There is a growing body of evidence that indicates that macrophages in the primary tumor promote tumor progression to the metastatic phenotype 
. This evidence comes from two sources: 1) clinical correlative data that shows in over 80% of the cases that a strong macrophage infiltrate is correlated with poor prognosis 
; 2) mouse models where genetic ablation of macrophages results in an inhibition of tumor progression and a reduced rate of metastasis 
. In part, this tumor promotion by macrophages is due to their role in regulating angiogenesis 
and to their ability to enhance tumor cell motility, invasion and intravasation (reviewed in 
). These macrophage activities therefore increase the potential for metastatic spread from the primary tumor not only by stimulating tumor invasiveness but also increasing the number of target vessels through which the tumor cells escape. These tumor-associated macrophages have been described as trophic macrophages 
or M2 
since they have a phenotype that suggests developmental, tissue remodeling as well as immune-regulatory functions that suppress cytotoxic immune responses 
Macrophages have also been found associated with tumors at metastatic sites 
. Earlier studies attempted to determine macrophage functions in distal metastasis events by their co-injection with tumor cells. Although the injection of tumor cells mimics the large number of tumor cells that are shed from late stage tumors 
mature macrophages are not observed in the circulation. Thus this system remains artificial and contradictory results have been observed 
. Recent studies however have indicated that bone marrow derived cells in addition to their roles in the primary tumor also promote metastasis through their effects at secondary sites. In particular, it has been shown that primary tumors influence the selection of metastatic sites through the secretion of factors that recruit bone marrow cells to these sites to create the so-called pre-metastatic niche in which circulating tumor cells settle and prosper 
. The bone marrow cells that populate these niches have not been fully characterized but are of myeloid origin 
. Despite these important studies, there remains no direct evidence for macrophages influencing the events subsequent to homing of metastatic cells 
. To test whether macrophages do have a role in these subsequent events, we used an experimental metastasis model to circumvent the influence of primary tumors. Using this system we show that macrophage play a significant role in the extravasation of metastatic cells as well as in their establishment and growth in the lung.
This conclusion that macrophages have a major impact on metastatic cell seeding and persistent growth was based upon studies using three different and independent methods of macrophage ablation. The first used mice carrying a null mutation (Csf1op
) in the major macrophage growth factor, CSF-1, to deplete the macrophages 
and these studies showed a profound inhibition of metastatic cell seeding and persistent growth. Interestingly, there was an effect on the metastatic index (a sum of metastasis number and size) according to the null allele frequency. This is unusual since previous studies showed that heterozygous mice have normal serum concentrations of CSF-1 and normal populations of macrophages in most tissues tested 
. Indeed in the lung we showed that+/Csf1op
mice also have normal resident macrophage numbers (Figure S3
). However, a radioimmuno-assay of lung tissue CSF-1 revealed a reduced CSF-1 expression in the heterozygote (data not shown). This indicates the strong dependence of host CSF-1 level for efficient tumor cell seeding and growth in lung. It will be interesting to determine if a heterozygote effect can be found in human populations with breast cancer. It should be noted that increased levels of circulating CSF-1 in human patients with breast, ovarian and endometrial cancer is correlated with poor prognosis 
The second method of macrophage depletion used the classical method of liposome-encapsulated Clodronate that causes macrophage death after its selective phagocytosis by mature macrophages 
but not by tumor or other cells 
. When the macrophages were depleted by this method spanning the period of metastasis assay, both tumor cell seeding and persistent growth were inhibited. Importantly macrophage depletion using this method after metastatic seeding resulted in the reduction of metastatic growth.
The third method was based upon the observation (see results
and below) that the recruited macrophage population was different from the lung resident CD11c+ macrophages in their expression of CD11b. This enabled the use of a suicide gene approach whereby the CD11b+positive macrophages were ablated by dint of their expression of the diptheria toxin receptor from the CD11b promoter 
. Using this method we showed that this CD11b+population of macrophages was required for both seeding and persistent growth. Importantly the resident population of macrophages were unaffected by this treatment showing that it is the newly recruited population that is responsible for the effects. Furthermore similar to the data obtained following L-Clodronate ablation, depletion of these macrophages using the diptheria toxin approach after the metastasis had been established also resulted in inhibition of growth. Thus we can conclude from these three approaches that the CD11b+population of mature macrophages that is recruited to the lung in response to the arrival of metastatic cells promote their subsequent establishment and growth. This ability to inhibit the growth of metastases once established also suggests that macrophage depletion will have significant therapeutic potential.
Our data showed the recruited macrophage population is characterized by a cell surface marker signature of F4/80+CSF-1R+CD11b+Gr1-CX3CR1high
and their recruitment to pulmonary metastases is independent of the metastatic cell type or the species of origin (mouse and human). It should also be noted that although this experimental model of metastasis is somewhat artificial in that a bolus of cells arrives to the lung within 5 minutes, a population of macrophages with a similar phenotype is also recruited to spontaneous metastases derived either from primary autochthonous mammary tumors or from xenotransplants of human breast cancer cells in immunocompromized mice. These macrophages display characteristic markers such as F4/80, Mac3, CSF-1R and are phagocytic because they up-take liposomes (see below) while being negative for granulocyte marker, Gr1. While none of these markers are unique to macrophages in themselves, the combination and their high level of expression together with the cells tissue location indicates that these cells are definitive macrophages 
. They probably differentiate from monocytes precursors that are restricted to blood and seed most macrophages but it cannot be ruled out that they cross-differentiate from other resident macrophages as has been described in some immune responses 
. This metastasis-associated CD11b+macrophage population is different from the classically defined inflammatory Gr1+CCR2+CX3CR1low
macrophages and Gr1-CCR2-CX3CR1high
tissue macrophages 
and is also distinct from other recently identified macrophage populations in the primary tumor microenvironment such as myeloid suppressor and pro-angiogenic macrophages 
. It is also different from the populations found in the primary tumors of PyMT mice in the expression of CXCR4 and Tie2 (our unpublished data). They are however, similar to a recently identified anti-inflammatory macrophage important in facilitating myogenesis in terms of Gr1 expression 
. This distinct phenotypes give further support to the notion of the tumor microenvironment educating the recruited macrophages to give functions that are advantageous to the growing tumor cells 
Our data shows that macrophages not only affect metastatic growth but also seeding. This ability of tumor cells to establish themselves at the metastatic site is considered one of the major rate limiting steps in metastasis 
. But there is considerable controversy about how tumor cells interact with blood vessels in their metastatic target organ and the subsequent steps in establishment 
. Because of the unique vasculature in the lung, conventional method of vessel labeling using dextran does not work well since this molecule leaks out easily. Furthermore, because lung is fragile, the vessel structure is often damaged during fixation and sectioning. Thus to examine the early steps we used an intact lung imaging system 
with methods that visualizes macrophages, blood vessels and tumor cells simultaneously followed by detailed quantitative analysis of extravasation events together with a QPCR method that accurately measures tumor cell number. For these methods single cell suspensions were carefully prepared to avoid emboli formation and these were not observed in the lung vasculature after tumor cell injection. Indeed as the individual tumor cells arrive in the lung they begin a process of attachment and invasion. While this process is inefficient the presence of the tumor cells stimulates the recruitment of macrophages that form intimate contacts with them as soon as they extrude through the vessel walls. Importantly, the rate of extravasation of the tumor cells was significantly reduced after macrophage depletion with a co-incident reduction in tumor cell viability. Once extravasation is completed tumor cell proliferation begins and there is a positive correlation of tumor cell growth with macrophage association at these early stages. Consistent with this role of macrophages in promoting growth, macrophage depletion at this time resulted in a two-fold increase in the population doubling time. Thus this novel imaging technique together with rigorous quantification shows that extravasation can be a rate limiting step in the metastatic process and also identifies this population of macrophages as a component of the microenvironment that plays a critical role in this step as well as in the subsequent growth of the surviving tumor cells. Further this imaging of early stage events in metastasis is consistent with our end point stereological measurements that show macrophages positively influence both tumor cell seeding and persistent growth.
Metastasis remains an intractable problem clinically and is therefore the major cause of death in cancer patients. Based on the data in current study, we suggest a model for the macrophage enhancement of metastasis at the distal target organ (). Following arrest of the tumor cells in capillaries of metastasis target organ, monocytes were quickly recruited and differentiated in situ into metastasis associated macrophage phenotype with a distinctly defined cell surface marker phenotype. This recruitment is at least in part under the influence of locally synthesized CSF-1, a well-documented growth and differentiation factor for macrophages. In addition as these macrophages express receptors for CCL-2 and VEGF (CCR2 and VEGFR1 respectively), both cytokines that are chemotactic for macrophages, it is likely that such signaling molecules will play a role in this recruitment process. These CD11b+macrophages recognizes extravasating tumor cells and our imaging shows that they interact with them directly and help them invade into the lung parenchyma. This is presumably through the secretion of proteases, growth, and motility and survival factors. In the absence of these newly differentiated macrophages, this process of tumor cell extravasation is very inefficient and the tumor cells rapidly die by apoptosis and thus the seeding efficiency is very low. Once extravascular the tumor cells continue to send signals to recruit and also possibly influence the differentiation of the macrophages into trophic ones 
that further enhance tumor cell viability and growth. When the tumors attain a certain size these macrophages are also likely to provide angiogenic factors as they have been documented to do in the primary tumor 
that then help in the vascularization needed for continuous metastatic growth. In this scenario several macrophage signaling pathways and functions are likely to be engaged at the different steps of tumor cell seeding, initial and persistent growth. These are continuous process and overlapping processes as ablation of macrophages after the metastatic lesions are established retards their growth significantly. These data suggests that macrophages themselves or their unique signaling pathways represent new therapeutic targets that may be efficacious in reducing cancer mortality.
Model for macrophage promotion of metastasis at distant sites.