Purpose: Tumor metastasis continues to be the major obstacle to cancer therapy and the leading cause of cancer-related death. Methods used to detect metastasis, especially occult metastases, have received a great deal attention. In this study, a novel selective peptide was assessed for its specific binding to metastasis.

Methods: The FliTrx bacterial peptide display system, an alternative to phage peptide display, was used to identify a 5-amino acid peptide termed TMTP1 (NVVRQ), which binds to the highly metastatic prostate cancer cell line PC-3M-1E8. The synthetic TMTP1 was tested in vitro for its binding specificity and affinity to highly metastatic cancer cells. The tumor targeting assays were done in vivo by i.v. injection of FITC-conjugated TMTP1 into tumor-bearing mice.

Results: TMTP1 specifically bound to a series of highly metastatic tumor cells, including prostate cancer PC-3M-1E8, breast cancer MDA-MB-435S, lung cancer PG-BE1, and gastric cancer MKN-45sci, in vitro and in vivo but not to the poorly metastatic or nonmetastatic cell line, including prostate cancer PC-3M-2B4, breast cancer MCF-7, lung cancer PG-LH7, or murine fibroblast cell NIH/3T3. FITC-TMTP1 strongly and specifically targeted the metastasis foci in tumor-bearing mice 24 h after i.v. peptide injection. Moreover, the occult metastases were specifically detected by FITC-TMTP1.

Conclusion: Our results suggest that TMTP1 is a potential strategy for the development of new diagnostic tracers or alternative anticancer agents for tumor metastasis.

Metastasis is responsible for most therapeutic failures in cancer treatment and leads to death in most cancer patients. Once cancer is diagnosed, it is important to know whether the disease is confined to the primary site or has spread either regionally or systemically (1). At initial diagnosis, approximately 70% of cancer patients can be cured surgical removal of tumors. Adjuvant radiation therapy and/or chemotherapy are beneficial for cancer patients postoperation (25). However, a proportion of patients with no evident systemic dissemination will develop recurrent disease or metastasis after “curative” therapy (6, 7). In these cases, the cancer has clearly undergone occult systemic spread and is undetectable by routine methods including careful clinical, pathologic, biochemical, and radiologic evaluation. Sometimes, the metastatic cancer cells form microfoci (<2 mm) and maintain a balance between proliferation and apoptosis without evoking clinical symptoms (811). Evasion of the host immune system by these tumor cells leads to systemic spread to tissues and eventually ruin the host (5, 1214). The 5-year survival rate in patients with recurrence or metastasis is <50% (15, 16). Although many new treatment modalities have been developed, they seldom improve the survival rate of patients with cancer metastasis (1720).

Diagnostic approaches for micrometastasis are essential to controlling metastasis. Heterogeneous tumor cells can be divided into clones with highly or poorly metastatic potential. Highly metastatic cancer cells, which are prone to spreading to other tissues and forming metastases, express unique molecular markers that distinguish them from cancer cells with poorly metastatic or nonmetastatic potential. These markers are potentially useful as diagnostic markers and therapeutic targets (2125). However, it is a major obstruction to detect such specific markers of metastasis. Peptide display technology has been used to study protein-protein interaction for over 20 years. The strength of this technology is the ability to identify interactive proteins and other molecules without prior knowledge of the nature of the interaction (26). The identified “homing” peptides are promising alternatives to currently used biomolecules for targeting metastatic cells because of their rapid blood clearance, increased diffusion and tissue penetration, nonimmunogenic nature, and ease of synthesis (2731).

In the present study, we used the bacterial peptide display system, FliTrx, an alternative to phage peptide display, to identify a new peptide that binds to highly metastatic cancer cells. In the FliTrx system (Invitrogen), a phagemid vector (pFliTrx) allows the display of peptides directly on the surface of Escherichia coli by using two proteins: the major bacterial flagellar protein (FliC) and thioredoxin (TrxA). Peptide libraries are cloned in frame within the active site loop of thioredoxin, which is inserted into the dispensable region of the FliC gene. The resulting fusion protein assembles into the flagella protruding from the bacterial cell surface. The dodecamer peptide library is constrained by a disulfide bridge, which results in a stable conformation for the synthetic peptide, and is resistant to degradation (32, 33). In vitro screening of the FliTrx library with the highly metastatic prostate cancer cell line PC-3M-1E8 derived a 5-amino acid peptide, NVVRQ (termed TMTP1). We carefully assessed the accuracy and specificity of TMTP1 binding to a series of cell lines of various metastatic potential and its targeting to the metastatic foci.

Cell lines. The paired highly metastatic and nonmetastatic human prostate cancer cell lines PC-3M-1E8 and PC-3M-2B4 and human lung cancer cell lines PG-BE1 and PG-LH7, as shown in Supplementary Table S1 (34), were kindly provided by Dr. Jie Zheng (Beijing University; refs. 35, 36). The human breast cancer cell lines MDA-MB-435S, MDA-MB-231, and MCF-7 and the murine fibroblast cell line NIH/3T3 were obtained from the American Type Culture Collection. All cell lines were maintained in complete RPMI 1640 containing 10% fetal bovine serum.

Construction of mouse models. Four-week-old BALB/c nu/nu mice were obtained from the SLAC Laboratory Animal Co. and injected s.c. into the posterior trunk with a single dose of 2 × 106 tumor cells to induce tumor (PC-3M-1E8, PC-3M-2B4, and MDA-MB-435S cell lines). Three weeks after tumor cell injection, the mice were euthanized to harvest the tumor fragments. Fresh tumor fragments (∼2 mm3) were then implanted s.c. into the posterior trunk of anesthetized mice.

The mouse model of MKN-45sci orthotopic gastric cancer, which has liver-specific metastasis potential, was kindly provided by Dr. Jinjun Li (Shanghai Cancer Institute, Medical College of Shanghai Jiao Tong University; refs. 37, 38). Fresh tumor fragments were obtained as described above. The stomach of the anesthetized mouse was exposed, and part of the serosal membrane was scraped with a forceps. One 1-mm3 tumor piece was fixed on the scraped site of the serosal surface with a 5-0 absorbent suture. The stomach was then returned to the peritoneal cavity, and the abdominal wall and skin were closed with 1-0 sutures.

The mouse model of MDA-MB-435S orthotopic breast cancer was induced by injection of a single dose of 2 × 106 tumor cells to the mammary fat pad.

Peptide library screening and FliTrx clones binding. A random 12-mer cyclic peptide display library, FliTrx (Invitrogen), was screened. Tumor-targeting FliTrx clones were isolated from the FliTrx library by combined in vitro screening according to the manufacturer's protocol. Quantification of binding selectivity was determined by counting the bacterial clones. The specific binding capacity of TMTP1 was confirmed by cross-inhibition experiments.

Peptide design and synthesis. The TMTP1 peptide (NVVRQ) and the control peptide, a scrambled variant (VNQRV, named svTMTP1) whose amino acid composition resembles that of TMTP1, were designed with the general structure GCGXXXXXGC (XXXXX = variable residue, C = cysteine, and G = glycine), which is flanked by two cysteine residues to allow for disulfide linkage and loop formation. Glycine residues provide a spacer function, permitting the physical formation of a loop structure. FITC was coupled to the peptides via an additional glycine at the NH2 terminus. The peptides were synthesized using Fmoc chemistry in a solid-phase synthesizer by Xi'an Huachen Bio-tech. They were purified by high-performance liquid chromatography, and their sequence and structure were confirmed by mass spectrometry.

Similarity searches. BLAST searches1

were done by using the Swiss-prot database to determine if TMTP1 displayed sequence similarity with previously identified human peptides or proteins.

Tumor targeting. Tumor-bearing mice were used for targeting experiments when the tumors had grown to a size of 1.0 to 1.5 cm3. FITC-TMTP1 (210 μg) was injected into the tail vein and allowed to circulate for 24 h. The mice were perfused with PBS through the left ventricle to remove blood and unbound peptides. Tumors and control organs were excised, and frozen sections were prepared and examined for fluorescence by laser scanning confocal microscopy (Olympus Fluoview FV1000). Quantification of the imaging results was accomplished by using Image-Pro Plus 5.1 (Media Cybernetics).

Real-time fluorescent whole-body imaging. MKN-45sci tumor-bearing mice were received 210 μg FITC-TMTP1 peptide i.v. at day 21 after orthotopic transplantation. Two hours after the injection, whole-body imaging was done by using a whole-body optical imaging system (39, 40). Selective excitation of FITC was produced through a D455/70 nm band-pass excitation filter (HB Optical). Emitted fluorescence was collected through a long-pass filter (520 nm; HB Optical) on a 3.3 MP digital camera (DC290; Kodak).

Identification of a peptide that specifically binds to a highly metastatic prostate cancer cell line. Targeting peptides were selected in vitro by screening of the FliTrx library with the highly metastatic prostate carcinoma cell line PC-3M-1E8 for four rounds. For each round, 1 × 108 FliTrx cells were added to a cell culture dish with the nonmetastatic human prostate cancer cell line, PC-3M-2B4, for negative selection, and then incubated for 1 h with PC-3M-1E8 cells for positive selection. After four rounds, 100 individual FliTrx clones were selected and their peptide-encoding inserts were sequenced and analyzed for potential repetitive peptide motifs. One of the peptide sequences enriched in the selected clone pool was NVVRQ, which we named TMTP1. The computer-generated mimic models and amino acid sequence of TMTP1 are shown in Fig. 1A to C. An in vitro binding experiment showed that TMTP1-displaying FliTrx clone 27 or 50 specifically bound to PC-3M-1E8 cells 12.6 or 11.7 times over control PC-3M-2B4 cells, respectively (Fig. 1D). A synthetic TMTP1 with the sequence GCGNVVRQGC was used in the competitive inhibition assays and completely inhibited specific binding of the clones to PC-3M-1E8 cells (Fig. 1D).

TMTP1 specifically binds to highly metastatic tumor cells in vitro. To determine whether TMTP1 peptide binds to highly metastatic tumor cells in vitro, FITC-conjugated TMTP1 was used to determine the specific binding in a series of tumor cell lines. Fluorescein imaging revealed that TMTP1 bound to all highly metastatic tumor cells tested, including prostate cancer PC-3M-1E8, breast cancer MDA-MB-435S, lung cancer PG-BE1, and gastric cancer MKN-45sci (Fig. 1E-H). By contrast, TMTP1 did not bind to the poorly metastatic or nonmetastatic tumor cells, including prostate cancer PC-3M-2B4, breast cancer MCF-7, lung cancer PG-LH7, or murine fibroblast NIH/3T3 cells (Fig. 1I-L). svTMTP1, whose amino acid composition resembles that of TMTP1, was synthesized as a control to identify the specific binding characteristics of TMTP1. FITC-conjugated svTMTP1 showed no positive binding to any highly metastatic tumor cells (Fig. 1M-P). Flow cytometric analysis further confirmed the specific binding of TMTP1 to highly metastatic PC-3M-1E8 and MDA-MB-435S tumor cells (Fig. 1Q and R). Similarly, the lower E-cadherin expression levels were detected by flow cytometry in both PC-3M-1E8 and MDA-MB-435S cells compared with the paired poorly metastatic PC-3M-2B4 and MCF-7 cells (Fig. 1Q and R).

TMTP1 binding location to tumor cells was time dependent. Confocal images showed that FITC-TMTP1 was bound to PC-3M-1E8 cell membranes 10 min after incubation (Fig. 2A) and was gradually internalized into the cytoplasm after 30 min (Fig. 2B). After 6 h incubation, strong fluorescein staining was detected in the cytoplasm, and fluorescein-labeled peptide was partially apparent in the nucleoplasm (Fig. 2C). No positive staining was detected in PC-3M-2B4 cells (Fig. 2D).

TMTP1 targets highly metastatic tumors in vivo. The initial in vitro experiments showed that TMTP1 specifically bound to highly metastatic tumor cells. We next sought to determine whether TMTP1 could target highly metastatic tumor cells in vivo. For this purpose, three highly metastatic tumor cell lines, PC-3M-1E8, MDA-MB-435S, and MKN-45sci, and one nonmetastatic tumor cell line, PC-3M-2B4, were used to induce s.c. tumors (PC-3M-1E8 and PC-3M-2B4) or orthotopic tumors (MDA-MB-435S and MKN-45sci) in BALB/c nude mice. Mice with tumors 1 to 1.5 cm3 in size were used for the in vivo targeting studies. FITC-TMTP1 (210 μg) was injected via the tail vein and allowed to circulate for 24 h. Tumors and control organs were then excised and processed for frozen sectioning. As shown in Fig. 3, 24 h after i.v. injection, TMTP1 specifically targeted all three highly metastatic tumors, including PC-3M-1E8 (Fig. 3A), MDA-MB-435S (Fig. 3B), and MKN-45sci (Fig. 3C). The fluorescence of TMTP1 appeared mainly in the cell cytoplasm (Fig. 3C, inset). By contrast, no TMTP1 fluorescence was detected in nonmetastatic PC-3M-2B4 tumor (Fig. 3D). Furthermore, animals injected with control peptide FITC-svTMTP1 had no detectable positive fluorescence in the PC-3M-1E8 (Fig. 3E) or MDA-MB-435S tumors (Fig. 3F).

To determine the distribution of TMTP1 in tumor-bearing mice, organs and tissues were removed, frozen sectioned, and examined for fluorescence 24 h after i.v. injection of FITC-TMTP1. Specific fluorescence conjugation was detected in PC-3M-1E8 xenografted tumor tissue (Fig. 4A). However, no specific fluorescence was detected in any normal organs and tissues (Fig. 4B-I). Weak fluorescence was observed in kidney tubuli presumably as a result of uptake of free FITC-TMTP1 from the glomerular filtrate (Fig. 4F). The homing assay with FITC-svTMTP1 showed no obvious fluorescence staining in PC-3M-1E8 xenografted tumor tissue or normal tissues (Fig. 4J). In PC-3M-1E8 xenografted tumor tissue, the fluorescence intensity of FITC-svTMTP1 was 23.5-fold lower than that of FITC-TMTP1 (Fig. 4J).

TMTP1 targets tumor metastases. The data above showed specific binding of TMTP1 to highly metastatic tumor cells in vitro and in vivo. An unresolved question was whether TMTP1 could target metastatic foci, particularly occult micrometastases in the host. This possibility was investigated by using two metastatic mouse models: BALB/c nu/nu mice with lymph node metastases after s.c. inoculation with prostate cancer PC-3M-1E8 cells (Fig. 5A) and mice with mesentery lymph node metastases (Fig. 5B) or liver metastases after orthotopic implantation of gastric MKN-45sci cancer cells. As shown in Fig. 5, the metastatic foci were identified by H&E staining (Fig. 5C and F), and FITC-TMTP1 specifically recognized the metastatic foci (Fig. 5D and G). The metastatic foci of prostate cancer PC-3M-1E8 and gastric cancer MKN-45sci were confirmed by immunohistochemical staining with anti-prostate-specific antigen antibody (Fig. 5E) and anti-CA72-4 antibody (Fig. 5H), respectively.

MKN-45sci gastric tumor-bearing mice were examined, in real time, by whole-body fluorescence microscopy. An abdominal tumor in the mouse receiving FITC-TMTP1 peptide showed a localized fluorescence (Fig. 6A), which was obviously distinguished from the background. On opening the abdominal wall, we found a remarkable accumulation of FITC-TMTP1 in gastric tumors and mesentery lymph node metastases (Fig. 6B). Tumors from mice that received control peptide FITC-svTMTP1 did not show any detectable fluorescence (Supplementary Fig. S1A-C). Imaging of dissected gastric tumors, mesentery metastases, and organs revealed bright fluorescence in the tumors and mesentery metastases even in the small metastatic foci in the visceral surface of the liver (Fig. 6C and D, red arrows). However, no fluorescence was detectable in other organs (Fig. 6C and D).

Mice with MKN-45sci orthotopic gastric cancer develop liver-specific metastasis(41, 42). Successive examination of the formative process of live metastatic foci in this mouse model revealed a very interesting phenomenon. By macrography, approximately 16 days after orthotopic implantation of gastric cancer MKN-45sci cells, multiple gray nodules were visible on the capsular surface of the liver. They typically appeared as nodules 0.5 to 1.5 mm in size, which were centrally umbilicated (Fig. 6E). They contained necrotic plaques with heavy lymph cell infiltration and infrequent pyknotic neoplastic cells inside (Fig. 6F). These foci were examined by three independent pathologists and were not determined to be typical metastatic lesions in morphology. Further immunohistochemical staining with anti-CA72-4 antibody, a specific marker to gastric tumor, showed positive staining of the neoplastic cells in these foci (Supplementary Fig. S3). The ability of FITC-TMTP1 to specifically bind to these microfoci (Fig. 6G) was unexpected. In our later examination, 21 to 28 days after tumor implantation, we saw typical metastatic lesions (Fig. 6H), which were confirmed by pathologists (Fig. 6I). FITC-TMTP1 accurately targeted these metastatic tumors (Fig. 6J). These findings show that TMTP1 has the remarkable capability to specifically target metastatic foci, particularly the very early stage of metastatic foci that could only be identified by specific immunohistochemical staining.

Similarity searching of human proteins for sequences containing the TMTP1 core. We searched SWISSPROT for human proteins containing the TMTP1 peptide core sequence (NVVRQ). Only four human proteins exactly matched the sequence: Ran-binding protein 6, pannexin-2, Ran-binding protein 5, and zinc finger MYM-type protein 3. Both Ran-binding proteins 6 and 5 have conserved armadillo/β-catenin-like repeat domains, which have been implicated in protein-protein interactions. Pannexin-2 belongs to the innexin family, which forms the structural components of gap junctions. Zinc finger MYM-type protein 3 contains a metallochaperone-like domain and is associated with cancer-causing chromosomal translocations. When we extended the searching range to include single positive substitution patterns such as [NDHS]VVRQ or N[VILM]VRQ, we obtained more than 10,000 hits but could not deduce any common characteristics.

Metastasis recurrence is considered to be the result of free tumor cells or micrometastases already present in secondary sites when the primary neoplasm is removed (811). The course of disease in most patients with metastasis is difficult to reverse (12). Therefore, methods used to detect and eliminate occult metastases in cancer patients have received much attention. Several molecular markers have been evaluated for their ability to detect micrometastases (21, 25, 28, 4346). Highly metastatic cancer cells, which are capable of aggressive invasion and stable resistance to chemotherapy, express unique molecular markers or lose some molecules such as E-cadherin, which is involved in the control of cell migration and invasion (41, 47). In this study, we screened a FliTrx library with the highly metastatic prostate cancer cell line PC-3M-1E8 in vitro to select peptides that home to cancer cells and their metastases.

The identified TMTP1 peptide, which contains the core sequence NVVRQ, was designed and synthesized with the structure GCGNVVRQGC. Glycine residues provide a spacer function, permitting the physical formation of a loop structure, and the flanking cysteines allow for the formation of a disulfide loop, thus constraining and mimicking the binding conditions of the initial selection process. TMTP1 specifically bound to highly metastatic tumor cells including prostate cancer PC-3M-1E8, breast cancer MDA-MB-435S, lung cancer PG-BE1, and gastric cancer MKN-45sci in vitro and in vivo. However, it did not bind to the poorly metastatic or nometastatic cell lines PC-3M-2B4, MCF-7, and PG-LH7 or the murine fibroblast cell line NIH/3T3. This indicates that TMTP1 possesses excellent targeting or homing properties for metastatic tumors. In vitro and in vivo binding experiments with the control peptide svTMTP1, whose amino acid composition resembles that of TMTP1, further showed the specific targeting property of TMTP1.

Occult metastases or micrometastases, by definition, are not detected by even the most sophisticated imaging techniques such as positron emission tomography. Some of these occult foci can remain dormant without any signs and symptoms for years (8, 10). Even if dormant, these occult metastases maintain the potential to proliferate and recur. When the balance tips toward malignant proliferation, malignant cells progressively corrode tissues and organs till they thoroughly wreck the host. TMTP1 specifically targeted celiac lymph node metastases of PC-3M-1E8 tumors, the seeding foci, and the mesentery lymph node metastases of MKN-45sci tumors. Remarkably, occult foci even less than 1 mm in diameter was specifically targeted.TMTP1 also specifically recognized atypical liver micrometastases that contained infrequent neoplastic cells. It is conceivable that some neoplastic cells from parts of these lesions may ultimately escape the immune surveillance of the host and become clinical metastases (48). Our study shows that TMTP1 has a remarkable ability to target very early stage of occult metastasis foci. This could significantly manage metastasis or recurrence by detecting and eradicating concealed tumor cells before they evade the host's immune surveillance and cause damage to the host.

The remarkable tumor-homing efficiency of TMTP1 may be due to its propensity to be internalized by cells. Cells that bind TMTP1 transport it across the cell membrane and into the cytoplasm and nucleoplasm. In this regard, TMTP1 is similar to the Tat peptide and other cell-penetrating peptides, which are also taken up by cells (42, 49). An important difference is that TMTP1 peptides are cell specific, targeting the highly metastatic cells that display the “receptor” for this peptide. Peptide internalization is likely to contribute to its effectiveness in becoming concentrated in the targeted tumors. We also observed that incubating PC-3M-1E8 cells with TMTP1 resulted in an obvious dose-dependent inhibition of cell viability and induced apoptosis of tumor cells.2

2

W. Yang, et al. unpublished findings.

Ongoing experiments are in progress to determine the mechanism in which TMTP1 induces apoptosis. In the meantime, we have prepared TMTP1-sTRAIL and TMTP1-DT fusion proteins, in which human soluble tumor necrosis factor-related apoptosis-inducing ligand or diphtheria toxin is fused with the TMTP1 peptide. The antitumor activities of these fusion proteins were evaluated in vitro and in vivo and have revealed promising anticancer activity.3
3

R. Liu, et al. unpublished findings.

It is important to identify the receptor that TMTP1 binds at the cell surface, which will help us to better understand the molecular properties of TMTP1. A BLAST search with the TMTP1 core sequence revealed four proteins, which do not contribute to metastasis, and the search for single positive substitutions yielded redundant proteins with no common biological characteristics. We are now working to isolate the TMTP1 receptor. Affinity chromatography (ProFound Pull-Down) and time-of-flight delayed extraction MALDI mass spectrometer (Bruker Autoflex) are being used for this purpose. The preliminary results of mass spectrometry hinted the aminopeptidase P, a GPI-linked membrane dipeptidese expressed on the surface of vascular endothelial cells and lymphoid cells, may be involved. However, this result needs to be further confirmed.

TMTP1 promises to be a useful tool for detecting occult metastases. Furthermore, eradication of occult micrometastases by coupling radioactive isotopes or anticancer agents to the TMTP1 peptide would be greatly beneficial in protecting cancer patients from metastasis or recurrence.

The authors declare that they have no competing financial interests.

Grant support: National Science Foundation of China grants 30672227, 30571950, 30500596, 30600667, and 30628029 and “973” Program of China grant 2002CB513100.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/).

W. Yang and D. Luo contributed equally to this article.

We thank Drs. Ping Wu, Ping Gong, Jianhui Wen, and Hongxing Tang and Yunqin Chen of Hunan University for fluorescence quantification support and Dr. Luo Qinmin for whole-body imaging.

1
Hawes D, Neville AM, Cote RJ. Detection of occult metastasis in patients with breast cancer.
Semin Surg Oncol
2001
;
20
:
312
–8.
2
Lerut T, Coosemans W, Decker G, et al. Surgical techniques.
J Surg Oncol
2005
;
92
:
218
–29.
3
Pfister DG, Benson AB III, Somerfield MR. Clinical practice. Surveillance strategies after curative treatment of colorectal cancer.
N Engl J Med
2004
;
350
:
2375
–82.
4
Mori T, Fujiwara Y, Sugita Y, et al. Application of molecular diagnosis for detection of peritoneal micrometastasis and evaluation of preoperative chemotherapy in advanced gastric carcinoma.
Ann Surg Oncol
2004
;
11
:
14
–20.
5
Luzzi KJ, MacDonald IC, Schmidt EE, et al. Multistep nature of metastatic inefficiency: dormancy of solitary cells after successful extravasation and limited survival of early micrometastases.
Am J Pathol
1998
;
153
:
865
–73.
6
Pantel K, Otte M. Occult micrometastasis: enrichment, identification and characterization of single disseminated tumour cells.
Semin Cancer Biol
2001
;
11
:
327
–37.
7
Townson JL, Chambers AF. Dormancy of solitary metastatic cells.
Cell Cycle
2006
;
5
:
1744
–50.
8
Cameron MD, Schmidt EE, Kerkvliet N, et al. Temporal progression of metastasis in lung: cell survival, dormancy, and location dependence of metastatic inefficiency.
Cancer Res
2000
;
60
:
2541
–6.
9
Holmgren L, O'Reilly MS, Folkman J. Dormancy of micrometastases: balanced proliferation and apoptosis in the presence of angiogenesis suppression.
Nat Med
1995
;
1
:
149
–53.
10
Naumov GN, MacDonald IC, Chambers AF, Groom AC. Solitary cancer cells as a possible source of tumour dormancy?
Semin Cancer Biol
2001
;
11
:
271
–6.
11
Naumov GN, MacDonald IC, Weinmeister PM, et al. Persistence of solitary mammary carcinoma cells in a secondary site: a possible contributor to dormancy.
Cancer Res
2002
;
62
:
2162
–8.
12
Chambers AF, Groom AC, MacDonald IC. Dissemination and growth of cancer cells in metastatic sites.
Nat Rev Cancer
2002
;
2
:
563
–72.
13
Naumov GN, Bender E, Zurakowski D, et al. A model of human tumor dormancy: an angiogenic switch from the nonangiogenic phenotype.
J Natl Cancer Inst
2006
;
98
:
316
–25.
14
Chambers AF, Naumov GN, Varghese HJ, Nadkarni KV, MacDonald IC, Groom AC. Critical steps in hematogenous metastasis: an overview.
Surg Oncol Clin N Am
2001
;
10
:
243
–55, vii.
15
Lehnert T, Golling M. Indications and outcome of liver metastases resection.
Radiologe
2001
;
41
:
40
–8.
16
Weiss L. Metastatic inefficiency: intravascular and intraperitoneal implantation of cancer cells.
Cancer Treat Res
1996
;
82
:
1
–11.
17
Ilson DH. Oesophageal cancer: new developments in systemic therapy.
Cancer Treat Rev
2003
;
29
:
525
–32.
18
Rubin P, Brasacchio R, Katz A. Solitary metastases: illusion versus reality.
Semin Radiat Oncol
2006
;
16
:
120
–30.
19
O'Day S, Boasberg P. Management of metastatic melanoma 2005.
Surg Oncol Clin N Am
2006
;
15
:
419
–37.
20
Valdespino V, Tsagozis P, Pisa P. Current perspectives in the treatment of advanced prostate cancer.
Med Oncol
2007
;
24
:
273
–86.
21
Barak V, Frenkel S, Kalickman I, Maniotis AJ, Folberg R, Pe'er J. Serum markers to detect metastatic uveal melanoma.
Anticancer Res
2007
;
27
:
1897
–900.
22
Korenberg MJ, Farinha P, Gascoyne RD. Predicting survival in follicular lymphoma using tissue microarrays.
Methods Mol Biol
2007
;
377
:
255
–68.
23
Wang N, Thuraisingam T, Fallavollita L, Ding A, Radzioch D, Brodt P. The secretory leukocyte protease inhibitor is a type 1 insulin-like growth factor receptor-regulated protein that protects against liver metastasis by attenuating the host proinflammatory response.
Cancer Res
2006
;
66
:
3062
–70.
24
Zhu X, Jiang J, Shen H, et al. Elevated β1,4-galactosyltransferase I in highly metastatic human lung cancer cells. Identification of E1AF as important transcription activator.
J Biol Chem
2005
;
280
:
12503
–16.
25
Maurya P, Meleady P, Dowling P, Clynes M. Proteomic approaches for serum biomarker discovery in cancer.
Anticancer Res
2007
;
27
:
1247
–55.
26
Brown CK, Modzelewski RA, Johnson CS, Wong MK. A novel approach for the identification of unique tumor vasculature binding peptides using an E. coli peptide display library.
Ann Surg Oncol
2000
;
7
:
743
–9.
27
Brissette R, Prendergast JK, Goldstein NI. Identification of cancer targets and therapeutics using phage display.
Curr Opin Drug Discov Dev
2006
;
9
:
363
–9.
28
Li G, Passebosc-Faure K, Feng G, et al. MN/CA9: a potential gene marker for detection of malignant cells in effusions.
Biomarkers
2007
;
12
:
214
–20.
29
Samoylova TI, Morrison NE, Globa LP, Cox NR. Peptide phage display: opportunities for development of personalized anti-cancer strategies.
Anticancer Agents Med Chem
2006
;
6
:
9
–17.
30
Hu S, Guo X, Xie H, et al. Phage display selection of peptides that inhibit metastasis ability of gastric cancer cells with high liver-metastatic potential.
Biochem Biophys Res Commun
2006
;
341
:
964
–72.
31
Landon LA, Zou J, Deutscher SL. Is phage display technology on target for developing peptide-based cancer drugs?
Curr Drug Discov Technol
2004
;
1
:
113
–32.
32
Lu Z, Murray KS, Van Cleave V, LaVallie ER, Stahl ML, McCoy JM. Expression of thioredoxin random peptide libraries on the Escherichia coli cell surface as functional fusions to flagellin: a system designed for exploring protein-protein interactions.
Biotechnology NY
1995
;
13
:
366
–72.
33
Zitzmann S, Kramer S, Mier W, et al. Identification of a new prostate-specific cyclic peptide with the bacterial FliTrx system.
J Nucl Med
2005
;
46
:
782
–5.
34
Fei P, Junyu N, Jiangfeng Y, et al. Monoclonal antibodies against human tumor metastasis suppressor gene-1 (TMSG-1): preparation, characterization, and application.
Hybrid Hybridomics
2004
;
23
:
318
–25.
35
Liu Y, Zheng J, Fang W, et al. Isolation and characterization of human prostate cancer cell subclones with different metastatic potential.
Zhonghua Bing Li Xue Za Zhi
1999
;
28
:
361
–4.
36
Zhu W, Zheng J, Fang W. Isolation and characterization of human lung cancer cell subline with different metastatic potential.
Zhonghua Bing Li Xue Za Zhi
1995
;
24
:
136
–8.
37
Yan M, Xu L, Yao M, et al. An establishment of orthotopic implant models for human gastric carcinoma in two of the animals immune deficiencies and their biological properties.
Shanghai Lab Anim Sci
2005
;
25
:
8
–12.
38
Yan M, Yao M, Liu Q, et al. An establishment of liver metastasis models for human gastric carcinoma in nude mice.
Lab Anim Comp Med
2005
;
25
:
72
–5.
39
Chen Y, Xiong T, Yu L, Zeng S, Luo Q. Whole-body fluorescent optical imaging based on power light emitting diode.
Conf Proc IEEE Eng Med Biol Soc
2005
;
2
:
1442
–5.
40
Xiong T, Zhang Z, Liu BF, et al. In vivo optical imaging of human adenoid cystic carcinoma cell metastasis.
Oral Oncol
2005
;
41
:
709
–15.
41
Jeschke U, Mylonas I, Kuhn C, et al. Expression of E-cadherin in human ductal breast cancer carcinoma in situ, invasive carcinomas, their lymph node metastases, their distant metastases, carcinomas with recurrence and in recurrence.
Anticancer Res
2007
;
27
:
1969
–74.
42
Lundberg P, Langel U. A brief introduction to cell-penetrating peptides.
J Mol Recognit
2003
;
16
:
227
–33.
43
Abdul-Rasool S, Kidson SH, Panieri E, Dent D, Pillay K, Hanekom GS. An evaluation of molecular markers for improved detection of breast cancer metastases in sentinel nodes.
J Clin Pathol
2006
;
59
:
289
–97.
44
Ferrari AC, Stone NN, Kurek R, et al. Molecular load of pathologically occult metastases in pelvic lymph nodes is an independent prognostic marker of biochemical failure after localized prostate cancer treatment.
J Clin Oncol
2006
;
24
:
3081
–8.
45
Norell H, Carlsten M, Ohlum T, et al. Frequent loss of HLA-A2 expression in metastasizing ovarian carcinomas associated with genomic haplotype loss and HLA-A2-restricted HER-2/neu-specific immunity.
Cancer Res
2006
;
66
:
6387
–94.
46
Donskov F, von der Maase H. Impact of immune parameters on long-term survival in metastatic renal cell carcinoma.
J Clin Oncol
2006
;
24
:
1997
–2005.
47
Annicotte JS, Iankova I, Miard S, et al. Peroxisome proliferator-activated receptor gamma regulates E-cadherin expression and inhibits growth and invasion of prostate cancer.
Mol Cell Biol
2006
;
26
:
7561
–74.
48
Kaplan RN, Riba RD, Zacharoulis S, et al. VEGFR1-positive haematopoietic bone marrow progenitors initiate the pre-metastatic niche.
Nature
2005
;
8
:
820
–7.
49
Wu RP, Youngblood DS, Hassinger JN, et al. Cell-penetrating peptides as transporters for morpholino oligomers: effects of amino acid composition on intracellular delivery and cytotoxicity.
Nucleic Acids Res
2007
;
35
:
5182
–91.

Supplementary data