Recruitment of a myeloid cell subset (CD11b/Gr1mid) via CCL2/CCR2 promotes the development of colorectal cancer liver metastasis*§


  • Potential conflict of interest: Nothing to report.

  • §

    Supported by Cancer Research UK (CRUK) with assistance from the CRUK/EPSRC Oxford Cancer Imaging Centre. L.Z. was funded by Cancer Research UK China Fellow Programme and now works in the Shandong Provincial Cancer Hospital and Institute, Jinan, China. A.N.G.-W. was funded by a Royal College of Surgeon of England Research Fellowship.


Liver metastasis from colorectal cancer is a leading cause of cancer mortality. Myeloid cells play pivotal roles in the metastatic process, but their prometastatic functions in liver metastasis remain incompletely understood. To investigate their role, we simulated liver metastasis in C57BL/6 mice through intrasplenic inoculation of MC38 colon carcinoma cells. Among the heterogeneous myeloid infiltrate, we identified a distinct population of CD11b/Gr1mid cells different from other myeloid populations previously associated with liver metastasis. These cells increased in number dramatically during establishment of liver metastases and were recruited from bone marrow by tumor-derived CCL2. Liver metastasis of Lewis lung carcinoma cells followed this pattern but this mechanism is not universal as liver colonization by B16F1 melanoma cells did not recruit similar subsets. Inhibition of CCL2 signaling and absence of its cognate receptor CCR2 reduced CD11b/Gr1mid recruitment and decreased tumor burden. Depletion of the CD11b/Gr1mid subset in a transgenic CD11b-diphtheria toxin receptor mouse model markedly reduced tumor cell proliferation. There was no evidence for involvement of an adaptive immune response in the prometastatic effects of CD11b/Gr1mid cells. Additionally, an analogous myeloid subset was found in liver metastases of some colorectal cancer patients. Conclusion: Collectively, our findings highlight the importance of myeloid cells—in this case a selective CD11b/Gr1mid subset—in sustaining development of colorectal cancer liver metastasis and identify a potential target for antimetastatic therapy. (HEPATOLOGY 2013)

Metastatic colorectal cancer (CRC) is a prominent cause of cancer mortality worldwide.1 Hepatic metastases are found in approximately 15% of CRC patients at primary diagnosis2 with 14% subsequently developing metastases.3 Development of new treatment modalities for CRC liver metastasis is urgently required and a greater understanding of the biology of this process will help establish new therapeutics aimed at downstaging the disease, improving operability, and prolonging survival.

Metastasis is a multistep process involving complex and continuous interactions between tumor cells and the host microenvironment.4 Several myeloid-derived cell types have been shown to play key roles in the metastatic cascade, including intravasation, extravasation,5 and colonization at secondary sites by stimulating tumor cell proliferation and angiogenesis and suppressing antitumor immunity.6-8 However, delineation of their roles in metastasis is complicated by the heterogeneity of myeloid phenotypes that appears to be both tumor- and organ-selective. Vascular endothelial growth factor receptor 1 (VEGFR1)+ hematopoietic progenitor cells accumulated at premetastatic sites to promote adherence and growth of lung Lewis carcinoma (LLC) and B16F1 tumor cells,9 while a Mac-1+ myeloid population with different markers was recruited by S100A8/A9 to premetastatic lung to promote LLC tumor migration.10 At later stages of metastasis, CD11b+/CD115+ inflammatory monocytes were recruited via CCL2/CCR2 to experimentally induced and spontaneous metastases of mammary tumors,11 and subsequently differentiated into CD11b+/Gr1 macrophages to promote tumor cell extravasation and growth.12 Such complexity highlights the importance of thorough characterization of heterogeneous tumor-infiltrating myeloid cells and the factors driving metastasis. Without detailed characterization, understanding the contribution of myeloid subsets to the metastatic process and identification of specific targets for therapeutic manipulation becomes difficult.

Although the development of lung metastasis is well studied, the role of myeloid infiltrates in liver metastasis has received less attention. Recently, Kitamura et al.13 identified an immature myeloid cell population, recruited via CCL9/CCR1, in a mouse model of colon cancer liver metastasis. Inhibition of their accumulation suppressed metastatic growth,13 thus reinforcing the idea that myeloid cells are important for metastatic development in the liver.

Here we report a different prometastatic CD11b/Gr1mid myeloid subset associated with CRC liver metastases. Recruitment of these cells was dependent on CCL2/CCR2 and its inhibition markedly reduced tumor burden. Moreover, depletion of the CD11b/Gr1mid subset significantly decreased tumor cell proliferation. Cells analogous to the CD11b/Gr1mid subset were identified in patients with CRC liver metastases, underscoring their potential for therapeutic manipulation.


ANOVA, analysis of variance; cDNA, complementary DNA; CRC, colorectal cancer; DT, diphtheria toxin; DTR, diphtheria toxin receptor; FACS, fluorescence-activated cell sorting; GFP, green fluorescent protein; IL, interleukin; KO, knockout; LLC, lung Lewis carcinoma; PBS, phosphate-buffered saline; SCID, severe combined immunodeficiency; TIMP-1, tissue inhibitor of metalloproteinase 1; VEGFR1, vascular endothelial growth factor receptor 1.

Materials and Methods

Animals, Cell Lines, and Patient Samples.

The sources of mice, cell lines, and patient samples are detailed in the Supporting Information. Animal procedures were performed in accordance with the UK Animal (Scientific Procedures) Act 1986 and followed local ethics review.

Liver Metastasis Model.

Tumor cells (5 × 105/100 μL phosphate-buffered saline [PBS]) were injected into the spleens of C57BL/6, CCR2 knockout (KO), severe combined immunodeficiency (SCID), or CD11b-diphtheria toxin receptor (DTR) mice. The spleens were removed, and the mice were sacrificed on day 14. To inhibit CCL2, C57BL/6 mice received daily intraperitoneal injections of CCL2 antibody (15 μg/mouse; R&D Systems) or rat immunoglobulin G2b control (R&D Systems) following tumor cell inoculation. CD11b-DTR or C57BL/6 mice received diphtheria toxin (7.5 ng DT/g body weight; List Biological Laboratories) or PBS via intraperitoneal injection on day 7 and 9, and sacrificed on day 11.

Adoptive Transfer.

Bone marrow cells were isolated from female C57BL/6-Tg(UBC-GFP)30Scha/J mice (provided by Prof. Richard Cornall, University of Oxford, UK), and 2 × 106 cells were transferred intravenously into C57BL/6 mice on day 11. Mice were sacrificed on day 12.

Preparation of Single Cell Suspensions.

Single cell suspensions were prepared from livers, bone marrow, and blood as described in the Supporting Information and were adjusted to 107 cells/mL for fluorescence-activated cell sorting (FACS) analysis.

FACS Analysis.

Antibodies used are detailed in the Supporting Information. FACS analysis was performed using a FACSCalibur flow cytometer (BD Biosciences) and analyzed with FlowJo software version 7.2.5 (Tree Star, Ashland, OR).

Gene Expression Arrays.

RNA was isolated with Trizol (Invitrogen) and complementary DNA (cDNA) (0.5 μg) synthesized using the SuperScript VILO cDNA kit (Invitrogen). cDNA was mixed with TaqMan Universal PCR Master Mix, added to the TaqMan Mouse Immune Array (Applied Biosystems) and reaction performed using an ABI Prism 7900HT system (Applied Biosystems). Data were analyzed using the ΔΔCT method and normalized to 18S RNA.


Immunohistochemical staining of tissue samples is described in Supporting Materials.

Cytokine Analysis.

Cytokine expression was assayed using the Proteome Profiler Mouse Cytokine Array (R&D Systems). Membranes were detected with streptavidin-Alexa 700 (Invitrogen) using a two-channel near-infrared Odyssey scanner (LI-COR, UK), and spot intensities were quantified using software developed by our laboratory. CCL2 was quantified using the mouse CCL2 (monocyte chemoattractant protein-1) enzyme-linked immunosorbent assay kit (eBioscience).

Statistical Analysis.

Data are expressed as the mean ± SEM and were analyzed using an unpaired Student t test or one-way analysis of variance (ANOVA) with a Bonferroni posttest. Correlation coefficients were calculated using nonparametric Spearman correlation analysis. P ≤ 0.05 was considered statistically significant.


Discrete CD11b/Gr1 Subsets Are Found in Liver Metastases of Colorectal Carcinoma.

Macroscopic liver metastases were observed 7 days after MC38GFP+ inoculation into C57BL/6 mice (Supporting Fig. 1A). CD11b+ myeloid cells in tumor-bearing livers were assessed via FACS analysis and were segregated based on Gr1 (Ly6G/Ly6C) expression (Supporting Fig. 1B). Three discrete subsets, subsequently described as CD11b/Gr1high, CD11b/Gr1mid, and CD11b/Gr1low cells (Fig. 1A), were identified at day 0, 7, and 14, respectively (Supporting Fig. 1C). These subsets were further characterized by morphology (Fig. 1B) and surface marker expression (Fig. 1C). CD11b/Gr1high cells had multilobed nuclei typical of granulocytes, whereas CD11b/Gr1mid and CD11b/Gr1low cells had ovoid nuclei typical of monocytes/macrophages (Fig. 1B). All CD11b/Gr1 subsets expressed Ly6C and CCR5, but F4/80 was detected only on CD11b/Gr1mid cells and Ly6G was detected only on CD11b/Gr1high cells. CD11b/Gr1low cells had bimodal expression of CD11c, CCR4, and CXCR4, and both CD11b/Gr1mid and CD11b/Gr1low cells expressed CCR2. VEGFR1 and CCR1, previously reported to be expressed by myeloid cells infiltrating lung9 and liver metastases,13 were not detected in any of the CD11b/Gr1 subsets (Fig. 1C). These subsets were negative for natural killer cell, T cell, and B cell markers (Supporting Fig. 1D).

Figure 1.

CD11b+ cells in liver metastases comprise distinct Gr1-expressing subsets. (A) Myeloid cells in tumor-bearing livers were separated by FACS analysis based on CD11b and Gr1. (B) Isolated CD11b/Gr1 subsets were cytospun for hematoxylin and eosin staining (magnification ×100) or (C) stained with antibodies to indicated surface markers (open histograms) or isotype controls (closed histograms), representative of two experiments. (D) RNA (5 μg) from each CD11b/Gr1 subset was used to detect expression of inflammatory mediators. Data are expressed as relative messenger RNA levels normalized to 18S RNA.

CD11b/Gr1mid and CD11b/Gr1low cells had similar cytokine messenger RNA profiles, with relatively high expression of proinflammatory mediators CCL2, CCL3, CCL5, interleukin (IL)-1α, IL-1β, IL-15, IL-18 and tumor necrosis factor (Fig. 1D), resembling a mixed M1/M2-like phenotype.14 CD11b/Gr1high cells expressed proinflammatory IL-1β, but expression of other cytokines was low.

CD11b/Gr1mid and CD11b/Gr1low Cells Increase During Metastatic Progression.

We then compared myeloid subsets in tumor-bearing and naïve livers. CD11b/Gr1mid percentages increased significantly 14 days after MC38GFP+ inoculation. A more modest increase in CD11b/Gr1low cells was observed, but CD11b/Gr1high cells remained constant (Fig. 2A). In light of this finding, and because CD11b/Gr1mid and CD11b/Gr1low cells expressed higher levels of classic proinflammatory genes (Fig. 1D), we focused further study on these two subsets.

Figure 2.

CD11b/Gr1mid cells are derived from bone marrow. (A) Percentages of CD11b/Gr1 cells in the liver 14 days after MC38GFP+ or PBS inoculation (n = 3/group of four experiments). *P < 0.05 (Student t test). (B) Percentages of CD11b/Gr1mid and (C) CD11b/Gr1low cells in bone marrow, blood, and liver at different intervals after MC38GFP+ inoculation (n = 3/group/time point). (D) GFP+ bone marrow cells in the liver of MC38 or PBS-inoculated recipient mice [(n = 3/group); *P < 0.05 (Student t test)] 1 day after adoptive transfer (E) were observed within liver metastases (magnification ×100; inset, magnification ×200) and (F) expressed CD11b, F4/80, and CCR2.

Percentages of CD11b/Gr1mid and CD11b/Gr1low cells in bone marrow, blood, and liver of tumor-bearing mice were analyzed at various time points during metastatic growth. Levels of CD11b/Gr1mid cells in bone marrow peaked at day 5, and decreased thereafter, which coincided with increasing levels in blood and liver. Circulating and hepatic CD11b/Gr1mid cell numbers continued to rise by day 14 (Fig. 2B). In contrast, bone marrow and circulating CD11b/Gr1low cell numbers remained constant with time while increasing in the liver abruptly from day 12 (Fig. 2C). These results suggest that the CD11b/Gr1mid subset is recruited from bone marrow during development of liver metastasis, whereas the CD11b/Gr1low population likely derived from expansion or differentiation of resident cells after metastases had established.

To confirm the bone marrow origin of the CD11b/Gr1mid subset, GFP+ cells isolated from bone marrow of GFP transgenic mice were transferred intravenously into C57BL/6 mice 11 days after MC38 or PBS inoculation. Significantly more GFP+ bone marrow cells were found in MC38-inoculated tumor-bearing livers compared with PBS-inoculated controls (Fig. 2D). These GFP+ cells were in the peritumoral regions of liver metastases (Fig. 2E), and were CD11b+, CCR2+, and F4/80+ (Fig. 2F), markers expressed only by the CD11b/Gr1mid population.

To investigate whether similar CD11b/Gr1mid and CD11b/Gr1low subsets are associated with liver metastasis of other cancer cell lines, we inoculated B16F1GFP+ and LLCGFP+ cells into C57BL/6 mice. Metastases were observed in the liver at day 14 when myeloid infiltrates were assessed. Formation of LLCGFP+ tumor colonies resulted in a significant increase in the CD11b/Gr1mid population, similar in extent to MC38GFP+ inoculation. In contrast, CD11b/Gr1mid cell numbers were not significantly altered after B16F1GFP+ colonization (Fig. 3A). LLCGFP+ inoculation also led to a substantial increase in CD11b/Gr1low cell numbers, whereas moderate increases were observed after B16F1GFP+ and MC38GFP+ inoculation (Supporting Fig. 3C). Thus, LLCGFP+ colonization was analogous to that of MC38GFP+ in recruiting CD11b/Gr1mid cells, whereas this recruitment was dispensable for B16F1GFP+ cells.

Figure 3.

Tumor-derived CCL2 promotes CD11b/Gr1mid accumulation in liver metastases. (A) Percentages of CD11b/Gr1mid cells in the liver 14 days after MC38, LLC, B16F1, or PBS inoculation (n = 3/group of two experiments). *P < 0.05 (ANOVA, Bonferroni posttest). (B) Relative cytokine levels in culture media of MC38, LLC, and B16F1 cells. (C) CCL2 concentration in MC38, LLC, and B16F1 culture media. *P < 0.05 (ANOVA, Bonferroni posttest). (D) Percentages of MC38GFP+ and (E) CD11b/Gr1mid cells at various intervals after MC38GFP+ inoculation significantly correlated with serum CCL2 (n = 3/time point). *P < 0.05 (Spearman correlation analysis).

Tumor Cells that Secrete CCL2 Promote CD11b/Gr1mid Recruitment.

To identify factors involved in recruitment of bone marrow-derived CD11b/Gr1mid cells to liver metastases, we compared the cytokine expression profile of MC38, B16F1, and LLC cells. MC38 cells expressed high levels of CCL2 and moderate levels of CXCL1, CXCL10, and tissue inhibitor of metalloproteinase 1 (TIMP-1). Moderate levels of CCL2, CXCL1, and TIMP-1 were also detected in culture medium of LLC cells. B16F1 cells produced moderate levels of CXCL10 and CCL5 but CCL2 was not detected (Fig. 3B). Additionally, we tested another B16 melanoma variant cell line, B16F10, and found it to have a similar cytokine expression profile as B16F1 (Supporting Fig. 3A), and to produce liver metastases independent of CD11b/Gr1mid and CD11b/Gr1low cells (Supporting Fig. 3B).

Because CCL2 has been widely reported as a chemoattractant for tumor-associated myeloid cells,15, 16 we compared its expression in culture media from the three cell lines. MC38 and LLC cells produced significantly more CCL2 than B16F1 cells (Fig. 3C). Moreover, serum CCL2 in C57BL/6 mice increased following MC38GFP+ inoculation and significantly correlated with increased numbers of MC38GFP+ tumor cells (Fig. 3D) and CD11b/Gr1mid cells (Fig. 3E) in the liver as metastasis progressed. These findings suggest CCL2 as a candidate for recruiting CD11b/Gr1mid cells to liver metastases.

CD11b/Gr1mid Recruitment Is Mediated by CCL2/CCR2.

CCL2 binds both CCR2 and CCR4,17 but only CCR2 is expressed by CD11b/Gr1mid cells (Fig. 1C). To examine whether CCL2/CCR2 is required for CD11b/Gr1mid recruitment, we attempted to inhibit CCL2 using a monoclonal blocking antibody. Essentially the same numbers of hepatic CD11b/Gr1mid and CD11b/Gr1low cells were found in MC38GFP+-inoculated mice following CCL2 blockade as in mice treated with isotype-matched antibody (Fig. 4A). However, serum CCL2 was significantly higher in α-CCL2–treated mice at day 6 than controls, and similar to controls at day 14 (Fig. 4B). These findings suggest a compensatory increase in CCL2 following pharmacological blockade, thus doses of blocking antibody administered may not be sufficient to inhibit CCL2-mediated effects during metastatic development.

Figure 4.

CD11b/Gr1mid accumulation is mediated by CCL2/CCR2. (A) Percentages of CD11b/Gr1low and CD11b/Gr1mid cells in the liver of C57BL/6 mice at day 6 and 14 post-MC38GFP+ inoculation following anti-CCL2 or immunoglobulin G (IgG) control treatment and (B) serum CCL2. (C) Percentages of CD11b/Gr1low and CD11b/Gr1mid cells in the liver of C57BL/6 mice at day 6, 9, and 13 after MC38CCL2 KD or MC38Lenti Ctrl inoculation and (D) serum CCL2. (E) Percentages of CD11b/Gr1low and CD11b/Gr1mid cells in the liver of CCR2 KO or wild-type (WT) controls at day 14 after MC38GFP+ inoculation and (F) serum CCL2 (n = 3-4/group of two experiments). *P < 0.05 (Student t test).

Given this result, we sought alternative approaches to abrogate CCL2 signaling. Transfection of MC38 cells with a lentivirus encoding short hairpin RNA targeting CCL2 (MC38CCL2 KD) decreased CCL2 expression by two-fold (Supporting Fig. 3D), and a significant reduction in serum CCL2 was observed at day 6, 9, and 13 in MC38CCL2 KD-inoculated mice compared with MC38Lenti Ctrl-inoculated controls (Fig. 4D). MC38CCL2 KD-inoculated mice had fewer hepatic CD11b/Gr1mid cells at day 6 and 9 (Fig. 4C), although by day 13 levels were similar to those of controls. Hepatic CD11b/Gr1low cell numbers in MC38CCL2 KD-inoculated mice were not noticeably different to those of controls at all three time points, indicating that CCL2 knockdown did not influence accumulation of this subset.

In addition to inhibiting CCL2, we investigated the effects of eliminating its cognate receptor, CCR2. Fewer CD11b/Gr1mid and CD11b/Gr1low cells were found in livers of CCR2 KO mice 14 days after MC38GFP+ inoculation compared with wild-type C57BL/6 controls (Fig. 4E). Serum CCL2 was significantly higher in CCR2 KO mice compared with controls (Fig. 4F), once again alluding to a compensatory up-regulation of CCL2 when CCL2/CCR2 signaling is inhibited. These data suggest that CCL2 expression by tumor cells and myeloid cell expression of CCR2 have a considerable impact on CD11b/Gr1mid recruitment to liver metastases.

CD11b/Gr1mid and CD11b/Gr1low Cells Support Tumor Growth.

When CD11b/Gr1mid recruitment was inhibited in MC38CCL2 KD-inoculated mice, we found fewer MC38 tumor cells in the liver compared with controls at day 6, with a significant decrease at day 9. However, tumor cell numbers in MC38CCL2 KD-inoculated mice were comparable to controls by day 13 (Fig. 5A), mirroring the lack of difference in CD11b/Gr1mid recruitment at this time point. Likewise, fewer MC38GFP+ cells were detected in CCR2 KO mice compared with controls (Fig. 5B), although the differences were not as striking. Overall, decreased accumulation of CD11b/Gr1mid and CD11b/Gr1low cells in liver metastases caused a substantial reduction in tumor burden.

Figure 5.

CD11b/Gr1mid cells support tumor growth. Percentages of MC38 tumor cells (A) in C57BL/6 mice at day 6, 9, and 13 after MC38CCL2 KD or MC38Lenti Ctrl inoculation, (B) in CCR2 KO and C57BL/6 (WT) mice at day 14 after MC38GFP+ inoculation, and (C) in CD11b-DTR mice at day 12 following PBS or DT treatment (n = 3-4/group of two experiments). *P < 0.05 (Student t test). Liver tissue sections from CD11b-DTR mice after PBS or DT treatment were stained for (D) Ki67 and (E) 5-bromo-2'-deoxyuridine and ratio of positive cells to tumor area determined (3-5 fields of view from five sections/group; *P < 0.05 (Student t test). (F) Percentages of CD11b/Gr1low and CD11b/Gr1mid cells in the liver of SCID mice 14 days after MC38GFP+ or PBS inoculation (n = 3-4/group) *P < 0.05 (Student t test).

In an attempt to deplete the CD11b/Gr1mid and CD11b/Gr1low subsets, CD11b-DTR mice bearing a human diphtheria toxin receptor (DTR) transgene driven by a CD11b promoter were used. Here, conditional ablation of CD11b+ cells can be achieved by diphtheria toxin (DT) administration.18 DT was administered to CD11b-DTR mice on day 7 and 9 after MC38GFP+ inoculation, a time when metastatic colonies had formed, and mice were sacrificed on day 11. DT administration markedly depleted CD11b/Gr1mid and CD11b/Gr1low cells in the liver compared with treatment with PBS (Supporting Fig. 4A) and had little effects on levels of T (CD3+) or B (CD19+) cells (Supporting Fig. 4B). Neutrophils were shown to be unaffected by DT in CD11b-DTR mice19 and numbers of CD11b/Gr1high cells were similarly unaffected (Supporting Fig. 4A). Livers of control mice had large metastatic colonies, whereas metastases were much smaller in DT-treated mice (Supporting Fig. 4C) and correspondingly, markedly fewer MC38GFP+ tumor cells were detected in livers of DT-treated mice (Fig. 5C). Administration of DT to wild-type C57BL/6 mice did not deplete CD11b/Gr1mid cells, affect the number of MC38GFP+ cells in the liver, or the formation of liver metastases compared with controls (Supporting Fig. 4D-F). Tumor cell proliferation was assessed by staining liver tissue sections of CD11b-DTR mice after DT or PBS treatment. A pronounced two-fold reduction in both bromodeoxyuridine incorporation (BrDu) and Ki67-positive cells (Fig. 5D,E) were observed after DT treatment compared with controls. Overall, depletion of the CD11b/Gr1mid and CD11b/Gr1low subsets minimized metastatic growth, causing an appreciable reduction in tumor burden.

Adaptive Immunity Is Not Essential in the Recruitment and Protumorigenic Effects of CD11b/Gr1mid and CD11b/Gr1low Cells.

We considered the possibility that CD11b+ cell depletion could instigate an adaptive immune response leading to decreased tumor burden. However, T cell and B cell numbers were comparable between DT-treated CD11b-DTR mice and controls (Supporting Fig. 4B). We also assessed myeloid infiltrates 14 days after MC38GFP+ inoculation in SCID mice (Supporting Fig. 5) and found myeloid subsets similar to those observed in wild-type C57BL/6 mice (Fig. 5F). Taken together, these findings suggest that accumulation of the CD11b/Gr1mid and CD11b/Gr1low subsets and decreased tumor growth after their depletion did not involve an adaptive immune response.

CD11b+/CCR2+ Cells Are Observed in Human CRC Liver Metastasis.

Elevated CCL2 has been reported in human primary and metastatic CRC tumor tissues,20 and we found significantly higher levels of serum CCL2 in patients with primary (Dukes class A and B) and metastatic (Dukes class C and D) CRC compared with healthy controls (Fig. 6A). To examine whether CD11b/Gr1mid and CD11b/Gr1low cells might also be important in liver metastasis of human CRC, we stained tissue sections for CD11b and CCR2, markers that characterize both subsets. Few CD11b+ cells were detected in normal liver tissues, consistent with previous reports.21, 22 CCR2+ cells were also infrequent in normal liver, and cells expressing both CD11b and CCR2 were not detected in five out of five tissue samples (Fig. 6B). However, CD11b+ cells were evident in the tumor core, at the invasive edge of tumor colonies, and in the tumor stroma of liver metastasis tissues. CCR2+ cells were also found at the invasive edge but were mostly localized in the stroma (Fig. 6C). Cells expressing both CD11b and CCR2 were detected in ∼30% of liver metastases (two out of seven tissues) (Fig. 6D).

Figure 6.

Cells characteristic of CD11b/Gr1mid and CD11b/Gr1low are detected in liver metastases from CRC patients. (A) Serum CCL2 in CRC patients with Dukes class A, B, C, and D (n = 30/group). *P < 0.05 (ANOVA, Bonferroni posttest). (B) Tissue samples of normal liver and (C) liver metastases stained for CD11b and CCR2, and for nuclei by 4',6-diamidino-2-phenylindole. Images acquired at magnification ×100 by epifluorescence imaging and represent 5-7 samples. (D) Colocalization of CD11b and CCR2 reactivity in liver metastasis tissues, acquired by confocal imaging (magnification ×200; 3-4 fields of view).


Myeloid cells are recruited via different mechanisms and can facilitate the metastatic process.11, 12 Here we report a CD11b/Gr1mid subset that is essential for the development of liver metastasis by some but not all tumor cell lines. CD11b/Gr1low cells may exert similar prometastatic effects, as elimination of the two subsets substantially reduced tumor growth. CD11b/Gr1mid cells likely derived from bone marrow progenitor cells, and given that the CD11b/Gr1low subset expressed similar inflammatory mediators and surface markers as CD11b/Gr1mid cells, these two subsets may arise from each other by differentiation.

CCL2 has been reported to recruit myeloid cells in cancer,23 and impairment of CCL2/CCR2 signaling inhibits tumor growth.24 Furthermore, CCL2 is up-regulated in primary and metastatic CRC20 and its expression in CRC patients is a predictive marker of liver metastasis25 and of postoperative hepatic recurrence.26 We show that CD11b/Gr1mid recruitment is mediated by tumor-derived CCL2, akin to the signaling mechanism reported in lung metastasis,11, 20 and represents an alternative pathway to the CCL9/CCR1-dependent mechanism that has been reported.13 On the other hand, B16F1 and B16F10 cells expressed little CCL2 and did not recruit CD11b/Gr1mid cells during liver colonization. The extent of redundancy of these myeloid populations and the circumstances in which different mechanisms are operative remains to be determined.

Although CCL2 is produced by other stromal cells in the tumor microenvironment,27 we found that inhibition of tumor-derived CCL2 adequately suppressed recruitment of CD11b/Gr1mid cells, suggesting that CCL2 is predominantly synthesized by tumor cells. However, although we observed a striking reduction in CD11b/Gr1mid recruitment and tumor burden by blocking CCL2 at earlier time points, these effects were mitigated at later stages. Our results suggest that targeting CCL2 may cause a compensatory up-regulation of the chemokine and/or provoke additional mechanisms to restore CD11b/Gr1mid recruitment. These data raise some questions about CCL2 as a therapeutic target.

We found that depletion of CD11b/Gr1mid and CD11b/Gr1low cells in CD11b-DTR mice markedly decreased tumor cell numbers with an overall reduction in tumor cell proliferation. Here, functions of these cells can be partially defined. In lung metastasis, CD11b+ monocytes were recruited early in the metastatic process to shape the premetastatic niche,9 whereas mobilization of CD11b+/CCR2+ monocytes facilitated extravasation of breast cancer cells.11 Extravasation of tumor cells occurs rapidly in the liver, unlike the lung,28 and we found the greatest influx of CD11b/Gr1mid cells in liver after tumor colonies had established. Moreover, CD11b/Gr1mid and CD11b/Gr1low cells were depleted after metastases had formed, so the ensuing reduction in tumor burden was independent of premetastatic niche formation and extravasation.

Persistent proliferation of tumor cells can occur as a consequence of immune evasion. Myeloid-derived suppressor cells are CCR2+ and have been shown to suppress T cell infiltration and proliferation.29, 30 Because the CD11b/Gr1mid and CD11b/Gr1low subsets expressed CCR2, we considered the possibility that their prometastatic effects are dependent on a T cell–mediated response. Nonetheless, this seems unlikely, because their depletion did not elicit evidence of an adaptive immune response and tumor burden and myeloid recruitment was analogous in SCID mice compared with immunocompetent animals.

We further considered the implications of these findings for humans. CD11b+/CCR2+ cells characteristic of the CD11b/Gr1mid and CD11b/Gr1low subsets were found in tissue samples from several CRC patients with liver metastasis but were absent in normal liver. Hence, selected cases of liver metastasis may provoke similar infiltration of the CD11b/Gr1mid and CD11b/Gr1low subsets, and because these cells were found clustered around the tumor region, they may play a pivotal role in metastatic tumor development in humans. It remains to be determined whether there will be stratification in liver metastases wherein some depend upon infiltration of myeloid cells while others do not.

Overall, our study underscores the importance of myeloid cells in CRC liver metastasis and demonstrates that a distinct CD11b/Gr1mid subset, expressly different from other myeloid populations that have been described, is recruited to liver metastasis to promote tumor cell proliferation. However, bypass mechanisms clearly exist to counteract certain blocking strategies, and a thorough understanding of how these CD11b/Gr1mid and CD11b/Gr1low subsets affect liver metastasis will allow us to uncover novel and more effective candidates for therapeutic targeting.