Promotion of tumor progression induced by continuous low-dose administration of antineoplastic agent gemcitabine or gemcitabine combined with cisplatin.

BACKGROUND AND OBJECTIVES
There are indications that certain antineoplastic agents at low dosages may exhibit abnormal pharmacological actions, such as promoting tumor growth. However, the phenomenon still needs to be further confirmed, and its underlying mechanisms have not yet been fully elucidated.


METHODS
Gemcitabine (GEM) and cisplatin (CDDP) were employed as representative antineoplastic agents to observe effects of continuous low-dose chemotherapy with GEM or GEM combined with CDDP (GEM+CDDP) on tumor formation and growthin xenograft tumor models in vivo. Tumor and endothelial cell functions, apoptosis, cell cycle analysis, as well as bone marrow derived cells (BMDCs) mobilization, were evaluated with transwell, MTT or flow cytometry analysis in vitro, respectively. Histological methods were employed to assess angiogenesis in tumor tissues.


RESULTS
The results showed that tumor formation and growth were both significantly promoted by GEM or GEM+CDDP at as low as half of the metronomic dosages, which were accompanied by enhancements of angiogenesis in tumor tissues and the release of proangiogenic BMDCs in the circulating blood. Additionally, GEM or GEM+CDDP at low concentrations dramatically facilitated the proliferation, migration, and invasion of tumor cells in vitro. Cell-cycle arrest, activation of associated apoptotic proteins, and inhibition of apoptosis were also observed in tumor cells.


CONCLUSIONS
These findings indicate that, the continuous low-dose administration of GEM and GEM+CDDP can promote tumorigenesis and tumor progression in vivo by inhibiting apoptosis, mobilizing BMDCs, and promoting angiogenesis in certain dose ranges. These findings urge further investigations to avoid the potential risks in current empiric continuous low-dose chemotherapy regimens with antineoplastic agents.


MAJOR FINDING
This study observes a previously neglected pharmacological phenomenon and investigates its mechanism of that the continuous low-dose administration of some antineoplastic agents in certain dose ranges can promote tumorigenesis and tumor progression in vitro and in vivo, through stimulation of tumor cell functions directly as well as enhancement of tumor angiogenesis by BMDCs recruitment indirectly. The results alert to a potential risk in current empirically based continuous low-dose chemotherapy regimens such as metronomic chemotherapy.


Promotion of tumor progression induced by
Background Chemotherapy still plays a pivotal role in current cancer treatments (1). However, the conventional chemotherapy with antineoplastic agents at the maximum tolerated dose (MTD) is typically administrated in short cycles, obligatorily separated by long treatment breaks (e.g., 3 weeks), and unfortunately accompanied by serious toxic side effects. The prominent toxicities, or even death, associated with high doses used in these schedules often restrict the ability to increase chemotherapeutic dosages and also impair treatment outcomes (2,3). This situation has led to the introduction and clinical testing of some high-frequency low-dosage regimens, such as metronomic chemotherapy and adaptive therapy to reduce toxicity and obtain optimum outcomes (4). Metronomic chemotherapy, characterized by continuous and dose-dense administration of chemotherapeutic drugs with lowered doses, is usually associated with better tolerance than conventional chemotherapy, and favorable response rates have been reported in various settings (5). Therefore, pre-clinical and clinical studies have been investigating the use of such low dose metronomic therapy as an augmentation or as a substitute for conventional regimens (6). However, numbers of fundamental issues in this eld relating to pharmacology or clinical applications have not yet been clari ed. As a result, the clinical development of metronomic chemotherapy has been impeded by numerous limitations, including the lack of its basic pharmacological data, ambiguity of its de nition and the resulting empiric design of treatment protocols.
The current metronomic chemotherapy protocols applied clinically are exclusively empirical in terms of dosing (e.g., half, one third, one tenth of the maximum tolerated dose) and scheduling (e.g., weekly, twice a week, thrice a week, daily) due to the lack of theory-based dose setting method.
Cancerous cells can lie dormant for years before initiating tumor outgrowth again (7,8). It is generally believed that there are many dormant microtumors in normal tissues. Meanwhile, tumor recurrence and metastasis after chemotherapy represents the main cause of cancer-related mortality (9). Though there is a lifelong risk of recurrence and metastasis, the risk of recurrence typically peaks during the rst 5 years after treatment, and especially during the rst 2 years (10). Cancer can recur in the same place as the original tumor or in other places in the body if the tumor cells spread. Moreover, it has been found that the recurrent tumor tissues may not be homologous with the primary tissue, suggesting that recurring tumors are more likely to derive from independent sources such as dormant microtumors (11). However, little is known about what triggers steady state original tumor cells to reawaken, as well as the causative factors or causes of these dormant non-homologous microtumor progression.
Antineoplastic agents at certain low dosages have been found to enhance tumor development in some animal models (12)(13)(14)(15)(16)(17). For example, low-dose bleomycin was reported to promote tumor metastasis and low-dose cyclophosphamide was found to enhance both tumor growth and metastasis in mice (12,18). It was also found that antineoplastic agents are carcinogenic even at very low doses (19,20). Healthcare workers exposed to antineoplastic agents polluted the environment suffer signi cantly increased fetal malformations, chromosomal abnormalities, and tumor occurrence (21,22). Moreover, studies have revealed that low-concentration antineoplastic agents can enhance the proliferation, adhesion, and migration of endothelial cells, suggesting that these agents promoting tumor angiogenesis and tumor progression (23).
Therefore, the question that naturally arises is whether a tumor's relapse is via activation of dormant microtumors by chemotherapy. Therefore, it is necessary to determine whether inappropriate chemotherapy regimens can lead to adverse treatment outcomes. For instance, the potential role of lowdose antineoplastic agents in tumor recurrence and metastasis.
GEM is powerful antineoplastic drug that is used for pancreatic cancer and breast cancer chemotherapy, either alone or in combination with several types of cytotoxic drugs [e.g., CDDP and paclitaxel], and is also an option in various other solid and hematological cancers (24,25). Importantly, GEM is one of the most commonly used antineoplastic agents in metronomic chemotherapy for solid cancers such as breast, liver and pancreatic cancer (26). However, very few studies have investigated the dose effect relationship of GEM and its combined chemotherapy under continuous low-dose administration conditions such as metronomic chemotherapy. Therefore, in the present study, GEM and GEM plus CDDP were selected as the observation objects to investigate the effects of continuous low-dose administration of antineoplastic agents on tumorigenesis and tumor progression in vivo. The pharmacological effects of low-concentration GEM or GEM + CDDP on the functions and apoptosis of tumor and endothelial cells were also investigated in vitro.

Material And Methods
Chemicals and antibodies To mimic chemotherapy, GEM or GEM + CDDP was administrated 24h after tumor xenograft and repeated every 48h. 0.9% sodium chloride solution was delivered parallelly as control. Each administration was divided into two 0.2mL injections with drug or saline via the abdominal route. Tumor volumes were determined every 2 days in C57BL/6J mice and every 5 days in BALB/C nude mice with the volumetric calculation formula: length×width 2 ×0.5. At each time point, mice were anesthetized with 20% urethane, and blood was collected from the vena cava with heparin anticoagulation. Tumors were harvested by detaching the surrounding connective tissue 2 weeks in C57BL/6J mice or 6 weeks in BALB/C nude mice post-implantation, respectively by detaching the surrounding connective tissue, weighed, and processed for immunohistochemistry.

Collection and isolation of bone marrow cells
Mice were anesthetized with urethane and then euthanized immediately. Bone marrow was extracted from the femurs and tibias of mice. A single bone marrow cell suspension was prepared by passing the bone marrow through a 21-gauge needle. Then the cells were pelleted by centrifugation and suspended in medium for bone marrow transplantation in further experiments at a density of 1×10 7 cells/mL.

Bone marrow transplantation
Recipient mice (8 weeks old) were lethally irradiated (9Gy) followed by bone marrow reconstitution by tail vein injection with 1×10 7 bone marrow cells isolated from green uorescent protein-positive (GFP + ) donor femurs. 8 weeks after bone marrow transplantation, the mice were used for tumor experiments.

MTT assay
Cells (3×10 4 /mL) were seeded in 96-well plates. 20µL of MTT (5mg/mL) was added into each well and incubated for 4h. Then, 150µL of DMSO was added into each well to solubilize formazan for 15min. The absorbance was determined at 490nm on a microplate reader (BioTek). Each experiment was repeated three times.
Clone assay 1×10 2 cells were seeded in triplicate wells of 6-well plate (n = 3) and cultured for 2 weeks. After xed with 4% paraformaldehyde for 20min, the colonies were stained using 0.1% crystal violet and counted.

Transwell migration and invasion assay
Cell migration and invasion were assessed using 8-µm-pore transwell compartments (Corning). 5×10 5 cells were suspended in serum-free medium in the upper compartment. The translocated cells were stained with 0.5% crystal violet for 20 min at 25℃ after cells were incubated at 37℃ for 24h. For invasion assays, Matrigel Matrix (BD Biosciences) was added to each well according to the manufacturer's instructions before 5×10 5 cells were suspended in the upper compartment. After cells were incubated at 37℃ for 24h, the translocated cells were incubated with 0.5% crystal violet for 20min at 25℃, and were counted under an upright microscope ( ve elds each chamber). Each assay was repeated in three independent experiments.

Wound-healing assay
Cells were plated in triplicate into 6-well plates. A standard 10µL pipette tip was used to scratch wound when the cells reached a density of 95%. Subsequently, the cells were cultured in FBS-free medium or treated with drugs. After 24h, the wound closure was captured by a microscope and calculated using the software of Image J (National Institutes of Health, Bethesda, MD, USA).

Cell cycle analysis
Cells were harvested 24h after treatments and xed in ice-cold 75% ethanol overnight at 4°C. Cells were treated with PI staining Kit according to the manufacturer's instructions after washing with PBS for two times, and then analyzed by FACS-Canto II ow cytometry (Becton Dickinson Company).

Apoptosis analysis
Cells were harvested 48h after treatments, washed and resuspended with cold PBS. The staining process was then conducted using the Annexin V-FITC/PI Apoptosis Detection Kit. The stained cells were analyzed using FACS-Canto II ow cytometry (Becton Dickinson Company).

Western blotting analysis
Cells were lysed with RIPA buffer containing protease inhibitor on ice. Then equal amounts of protein lysates were electrophoretically separated by 10% SDS-PAGE and transferred to polyvinylidene uoride membranes (Millipore). After blocking with 5% nonfat dried milk for two hours, the membranes were incubated with primary antibodies overnight at 4℃. Another incubation with a horseradish peroxidaseconjugated secondary antibody was performed in the following day for two hours at 37℃, after when the protein bands were detected using the Super Signal West Pico kit (Thermo Fisher Scienti c). All Western blot experiments were repeated at least three times.

Histology and immunohistochemistry
Tissues were embedded in OCT and snap frozen in liquid nitrogen or xed in 10% buffered formalin followed by para n embedding. IHC staining was applied to 3-µm thick sections of para n-embedded tissue specimens with a PV-9001 Detection Kit (ZSGB-BIO). Brie y, formalin-xed para n-embedded tissues were depara nized in xylene for 10min and rehydrated in a graded series of ethanol solutions. The tissues were immersed in 0.01M citric acid buffer at 121℃ for 5min and then cooled and washed with 0.1M PBS for 3 times. After treated with 3% hydrogen peroxide for 10min, the tissues were incubated with primary antibodies overnight at 4℃, followed by the secondary antibody for 30min at 25℃. Sections were stained with 3,3'-diaminobenzidine and counterstained using hematoxylin for 5 seconds, dehydrated in a graded series of ethanol solutions, immersed in xylene, and examined with a microscope (Axio Image M2, Zeiss). The microvascular area (MVA) in IHC stained images of each group was analyzed with Image-Pro Plus software.

Flow cytometry
Cells isolated from circulating blood were incubated with uorescein isothiocyanate (FITC) anti-mouse CD11b antibody, PE anti-mouse Gr-1, and PE anti-mouse CD61, following a standard laboratory protocol (27). The cells were subjected to ow cytometer on FACS-Canto II ow cytometry (Becton Dickinson Company), and data were analyzed with Cell Quest Software.

Statistical analysis
Statistical analysis was performed using SPSS software. Data were expressed as mean ± standard deviation values. One-way ANOVA or Fisher's exact test was used for comparison tests. Probability values of P < 0.05 were considered statistically signi cant.

Results
Continuous low-dose administration of GEM and/or CDDP facilitated tumor formation and promoted tumor growth in vivo Surprisingly, we found that continuous low-dose administration of GEM and GEM + CDDP signi cantly promoted tumor growth in xenograft tumor models in vivo.
As shown in Fig. 1a in the 5 mg/kg GEM and 4.8/0.24 mg/kg GEM + CDDP groups after 2 weeks treatment were reduced by 36.6% and 39.1%, respectively, relative to control (both p < 0.05). Strikingly, in MCF-7 mediated nude mouse tumor model, out of six tested mice, we found some tumors (from 1 to 4) developed from either GEM or GEM + CDDP low-dose treatment groups but not in the control group 6 weeks after implanting equal number of MCF-7 tumor cells (Fig. 1d). Together these ndings indicate that the continuous lowdose administration of GEM and GEM + CDDP can promote signi cant tumor formation and growth in vivo. We also performed the colony formation and wound healing assays. As shown in Fig. 2d, colony formation was markedly enhanced in B16 cells at 1×10 − 3 µM GEM and 1×10 − 3 /5×10 − 5 µM GEM + CDDP groups relative to control (both p < 0.01). Similar results were seen in the wound-healing assay showing that low-concentration 1×10 − 3 µM GEM or 1×10 − 3 /5×10 − 5 µM GEM + CDDP promoted the proliferation of B16 cells (both p < 0.01) (Fig. 2e). In addition, the cell-cycle assay revealed that 1×10 − 3 µM GEM and/or 5×10 − 5 µM CDDP markedly decreased the fraction of MCF-7 and B16 cells at the G2 and M phases of the cell cycle, while 1×10 − 5 µM GEM and/or 5×10 − 7 µM CDDP decreased the fraction of T-47D cells at the G0 and G1 phases (each p < 0.01) (Fig. 2i-k).
Those ndings strongly suggest that the tumor growth acceleration induced by low-concentration GEM and GEM + CDDP is associated with inhibition of the expression of proapoptotic proteins and promotion of the expression of antiapoptotic proteins.
Low-dose GEM and GEM + CDDP promoted tumor angiogenesis in vivo, and inhibited the proliferation of endothelial cells in vitro As shown in Fig. 4a-b, the areas of CD31 + vessels in 1.25 mg/kg GEM and 0.6/0.03 mg/kg GEM + CDDP groups were signi cantly enhanced in B16 tumor tissues relative to control (both p < 0.01), as was also the case in laminin positive vessels (both p < 0.01). This implies that GEM and GEM + CDDP can promote tumor angiogenesis under continuous low-dose administration conditions. However, we found that GEM and GEM + CDDP inhibited endothelial cells proliferation in the concentration range from 1×10 − 2 /5×10 − 4 µM to 10/0.5 µM in vitro (each p < 0.01), with no stimulation role of GEM or GEM + CDDP found at lower concentrations (Fig. 4c).
These results suggest that GEM and GEM + CDDP can promote angiogenesis in B16 tumor tissues at lowdosage via pathways independent of endothelial cells stimulation.
These results indicate that continuous low-dose administration of GEM and GEM + CDDP can signi cantly mobilize proangiogenic Gr-1 + CD11b + and CD61 + BMDCs into the circulating blood in tumorbearing mice models.

Continuous low-dose administration of GEM and GEM + CDDP promoted the recruitment of BMDCs in tumor tissues
Tumor growth in C57BL/6J mice with GFP bone marrow was observed after 4-weeks of treatment with frequency of once every 2 days with antineoplastic agents. Tumor weights in the 1.25 mg/kg GEM and 0.6/0.03 mg/kg GEM + CDDP groups were increased by 342.5% and 344.9%, respectively, relative to control (both p < 0.01), while they were decreased by 61.8% and 12.9% in the 5 mg/kg GEM and 4.8/0.24 mg/kg GEM + CDDP groups, respectively (Fig. 6a).
The densities of GFP + BMDCs were analyzed using immunohistochemically of tumor tissue sections.
Compared with the control group, the numbers of GFP + cells were increased by 185.8% and 227.8% after treatment with 1.25 mg/kg GEM and 0.6/0.03 mg/kg GEM + CDDP, respectively, compared to control group (both p < 0.01) (Fig. 6b).
These results indicate that the continuous low-dose administration of GEM and GEM + CDDP can promote the recruitment of GFP + BMDCs in tumor tissues.
Continuous low-dose administration of GEM and GEM + CDDP promoted the expression of proangiogenic proteins in tumor tissues To con rm the role of low-dose GEM and GEM + CDDP in promoting angiogenesis, we also surveyed the expression changes of some angiogenic proteins in B16 tumor tissues by Immunoblotting. As shown in Fig. 7, the protein expression levels of MMP-9, VE-cadherin, VEGFR 1 , and VEGFR 2 in the 1.25 mg/kg GEM group were increased by 66.6%, 124.3%, 87.6%, and 54.5%, respectively, relative to control (each p < 0.01).
These results suggest that the continuous low-dose administration of GEM and GEM + CDDP can promote the expression of proangiogenic proteins.
ATRA inhibited enhancement of B16 tumor growth and mobilization of BMDCs induced by continuous low-dose administration of GEM As shown in Fig. 8a (Fig. 8b). Meanwhile, as shown in Fig. 8c, the areas of laminin positive vessels in 1.25 mg/kg GEM were signi cantly enhanced and decreased in 30 mg/kg ATRA in B16 tumor tissues relative to control (both p < 0.01).
These results suggest that low-dose GEM induces mobilization of proangiogenic BMDCs.

Discussion
Low-dose chemotherapy such as metronomic chemotherapy with main advantages of no prolonged drug-free breaks, the potential for delayed development of resistance, low toxicity pro le, and convenient use makes it a desirable alternative in clinical applications to overcome the shortcomings of e cacy and toxicity in conventional chemotherapy (28,29). Nevertheless, owing to the investigational nature of this approach, its regimens are highly empirical in terms of the optimal dose and schedule for the drugs administered. Therefore, greater knowledge of the dose-effect relationship of metronomic chemotherapy is critical to the success of this treatment strategy. However, few studies have focused on the doseresponse relationships of low-dose antineoplastic agents, even though clues had shown anomalous pharmacological actions in certain antineoplastic agents under such conditions (30). In present study, both in vitro and in vivo data suggest that inappropriate antineoplastic agents' dosage and treatment regimen setting may lead to opposite therapeutic outcomes. Therefore, inappropriate low-dose chemotherapy regimens may lead to the risk of adverse outcome, which is often ignored, especially under the condition of empirical setting of treatment. It should be noted that the results of present study do not negate the low-dose chemotherapy strategies, but call for more dose-response relationship studies and mechanisms understanding of various antineoplastic agents.
Few studies by our groups and others (16) have implied that antineoplastic agents might promote or induce tumor growth under certain low-dose conditions, these phenomena were thus investigated in vivo in the present study with lower dose GEM and/or CDDP. GEM is applied to solid tumors with common murine in vivo single dose of 100-120mg/kg and clinical dose of 25 mg/kg (1000 mg/m 2 ) (31,32). Numerous treatment schedules of metronomic GEM have been applied in mice, from continuously 1 mg/kg/day for 28 days to continuous 3.3 mg/kg/day for 21 days (31,(33)(34)(35). In order to investigate the aberrant dose-effect relationship of low-dose GEM and tumor growth, as well as considering the current status of empirical dose setting for metronomic chemotherapy in clinic, effects of GEM on tumor growth were observed in the present study at dose ranges below or above the reported dose of metronomic chemotherapy (36, 37). As a result, the single GEM dosage in continuous low-dose administration in vivo were set as about half to twice these doses in the mice model in the present study. The concentrations of promoting tumor cell function in the culture medium were selected as the GEM experiment concentrations in vitro (Fig. 2a). The dose ratio of GEM to CDDP was set according to the reported clinical dosages (38).
The results veri ed that GEM could exert inhibition effects at high-doses (equivalent to one to two times of the metronomic doses every 2 days) or promotion effects at low-doses (equivalent to half of the metronomic doses every 2 days) on tumor growth in tumor-bearing mice model. A similar biphasic effect dose-response relationship was observed in GEM + CDDP groups in vivo.
The present study also found that the antineoplastic agents GEM either alone or in combination with CDDP stimulated the functions and inhibited the apoptosis of tumor cells at low concentrations, indicating a direct role in promoting tumor growth. Previous studies have reported that CDDP enhanced the adhesion of endothelial cells and thereby might promote tumor angiogenesis (23). Nevertheless, no signi cant promotion effects of CDDP alone on either tumor growth in vivo or tumor cell functions in vitro, as well as endothelial cells proliferation, were observed at the selected low dosages or concentrations in the present study. However, when CDDP was applied in combination with GEM at low dosages in vivo or low concentrations in vitro, it demonstrated an enhancement effects on tumor cell proliferation and survival than GEM alone. These results suggest that combination of antineoplastic agents may shift the drugs' optimal concentration and a completely adverse outcome. Therefore, the present experimental results strongly support that dosage setting in chemotherapy regimen should be based on precise sources and bases, so as to ensure dosages and schedules within the therapeutic window. Thus, it could avoid inconsistent or opposite therapeutic effects of empiric metronomic chemotherapy with the treatment goals.
It is widely recognized that dormant microtumors are present in normal tissues, which may be activated and result in cancer progression under certain stimulation conditions (7,8). Moreover, studies have often found that genotypes of recurring tumors might differ from that of the primary tumor, implying that a heterogeneous tumor probably originated from sources other than primary tumor (11). When nude mice were subcutaneously inoculated with 1×10 5 MCF-7 tumor cells, the number of cells required was slightly smaller than normal. Then stimulation with low-dose GEM and/or CDDP for 6 weeks distinctly increased the numbers of tumor formation from zero (out of six) in the control group to two in the GEM group and four in the GEM + CDDP group. These results indicate that low-dose antineoplastic agents might promote the formation and progression of residual primary tumor or dormant microtumor. Given the fact that occupational exposure to low-dose antineoplastic agents increases the incidence of tumors, these results consequently support the speculation that that improper low-dose chemotherapy regimens lead to the activation of dormant microtumors by antineoplastic agents, which may promote tumor recurrence and metastasis. Similar phenomena have been reported in tumor radiotherapy, in which low-dose radiotherapy promoted tumor recurrence (39). The present data suggest that there is an urgent need to further investigate the pharmacological responses of antineoplastic agents in clinical applications, as well as the role of antineoplastic agents in tumor relapse, in continuous low-dose regimens such as metronomic chemotherapy.
Angiogenesis plays a key role in the development of solid tumors (40). It has been demonstrated that some antineoplastic agents can directly promote the proliferation, adhesion, and migration of endothelial cells at certain concentrations in vitro, thereby promoting tumor angiogenesis (23). The present study found that antineoplastic agents GEM and CDDP either alone or in combination inhibited endothelial cell functions in vitro and stimulated angiogenesis in vivo. Tumor angiogenesis is a complex multi-step process regulated by various factors and orchestrated by a number of intersecting pathways (41). Thus, even though the endothelial-cell functions were suppressed by antineoplastic agents directly in the present study, the ultimate outcome of tumor angiogenesis still might be enhanced through an indirect pathway. We previously found that BMDCs recruited into angiogenic local tissues played a dominant role in promoting angiogenesis (42,43). There is also evidence that antineoplastic agents stimulate the mobilization of proangiogenic BMDCs and their release from bone marrow, and enhance the recruitment of BMDCs in tumor tissues (14,44). The retention of proangiogenic BMDCs in local tissue can then promote tumor angiogenesis via the release of stimulating factors (43). The cytometer ow analysis results obtained in this study showed that the antineoplastic agents GEM and GEM + CDDP at low doses signi cantly increased the circulating level of proangiogenic Gr-1 + CD11b + and CD61 + BMDCs. The densities of GFP + BMDCs in tumor tissue sections obtained from mice with bone-marrow transplantation were higher for treatments with GEM and GEM + CDDP than in the control group, indicating that low-dose GEM and GEM + CDDP can promote the recruitment and retention of BMDCs in tumor microenvironment tissues. The expression of proangiogenic receptors such as VEGFR 1 and VEGFR 2 and proangiogenic proteins such as MMP-9 and VE-cadherin in tumor tissues also were found to be promoted by low-dose GEM and GEM + CDDP, which suggested an enhancement role of theses antineoplastic agents in angiogenesis, too. Thus, the promotion of tumor angiogenesis by low-dose GEM or GEM + CDDP might be associated with the recruitment of proangiogenic Gr-1 + CD11b + and CD61 + BMDCs and changes in the tumor microenvironment indirectly. It is speculated that occupational injuries of cancer incidence increase caused by antineoplastic agents may share similar mechanisms.
Previous studies have shown that ATRA combined with uorouracil can signi cantly inhibit tumor growth in vivo (45). It was also reported that ATRA induced immature myeloid cells to differentiate into mature dendritic cells, macrophages, and granulocytes, as well as greatly reduced the number of Gr-1 + CD11b + BMDCs in tumor tissues (46,47). The present study found that 30 mg/kg ATRA markedly inhibited the enhancement tumor growth induced by 1.25 mg/kg GEM and decreased the release of Gr-1 + CD11b + and CD61 + BMDCs in the circulating blood. These ndings suggest that ATRA can inhibit tumor growth by reducing the mobilization of proangiogenic BMDCs, thereby attenuating low-dose GEM's effect on promoting tumor growth.
It should be noted that Gr-1 + CD11b + is also a marker for myeloid-derived suppressor cells, which represent a heterogeneous population of immature myeloid cells capable of modulating immune responses (48, 49). Myeloid-derived suppressor cells are mobilized and recruited to tumor tissues to aid in establishing an immunosuppressive tumor microenvironment that makes it easier for tumor cells to avoid immunological detection (50,51). This means that the promotion of tumor development and growth by GEM and/or CDDP might also be associated with immunosuppression of the tumor microenvironment, further research is needed into this.
Antineoplastic agents are usually applied in combination chemotherapy to increase therapeutic effect and reduce the toxic side effects (52, 53). GEM as a cell-cycle-speci c drug that is usually combined with the nonspeci c cell-cycle drug CDDP administration in breast cancer, to achieve a synergistic effect (54). The data obtained in the present study have shown that the promotion of tumor development and growth by GEM under continuous low-dose administration was also presented or even enhanced in the GEM + CDDP group in vivo. The enhancement of tumor cell functions such as proliferation, migration, and invasion were found in the GEM groups, and further ampli ed in the GEM + CDDP group in vitro. In addition, low concentrations of GEM + CDDP resulted in the increased mobilization of proangiogenic BMDCs and expression of proangiogenic protein, decreased proapoptotic factors, and increased the expression levels of antiapoptotic factors, resulting in the enhancement of angiogenesis and inhibition of tumor-cell apoptosis. Thus, combining low-dose antineoplastic agents might still result in similar or even greater facilitation of tumor formation or growth compared with when GEM is used alone. This aspect needs to be further explored due to its potential for widespread clinical applications.

Conclusions
Together, the present ndings suggest that GEM or GEM + CDDP can facilitate tumor development and     Effects of continuous low-dose administration of GEM and GEM+CDDP once every 2 days on mobilization of Gr-1+CD11b+ and CD61+BMDCs in circulating blood of mice. a Gr-1+ CD11b+BMDCs counts in the circulating blood of MCF-7 tumor bearing nude mice treated with GEM and/or CDDP. b-c Gr-1+CD11b+BMDCs and CD61+BMDCs counts in the circulating blood of B16 tumor bearing C57BL/6J mice treated with GEM for 2 weeks. d-e Gr-1+CD11b+BMDCs and CD61+BMDCs counts in the circulating blood of B16 tumor bearing C57BL/6J mice treated with GEM+CDDP for 2 weeks. f B16 tumor volume and weight of subcutaneous xenografts in mice with bone-marrow transplantation and GFP+ BMDCS in tumor tissue. Scale bar,100 μm.*, p < 0.05; **, p < 0.01.

Figure 6
Effects of continuous low-dose administration of GEM and GEM+CDDP once every 2 days for 4 weeks on recruitment of GFP+BMDCs in B16 tumor tissues of C57BL/6J mice with GFP+ bone marrow. a B16 tumor volume and weight of subcutaneous xenografts in mice with bone-marrow transplantation. b GFP+ BMDCS in tumor tissue. Scale bar,100 μm. *, p < 0.05; **, p < 0.01 vs control.

Figure 8
Effects of ATRA on the enhancement of B16 tumor growth and mobilization of proangiogenic BMDCs induced by continuous low-dose administration of GEM in C57BL/6J mice. a B16 tumor weights and volumes in C57BL/6J mice. b Gr-1+CD11b+ and CD61+BMDCs counts in the circulating blood of C57BL/6J mice treated with GEM and GEM combined with ATRA. c Effects of GEM and GEM combined with ATRA on micro-vessel of B16 (Laminin). Scale bar, 100μm*, p < 0.05; **, p < 0.01.