thereby inhibits the supressive effect of MDSCs on T cell-mediated responses against cancer. However, the final step is not directly tested. One simple way to do that would be to conduct the VTA activation protocol in SCID or nude mice that lack functional T cells (in parallel with their parental strains bearing functional T cells).
Upregulated TNF production is implied to mediate VTA/SNS effects on T cell proliferation/activation, but that too is not directly demonstrated. Experiments with TNF knockout mice or anti-TNF neutralizing antibody would support that claim.
According to the experimental schematic in Much is known about how SNS activation in bone marrow enhances myeloid differentiation (e.g., from studies by P Frenette et al., J Sheridan's group and particularly a study by Powell et al, and the group of Nahrendorf & Swirski). That material should be cited and could help design experiments identifying the molecular mechanism by which VTA-induced SNS downregulation inhibits MDSC development/distribution. One approach might involve analysis of the myeloid growth factors already demonstrated to be upregulated by betaadrenergic signaling (e.g., as in Sheridan's and Frenette's studies).
The Discussion needs to acknowledge some limitations on the scope of conclusions that can be drawn from these studies. E.g., LLC and B16 are poor models of tumor initiation or metastasis; they assess predominately primary tumor growth rates in already-established cancer cells. There is no verification of relevance to human cancer, or to tumor growth in orthotopic tissue settings. Tumor injections into blood sc are not broadly representative of normal conditions where tumors initiate in solid tissues. The number of cell lines examined is limited, and the present models may overestimate the breadth of effects on other tumor types that are not highly immunogenic or lung localized. All that said, the authors are to be complimented for their careful interpretation of these studies as a test of basic physiological relationships among the nervous, immune, and tumor systems, as opposed to claiming a mechanism for the controversial clinical relationships between positive mood and cancer progression.
We would like to thank the referees for their time and attention in reviewing our manuscript entitled "Modulation 4 of anti-tumor immunity by the brain's reward system".

5
In this manuscript, we provided the first demonstration that reward system activity can alter the anti-tumor 6 immune response. Thanks to the reviewers' comments, we clarified the manuscript and performed several 7 additional key experiments that substantiated our findings and expanded our understanding of the mechanisms 8 underlying this effect. Below is a summary of these main new experiments: 9 1. To demonstrate the involvement of the sympathetic nervous system in mediating reward system effects, we 10 originally used 6OHDA. We now added another experiment using the b-adrenergic blocker, Nadolol (Fig 2b).

11
This manipulation, similarly to the effect of 6OHDA, eliminated the beneficial effects of the reward system on 12 tumor growth. 13 2. We originally applied an adoptive transfer experiment to demonstrate that MDSCs were sufficient to mediate 14 the VTA-induced attenuation in tumor growth. We now added an experiment using anti-Gr-1 antibody to deplete Granzyme B (Fig 4a). Moreover, this effect appears to depend on MDSCs, because depletion of MDSCs 23 eliminated the effect of VTA activation on Granzyme B levels.

24
In addition, as detailed in the point by point response, we address all the specific concerns raised by the  (   Granzyme B levels by CD8 T cells (Fig 4a).

85
4) The reviewer correctly indicates that there is no reliable tool to differentiate between neutrophils and PMNto address this issue, at least in part, we now provide evidence that VTA activation did not affect the relative Figures 1-2.

97
This should be stated in each of the figure legends.

98
Thank you for highlighting this omission. We added full descriptions of the control groups in all figure legends.

staining in other brain regions that have also been proposed to wire the brain's reward system, including
105 the nucleus accumbens, lateral hypothalamus frontal cortex, etc.

106
As suggested by the reviewer, we now provide representative immunofluorescence images of the nucleus 107 accumbens, lateral hypothalamus, and frontal cortex from the brains of mice injected with a virus (Fig. S2). There 108 was no mCherry expression in these brain regions, demonstrating that DREADD expression was restricted to 109 the VTA.

119
This is an important question. Therefore, as a complementary approach, we used Nadolol, a b-adrenergic 120 antagonist that does not cross the BBB, an approach suggested by Reviewer #3 to consolidate the involvement 121 of the SNS in mediating the beneficial effects of VTA activation on tumor weight. In analogy to the effect of 122 6OHDA, Nadolol treatment eliminated the effects of VTA activation on tumor growth (Fig. 2b). These findings 123 further support the mechanistic involvement of the sympathetic nervous system in VTA-induced tumor 124 suppression.

125
Moreover, to confirm that catecholaminergic neurons in the brain were not affected by the 6OHDA treatment, we 126 demonstrate that there was no significant difference in the number of TH + cells in the VTA of tumor bearing mice and controls in the relative abundance of PMN-MDSCs identified by their expression of CD11b and Ly6G versus Figure 3C to 143 properly evaluate these differences.

144
As suggested by the reviewer, we added a representative histogram showing the change the in p-CREB levels 145 following b-agonist treatment (Fig S6). Moreover, we agree with the reviewer's comment that changes in p-CREB

146
MFI are not striking. One factor that can explain this phenomenon, is that the cells used for the analysis were 147 isolated from the bone marrow, a niche with relatively high 148 NA levels (Fig S5). To substantiate this claim, we provide

162
We have now expanded the description of the experiment in Figure legend 3i, 3k and in the Methods (page 25).

163
As advised by the reviewer, we now provide the relevant controls in Fig S11,

166
Regarding the use of a dose response: Although a dose curve was applied while calibrating the initial 167 experimental protocol, due to the limited availability of the MDSCs from VTA activated mice and controls, we had 168 to limit the experimental design to a single dose.

176
The reviewer correctly states that changes in CD69 expression do not necessarily represent a functional change 177 in an effector cell. To directly address the reviewer's comment, we further characterized tumor and spleen CD4 178 and CD8 T cell abundance, IFN-γ and TNF-a expression, as well as Granzyme B expression by tumor CD8 T 179 cells. We did not observe any difference in the abundance of the T cells or their TNF-a/IFN-γ levels ( Fig S12).
anti-tumor immune response.

(1) It is unclear how NA levels were measured in Figure 2? qRT-PCR?
Levels of NA were measured by ELISA and validated by immunohistochemistry. We now clarify this point in the legend of Figure 2.

205
We would like to thank the reviewer for the helpful comments and suggestions.

: Durability of VTA activation (as measured by c-Fos expression) is only presented at 14 days
208 post tumor implantation. Because some experiments extended to 21 and 28 days (Fig. 1g and 1h

212
To address the reviewer's comment, we now provide c-Fos calibration on day 28. Our results demonstrate that 213 DREADD manipulation was also effective at this time point (Fig S3; p<0.012).

214
The reviewer also correctly suggests that activity of other brain regions may also affect tumor growth. In fact, a 215 major focus of my lab is to characterize the peripheral immune changes induced by activity of different brain 216 regions, as they serve different roles in an organism's behavior and physiology. However, as each of these brain 217 areas requires extensive characterization, we consider such a survey to be beyond the scope of the current 218 study.

219
A third point raised by the reviewer is that the observed effects could be due to generalized CNS DREADD 220 expression. To address this comment, we provide in Figure S2, an immunohistochemical analysis of mCherry 221 labeling (a marker of DREADD expression in our system) to demonstrate that there is no neuronal expression of 222 mCherry beyond the VTA dopaminergic neurons.

Fig. 2: Tumors in 6OHDA-treated mice (irrespective of VTA activation) appear to be twice as large as
225 the untreated mice (Fig. 2b vs. 1f)

228
As the reviewer correctly points out, the tumors in the 6OHDA-treated mice are larger than in un-treated controls.

229
This is in line with the literature indicating that sympathetic activity plays a role in tumor growth (e.g. L. Horvathova et al, 2016 andSW. Cole et al, 2015). In this study, we used 6OHDA to eliminate peripheral sympathetic activity 231 in order to determine whether the SNS is required to mediate the signal between the brain's reward system to revised version, we took an additional approach to validate SNS involvement in mediating the effects of the VTA, which was suggested by Reviewer #3. Instead of 6OHDA, we treated VTA activated mice and their controls with highly feasible studies would be required to more definitively support it. Please consider the remarks below as encouraging suggestions to enhance the impact of this already promising work.

282
We would like to thank the reviewer for the careful evaluation of the study and the helpful comments.

304
Thank you for pointing out these issues with the papers that we cited. We now replaced the papers that we 305 initially referenced, with those suggested by the reviewer.

318
This was added to the text as part of a more comprehensive description of the tumor model, both in the main text

signaling (preferably with an agent that does not cross BBB, e.g., nadolol) also blocks effects of VTA activation on tumor growth.
demonstrate that VTA activation indeed requires sympathetic activity, specifically, b-adrenergic signaling (Fig  experiment as it is abrogated by 6OHDA (i.e., a 2 x 2  other processes which are MDSCs independent are also involved. We now discuss this point on page 6.

is sufficient to mimic effects observed in Fig 1. Several questions arise, including whether similar effects occur with adoptive transfer M-MDSCs?, whether comprehensive numerical measures of tumor size are
reasons. It is extremely difficult to purify a sufficient number of M-MDSCs to achieve effective transfer. Therefore,

379
for the transfer experiment, we isolated the MDSCs based on their expression of Gr-1 and CD11b. We now clarify 380 this issue on page 6. Nevertheless, we could partially address the question of MDSC subtypes, by demonstrating 381 that VTA-activation did not alter the relative proportion of PMN-MDSCs and M-MDSCs used for the transfer (Fig.   382 S7).

383
The additional numerical measurement of the tumor size, which were suggested by the reviewer, are now 384 provided in Fig. 4f.

385
Finally, to determine whether MDSCs inhibition in VTA-activated or control mice affected tumor growth, we

396
The reviewer is correct that understanding the underlining molecular mechanism will be beneficial. However,

397
following consultation with the Editor, we consider the specific molecular characterization will be beyond the 398 scope of the present study. We discuss this limitation on page 7 and added the proposed citations.

411
We agree with the reviewer that the present study serves mainly as an initial indication of the potential 412 of brain activity, and specifically the reward system, to affect tumor growth. Accordingly, we now expanded the 413 description of the relevant limitations of our results and modified the text to state: "It is important to note that we 414 do not consider such robust and specific modulations of VTA activity (as induced here by DREADD activation)

415
to be physiological and the relevance to human cancer is still unknown. Moreover, in this study, we used two growth." 421 422