Using the Lewis lung carcinoma (LL2) implantation mouse model, we characterized the hematopoietic compartment in the tumor stroma. Figures 1A–C and Table 1 show the results of staining for hematopoietic cell surface antigens in dissociated tumor masses arising after the subcutaneous injection of 10⁶ LL2 cells into C57BL/6 CD45.1 host mice at various time points post-implantation. The composition of hematopoietic cells in tumor stroma differed from that from host peripheral blood (PB) or BM (Table 1), indicating that there exists a unique hematopoietic compartment in tumor stroma. The percentage of hematopoietic cells in the LL2 tumor stroma modestly increased from 20.6% at day 12 to 25.3% at day 19 after tumor implantation (Fig. 1A). The relative composition of myeloid cells (Mac-1⁺ and Gr-1⁺) increased whereas that of lymphoid cells (T cells as Thy1.2⁺ and B cells as B220⁺) and erythroid cells (Ter119⁺) in this compartment decreased over time (Fig. 1A).

We also measured the existence of hematopoietic progenitors and phenotypic HSCs in the tumor stroma. The number of progenitors for granulocytes and monocytes (CFU-GM) was low, with 3.5 per 1,000,000 tumor cells at day 19 post-implantation (Table 1). Interestingly, in the LL2 tumor stroma, we found phenotypic HSCs, measured as CD45⁺Lin⁻Sca-1⁺Kit⁺ cells (Fig. 1B). The flow cytometry analyses indicate that there were few Lin⁻Kit⁺ cells in the tumor stroma, concordant with a previous report [3]. The average frequency of these phenotypic HSCs increased over time from 0.0004% at day 12, to 0.0016% at day 15, and to 0.006% at day 19 (Fig. 1C). In particular, the frequency of phenotypic HSCs detected in the tumor stroma (0.006±0.001%) at day 19 was 1/13 or 1/33 of that in BM of normal mice or tumor-bearing mice respectively (0.08±0.05 and 0.20±0.02% respectively) (Table 1). Since the tumors were perfused, and the cell composition including the percentage of HSCs in tumor stroma was quite different from that in peripheral blood, our result indicates that phenotypic HSCs reside in tumor stroma and are not contaminant from blood.

Since the surface phenotype of HSCs in extramedullary organs can be different from that of BM HSCs [9], we used BM reconstitution analysis, the “gold standard” for measuring HSC repopulating activity, to determine whether functional HSCs existed in the tumor stroma. As we performed before [10], [11], [12], [13], [14], [15], [16], we examined the HSC activity of the donor cells in competitive reconstitution analyses, in which lethally irradiated recipient mice were co-transplanted with both the donor cells to be tested and WT BM competitors. The competitor cells serve as an internal control and as a supply of hematopoietic cells until the transplanted stem cells can generate sufficient mature lymphoid and myeloid cells for survival. Donor and recipient mice are genetically identical except for the CD45 surface protein that is found on nucleated peripheral blood cells and that is not involved in hematopoiesis or stem cell activity; donor cells carried the marker CD45.1, while recipient mice and supportive cells expressed CD45.2. An extremely stringent 18-month competitive reconstitution analysis and a secondary transplantation showed that the tumor stroma contained long-term repopulating HSCs (Fig. 1D). These cells were capable of repopulating both lymphoid and myeloid lineages in long-term reconstitution analysis (Fig. 1E–F), attesting that these are functional HSCs. The relative low repopulation of the donor cells in all lineages compared to BM competitors suggests these HSCs in the tumor stroma have weak engraftment to the recipient BM. Nevertheless, our data, for the first time to our knowledge, demonstrated that functional HSCs exist in tumor stroma.

To determine whether the hematopoietic cells in tumor stroma originated from the BM or from the local environment, we transplanted CD45.1 C57BL/6 donor BM cells into lethally irradiated CD45.2 C57BL/6 mice. At 4 months post-transplant, the recipient BM was completely repopulated by the donor CD45.1 cells, whereas peripheral tissues contained mostly CD45.1 cells with certain CD45.2 cells as reported [9]. We implanted LL2 tumor cells subcutaneously into these mice. Three weeks later, we isolated tumors and used flow cytometry to characterize the hematopoietic cells in stroma. We found that all of the tumor stromal CD45⁺ cells were CD45.1⁺ (Fig. 1G). Accordant with previous studies [3], we concluded that hematopoietic cells in the LL2 stroma are derived from BM.

Next we sought to determine whether the host hematopoietic compartment was affected by tumor growth. Total cellularity and total hemtopoietic cell counts of tumor-bearing mice did not significantly change over time (Fig. 2A–B). However, in both BM and PB of tumor-bearing mice we observed significantly increased percentages of myeloid (Mac-1⁺ and Gr-1⁺) cells but decreased lymphoid (Thy1.2⁺ and B220⁺) and erythroid (Ter119⁺) cells as a function of time post-implantation (Fig. 2C–D). This trend is similar to what occurred in the hematopoietic lineages in tumor stroma. In addition, we analyzed the frequencies and numbers of hematopoietic progenitors and phenotypic HSCs in the tumor-bearing mice and healthy controls. CFU-GM increased approximately 2-fold whereas CFU-E and CFU-Pre-B decreased at least 50% in tumor-bearing mice compared to normal mice (Fig. 2E), concordant with the total increase of myeloid cells and decrease of lymphoid and erythroid cells. The BM and PB enriched phenotypic HSCs as Lin⁻Sca-1⁺Kit⁺CD34⁻Flk-2⁻ or Lin⁻Sca-1⁺Kit⁺ cells increased over time. Numbers of BM HSCs increased about 3-fold and PB HSCs increased 7-fold at day 19 relative to numbers on day 12 (Fig. 2F–G). This suggests that the presence of a tumor induces in vivo expansion of HSCs and progenitors in the BM, which leads to increased HSCs and progenitors in the circulation. Consistent with this result, the tumor-bearing mice display splenomegaly, with 2-fold increase of spleen size and weight compared to healthy controls (data not shown).

Because numbers of HSCs in the tumor-bearing mice significantly increased, we tested whether the cell cycle of HSCs in the host BM changed. As shown in Figure 2H, there was no significant changes in fractions of cells in given cell cycle stages in HSCs in the BM of tumor-bearing mice at day 19 compared to normal mice, suggesting that the tumor did not significantly alter the quiescence of host BM HSCs at the late stage of cancer development.

We developed an assay to compare the abilities of different hematopoietic populations to collaborate functionally with tumor cells to affect the tumor development. In this assay, we use FACS to isolate a certain population of hematopoietic cells and mix them with a fixed number of LL2 cancer cells prior to implantation into the C57BL/6 mice. The kinetics of tumor growth was determined to evaluate the tumor-promoting ability of the co-implanted hematopoietic cells. In the experiment summarized in Figure 3A, we co-implanted 1×10⁵ LL2 cells with 1×10⁴ CD45.1 total BM cells or with enriched normal BM HSCs as Lin⁻Sca-1⁺Kit⁺ cells subcutaneously into C57BL/5 CD45.2 host mice. During the 3-week period of analysis, the tumor size was measured. These data show that both total BM cells and the enriched HSC population positively regulated tumor growth. It is noteworthy that the co-implantation of total BM cells with LL2 cells led to similar level of tumor size as the same number of transplanted LSK cells. Clearly, differentiated hematopoietic cells did promote tumor progression. This suggests that additional HSCs in the tumor local environment do not necessarily further promote tumor growth.

We further determined the effect of hematopoietic cells isolated from the tumor stroma on cancer development. Because it was technically difficult to isolate the low frequent Lin⁻Sca-1⁺Kit⁺ cells from these tumors, we isolated hematopoietic cells from LL2 tumor stroma or host PB as a control and co-injected them with GFP⁺LL2 tumor cells into secondary mice. We found that, although the host PB CD45⁺ cells stimulated LL2 tumor growth compared to LL2 cells alone, CD45⁺ cells isolated from previously existing tumor stroma had significantly increased ability to enhance LL2 tumor growth than these PB CD45⁺ cells (Fig. 3B, p<0.05). This result suggests that 1) hematopoietic cells in tumor stroma enhance tumor growth, 2) hematopoietic cells from tumor stroma are different from those from PB in promoting tumor growth, not unexpectedly as the compositions of these two sources of cells are different (Table 1), and 3) tumor stroma has certain “educating” effect on hematopoietic cells that leads to tumor development.

To further validate the pathological effect of the hematopoietic cells in the tumor stroma, we studied the role of IGF signaling in these hematopoietic compartment. We previously showed that the the receptor for IGF-2 is expressed on the surface of all HSCs [10]. IGF type I receptor (IGF-IR) is the signaling receptor for insulin-like growth factor 1 and 2 (IGF-1 and IGF-2). The IGF pathway has been reported to play important roles in the development of a range of malignancies, including both non-small cell lung cancer and small cell lung cancer (see review [17]). For example, elevated plasma levels of IGF-1 and single nucleotide polymorphisms within the IGF axis are associated with an increased risk of lung cancer [17]. The activation of IGF-IR facilitates malignant transformation and the majority of IGF-2 transgenic mice develop lung cancer by 18 months of age [17]. Of note is that these studies were focused on the activity of IGF-IR that is expressed on cancer cells. We sought to test whether IGF-IR expressed by the hematopoietic stroma plays any role in the development of the LL2 tumors.

Although IGF-IR^−/− mice die after birth [18], [19], we were able to collect IGF-IR^−/− HSCs from the fetal liver. Then we reconstituted recipient mice with IGF-IR^−/− or wild-type fetal liver HSCs. Four months later, the recipients were fully repopulated by the donor IGF-IR^−/− or wild-type HSCs, and we implanted LL2 cancer cells into these mice. The IGF-IR^−/− HSCs had no apparent defect in engrafting the recipient mice (Fig. 4A), nor did they have a noticeable skew in lineage differentiation compared to HSCs from wild-type mice (Fig. 4B). Nevertheless, LL2 tumors grew significantly more slowly in these IGF-IR^−/− HSC reconstituted mice than in mice engrafted with wild-type HSCs (Fig. 4C, p<0.05), accompanying with dramatically decreased metastasis to the lung (Fig. 4D). This experiment suggests that the lack of IGF signaling in the hematopoietic compartment of tumor stroma hampers the solid tumor development. This novel result complements the conventional view that IGF signaling in tumor cells per se is important for cancer development.

All animal experiments were performed with the approval of UT Southwestern Committee on Animal Care (APN# 2007-0068).

Murine Lewis lung carcinoma (LL2) cells were obtained from the ATCC and cultured under standard conditions. Retroviral MSCV-GFP was introduced into LL2 cells to produce stable GFP⁺ LL2 cells. C57BL/6 CD45.2 and CD45.1 mice were purchased from the Jackson Laboratory or the National Cancer Institute. IGF-IR^−/+ mice as previously described [19] were in pure C57BL/6 background. All animals were maintained at the University of Texas Southwestern Medical Center animal facility and animal experiments were performed with the approval of UT Southwestern Committee on Animal Care. Tumor cells were injected subcutaneously into mice and mice were maintained for about 3 weeks. Tumor size was measured on the flanks of live mice using calipers; volume was calculated as (length of tumor)×(width of tumor)²/2. To analyze lung metastasis, entire lungs were harvested and single cell suspensions were prepared by collagenase treatment. GFP⁺ LL2 cells originating from the distant implanted tumor were counted by flow cytometry analysis. The flow cytometry result was confirmed by counting GFP⁺ surface foci of the harvested lung under a dissecting microscope.

Mice were perfused with cold PBS and primary tumors were removed and chopped with a McIlwain Tissue Chopper (Mickle Laboratory Engineering Company). The tissue was washed with PBS and then placed in collagenase-dispase medium (Liver Digest Medium, Invitrogen) at 37°C for 90 min as we described previously [21]. Cells passed through a 70-µm strainer were used for further flow cytometry analysis or sorting.

BM and PB cells were isolated from 5–8 week old C57BL/6 mice. Lin⁻Sca-1⁺Kit⁺ or Lin⁻Sca-1⁺Kit⁺CD34⁻Flk-2⁻ cells were isolated by staining with a biotinylated lineage cocktail (anti-CD3, anti-CD5, anti-B220, anti-Mac-1, anti-Gr-1, anti-Ter119, and anti-7-4; Stem Cell Technologies) followed by streptavidin-PE/Cy5.5, anti-Sca-1-FITC, and anti-Kit-APC, and anti-CD34-PE and anti-Flk-2-PE if necessary. To analyze hematopoietic lineages and repopulation of mouse HSCs, mouse peripheral blood cells were collected by retro-orbital bleeding, followed by lysis of red blood cells and staining with anti-CD45.2-FITC, and anti-CD45.1-PE, and anti-Thy1.2-PE (for T-lymphoid lineage), anti-B220-PE (for B-lymphoid lineage), anti-Mac-1-PE, anti-Gr-1-PE (cells co-staining with anti-Mac-1 and anti-Gr-1 were deemed to be of the myeloid lineage), or anti-Ter119-PE (for erythroid lineage) monoclonal antibodies. All antibodies were from BD Pharmingen. The “percent reconstitution” shown in figures was based on the staining results of anti-CD45.2 and anti-CD45.1. In all cases flow cytometry analysis of hematopoietic lineages was also performed to confirm multilineage reconstitution as we described [14], [15], [16].

CD45⁺ cells from LL2 tumor stroma or normal BM cells were resuspended in IMDM with 2% FBS and were then seeded into methylcellulose medium M3334 (StemCell Technologies) for CFU-E, M3434 (StemCell Technologies) for CFU-GM, or M3630 (StemCell Technologies) for CFU-Pre-B assays, according to the manufacturer's protocols and as described [16].

The cell cycle analysis with Hoechst and pyronin Y staining was performed as described [16]. Briefly, the Lin⁻Sca-1⁺Kit⁺Flk2⁻CD34⁻ cells were collected in Hank's buffered salt solution medium containing 10% FBS, 1 g/liter glucose, and 20 mM Hepes (pH 7.2). Cells were washed, Hoechst 33342 (20 µg/ml, Invitrogen) was added, and cells were incubated at 37°C for 45 min after which pyronin Y (1 µg/ml, Sigma) was added. Cells were incubated for another 15 min at 37°C, washed, and resuspended in cold PBS. Samples were immediately analyzed by flow cytometry (BD Biosciences, FACSAria).

The indicated numbers of mouse CD45.2 or CD45.1 donor cells were mixed with 1 or 2×10⁵ freshly isolated CD45.1 or CD45.2 competitor BM cells, and the mixture were injected intravenously via the retro-orbital route into each of a group of 6–9 week old CD45.1 or CD45.2 mice previously irradiated with a total dose of 10 Gy. For secondary transplantation, CD45.1⁺ cells were collected from primary recipients and 10⁶ cells were injected with 10⁵ CD45.2 competitors into lethally irradiated secondary recipient mice. To measure reconstitution of transplanted mice, peripheral blood was collected at the indicated times post-transplant and the presence of CD45.1⁺ and CD45.2⁺ cells in lymphoid and myeloid compartments were measured [14], [16].

Data are expressed as mean ± SEM. Tumor growth curves for different experimental groups were compared using the Generalized Estimating Equations (GEE) method with AR(1) correlation structure. Tumor sizes among different experimental groups were also compared at each time points using t-test. SAS 9.1.3 was used for the analysis. Data were considered statistically significant if p<0.05.