Appendices Contents Appendix 1: Calculations of proliferation and estimates of size B Appendix 2: Immune profiling gating strategy . E Appendix 3: Table of chemokine/cytokine expression F Appendix 4: Mitf expression in LN G Appendix 5: Alternate inhibitors of HGFR, EGFR and TGF-βR1 . H Appendix 6: Dct expression in lungs of CD8-depleted mice at seven weeks . I Appendix 7: Dct expression in lungs of tumor-bearing MMP9-KO mice . J A Appendix 1: Calculations of proliferation and estimates of size Effect of reduced mitotic index on tumor diameter: At early time points, tumor growth is assumed to be exponential and described by the following equation (1): (1) 𝑘𝑘𝑘𝑘 𝑁𝑁 = 𝑁𝑁0 𝑒𝑒 ( 𝜏𝜏 ) Where N is the number of melanoma cells at the end of the experiment, N is the number of melanoma cells at the onset of the experiment, k is the fraction of proliferating cells, t is the duration of the experiment (6 or 19 weeks), τ is the duration of the cell cycle (28h, according to Gordon 1980, Canc Res 40, 4467-4472). The ratio N c /N d between the tumor mass in control vs depleted mice is given by: (2) 𝑁𝑁𝑐𝑐 [𝑘𝑘 𝑐𝑐 −𝑘𝑘 𝑑𝑑 ]𝑡𝑡 ) 𝜏𝜏 �𝑁𝑁 = 𝑒𝑒 ( 𝑑𝑑 Where k c is the fraction of proliferating cells in the tumor of control mice and k d is the fraction of proliferating cells in the tumor of PMN-MDSC-depleted mice. For tumors with ellipsoid shape, the longer diameter is proportional to the cubic root of the cell number. Therefore the ratio between the diameters of tumors from control and depleted mice is given by: (3) 𝐷𝐷𝑐𝑐 [𝑘𝑘 𝑐𝑐 −𝑘𝑘 𝑑𝑑 ]𝑡𝑡 ) 𝜏𝜏 �𝐷𝐷 = {𝑒𝑒 ( 𝑑𝑑 } k c and k d are deduced from MI, the percentage Ki67+ cells, and S, the duration of Ki67 expression during cell cycle (estimated to be 14h based on Lopez 1991 Cytometry 12:42-49) according to: B (4) 𝑘𝑘 = 𝑀𝑀𝑀𝑀 ( 𝜏𝜏�𝑆𝑆 ) From (3) and (4), one gets: (5) 𝐷𝐷𝑐𝑐 [𝑀𝑀𝑀𝑀 𝑐𝑐 −𝑀𝑀𝑀𝑀 𝑑𝑑 ] 𝑡𝑡 ) 𝑆𝑆 �𝐷𝐷 = {𝑒𝑒 ( 𝑑𝑑 } We measured MI c =0.035 and MI d =0.017. At seven weeks of age, the expected D c /D d was 1.7 and no significant difference could be detected experimentally. At 20 weeks of age, the expected D c /D d was 5, in good agreement with the 4.7 value measured experimentally. Decreased ectopic expression of Dct is not due to reduced cancer cell proliferation At early time points, the population of melanoma cells in the lung can be described by the following equation (1): (1) 𝑘𝑘𝑘𝑘 𝑁𝑁 = 𝑁𝑁0 𝑒𝑒 ( 𝜏𝜏 ) Where N is the number of melanoma cells at the end of the experiment, N is the number of melanoma cells at the onset of the experiment, k is the fraction of proliferating cells, t is the duration of the experiment (6 weeks), τ is the duration of the cell cycle (28h, according to Gordon 1980, Canc Res 40, 4467-4472). C It is assumed that Dct expression is proportional to N. Therefore, (2) 𝐷𝐷𝐷𝐷𝐷𝐷𝑐𝑐 [𝑘𝑘 𝑐𝑐 −𝑘𝑘 𝑑𝑑 ]𝑡𝑡 ) 𝜏𝜏 �𝐷𝐷𝐷𝐷𝐷𝐷 = 𝑒𝑒 ( 𝑑𝑑 Where Dct c /Dct d is the ratio of Dct expression in the lungs of control vs PMNMDSC-depleted mice, k c is the fraction of proliferating cells in the lungs of control mice and k d is the fraction of proliferating cells in the lungs of PMN-MDSC-depleted mice. (3) 𝑘𝑘𝑐𝑐 = 𝑀𝑀𝑀𝑀𝑐𝑐 ( 𝜏𝜏�𝑆𝑆 ) Where MIc is the fraction of Ki67+ cells in the lung of control animals (1.9%, see Eyles et al. 2010 J. Clin. Invest.), S is the duration of Ki67 expression during the cell cycle (estimated to be 14h based on Lopez 1991 Cytometry 12:42-49) We tested the hypothesis that PMN-MDSC depletion would reduce the proliferation of disseminated cancer cells by 80% (reduction seen in primary tumor). Therefore (4) 𝑘𝑘𝑑𝑑 = 0.2𝑘𝑘𝑐𝑐 By combining (2), (3) and (4), one finds (5) 𝐷𝐷𝐷𝐷𝐷𝐷𝑐𝑐 �𝐷𝐷𝐷𝐷𝐷𝐷 = 𝑑𝑑 Therefore, even an 80% inhibition of cancer cell proliferation in the lungs could not account for the observed 5.5-fold decrease in Dct expression. We therefore favor the interpretation that the decrease observed in Dct expression is mainly due to reduced colonization of the lungs by cancer cells rather than reduced proliferation. D Appendix 2: Immune profiling gating strategy Identification of immune subsets in single-cell suspension of RETAAD tumors The figure shows the strategy used to enumerate immune cells in tumors. E Appendix 3: Table of chemokine/cytokine expression Differential expression of chemokines and cytokines in primary tumors and metastases Gene expression was measured by low density qRT-PCR arrays (except for CXCL2 which was done by individual qRT-PCR). Expression values were normalized to GAPDH and presented as a log (expression in primary tumor/metastases). p-values were calculated using two-tailed paired t-test. Data was from 11 paired primary tumors and metastases. Each pair of primary and metastatic tumors was taken from the same individual mouse. F Appendix 4: Mitf expression in LN PMN-MDSC depletion reduces tumor dissemination Reduced expression of Mitf in the mandibular LN of mice depleted of PMN-MDSC (depletion scheme A). Data was from 12 control and 20 PMN-MDSC depleted lymph nodes. Bars represent mean ± SEM **p . Appendix 2: Immune profiling gating strategy Identification of immune subsets in single-cell suspension of RETAAD tumors The figure shows the strategy used to enumerate immune cells in tumors (2) � = ( [ − ] ) Where k c is the fraction of proliferating cells in the tumor of control mice and k d is the fraction of proliferating cells in the tumor of PMN-MDSC-depleted. the number of melanoma cells at the end of the experiment, N 0 is the number of melanoma cells at the onset of the experiment, k is the fraction of proliferating cells, t is the duration of