|Home | About | Journals | Submit | Contact Us | Français|
A tumor homing peptide, LyP-1, selectively binds to tumor-associated lymphatic vessels and tumor cells in certain tumors and exhibits an anti-tumor effect. Here, we show that the protein known as p32 or gC1qR is the receptor for LyP-1. Various human tumor cell lines were positive for p32 expression in culture, and the expression was increased in xenograft tumors grown from the positive cell lines. FACS analyses with anti-p32 antibodies showed that p32-positive cell lines expressed p32 at the cell surface. These cells bound and internalized LyP-1 peptide in proportion to the cell surface expression level, which correlated with malignancy rather than total p32 expression in the cells. Like the LyP-1 peptide, p32 antibodies highlighted hypoxic areas in tumors, where they bound to both tumor cells and cells that expressed macrophage/myeloid cell markers and often appeared incorporated into the walls of tumor lymphatics. Significant p32 expression was common in human cancers and the p32 levels were often greatly elevated compared to the corresponding normal tissue. These results establish p32, particularly its cell surface-expressed form, as a new marker of tumor cells and tumor-associated macrophages/myeloid cells in hypoxic/metabolically deprived areas of tumors. Its unique localization in tumors and its relative tumor specificity may make p32 a useful target in tumor diagnosis and therapy.
Specific molecular signatures expressed by tumor cells and tumor vasculature distinguish tumors from their non-malignant counterpart tissues. Tumor-associated antigens such as certain growth factor and cytokine receptors, membrane-type matrix metalloproteinases, and cell adhesion molecules are highly expressed in many tumors. The biochemical features that distinguish tumor vasculature from the vasculature of normal tissues include the expression of various angiogenesis-related molecules (1, 2). Tumor lymphatics are also specialized, since they express markers that are not present in the lymphatics of normal tissues (or in tumor blood vessels) (3, 4).
The distinct protein profile of tumor vessels and tumor cells can be exploited in ligand-directed targeting of diagnostic and therapeutic agents. Targeting can improve the specificity and efficacy of a compound while reducing side effects (5-7). These successes emphasize the need to find additional molecules that recognize selectively expressed markers in tumors.
In vivo screening of phage libraries that display random peptide sequences on their surface has yielded a number of specific homing peptides for tumor vasculature and tumor cells (3, 4, 8-10). Identification of receptors for homing peptides provides new tumor markers, and may also reveal signaling pathways that, if interrupted, affect tumor growth/malignancy. LyP-1, a cyclic nonapeptide that specifically recognizes lymphatic vessels in certain tumors (3), is a case in point. Lymphatic vessels are an important conduit for the spread of solid tumors, and their abundance in and around tumors correlates with propensity to metastasize (11, 12).
The LyP-1 peptide provides a marker for these vessels, but also binds to tumor cells, offering the potential to selectively target both tumor lymphatics and tumor cells. Moreover, the target molecule (receptor) for the LyP-1 peptide appears to be involved in tumor growth because systemic administration of LyP-1 inhibits tumor growth in mice (13). The mechanism of the anti-tumor effect is unknown, but LyP-1 appears to be cytotoxic against tumor cells undergoing stress, as LyP-1 accumulation coincides with hypoxic areas in tumors and tumor starvation enhances its binding and internalization in cultured tumor cells (13). These unique properties of the LyP-1 system prompted us to search for the tumor cell receptor for this peptide.
Here we identify p32/p33/gC1qR/HABP1 (p32) as the cellular receptor for LyP-1. This protein was originally isolated based on its co-purification with the nuclear splicing factor SF-2 (14). It was subsequently shown to bind the complement component C1q and designated the gC1q receptor (gC1qR) (15). Numerous extracellular and intracellular proteins have been reported to bind to p32; indeed, p32 may be a multifunctional chaperone protein (16, 17).
The p32 protein is a doughnut-shaped trimer (18) that is primarily localized in the mitochondria (19-21), but has also been reported to be present at the cell surface (15, 22, 23) and in the nucleus (14, 24). The role of p32 in mammalian cells is unclear. The yeast p32 homologue has been reported to regulate oxidative phosphorylation (19), and a link to autophagy has been proposed (25, 26). Elevated expression of p32 in cancer has also been noted (27-29).
Here we show that p32 is primarily localized in hypoxic/nutrient-deprived regions within tumors and that in addition to tumor cells, a tumor-associated macrophage/myeloid cell subpopulation closely linked to tumor lymphatics expresses high levels of p32. We also report that the expression of p32 at the cell surface, where it can be targeted with p32-binding antibodies and peptides, adds a second level of tumor specificity to p32 expression. The unique expression pattern of p32 in tumor cells, tumor lymphatics and tumor-associated macrophages/myeloid cells makes p32 a potential target for the diagnosis and therapy of cancer.
Mouse monoclonal 60.11 and 74.5.2 anti-p32 antibodies were purchased from Chemicon (Temecula, CA). Rat monoclonal anti-mouse CD-31, rat anti-MECA-32, rat anti mouse CD-11b and R-Phycoerythrin (R-PE)-conjugated rat anti-mouse Gr-1 were from BD-PharMingen (San Jose, CA), the anti-epithelial membrane antigen, EMA (clone E29), was from Chemicon, anti-CD68 from Oncogene Research Products (San Diego, CA) and anti β-actin from Sigma-Aldrich (St. Louis, MO). Rat anti-podoplanin antibody was kindly provided by Drs. T. Petrova and K. Alitalo (University of Helsinki, Finland). Polyclonal anti LYVE-1 antibody has been described (3). ChromPure Rabbit IgG (whole molecule) was from Jackson ImmunoResearch Laboratories (West Grove, PA) and purified Mouse IgG1 (mIgG) from BD-Pharmingen. Purified polyclonal anti-full-length p32 was a gift from Dr. B. Ghebrehiwet (Stony Brook University, NY). Polyclonal anti-peptide antibody against p32 was generated in rabbits against a mixture of peptides corresponding to amino acids 76-93 of mouse (TEGDKAFVEFLTDEIKEE) and human (TDGDKAFVDFLSDEIKEE) p32 protein. The peptides were coupled to keyhole limpet hemocyanin (Pierce, Rockford, IL) via a cysteine residue added at their N-termini. The antibody was affinity purified on the peptides coupled to Sulfolink Gel (Pierce) via the cysteine residue. Dr. A. Strongin (Burnham Institute for Medical Research, La Jolla, CA) kindly provided human p32 cDNA in pcDNA3.1 Zeo and pET-15b vectors. Oligonucleotide duplexes for transient siRNA knockdown of p32 (C1QBP-HSS101146-47-48 Stealth RNAi) and negative control duplexes (Stealth RNAi control low GC and medium GC) were purchased from Invitrogen (Carlsbad, CA). Tissue Arrays (core diameter 0.6 mm) of paraformaldehyde fixed and paraffin-embedded tumor and normal tissue samples were from Applied Phenomics LLC (Tartu, Estonia).
MDA-MB-435, MDA-MB-231, MCF7, C8161, BT549, HL60, Raji and 4T1 cells were maintained in DMEM supplemented with 10% FBS and 1% Glutamine Pen-Strep (Omega scientific) at 37°C/5%CO2. The MCF10-A cell line was from ATCC (Manassas, VA), while the MCF10-AT and MCF10-CA1a cells were from The Barbara Ann Karmanos Cancer Institute (Detroit, MI). Cells were maintained in DMEM supplemented with 10% FBS, 1% Glutamine Pen-Strep and 100ng/ml human EGF (Sigma-Aldrich, Saint Louis, MI).
To produce MDA-MB-435 and 4T1 tumors, BALB/c athymic nude or normal BALB/c mice were orthotopically injected into the mammary fat pad with 2×106 MDA-MB-435 cells suspended in 100μl of PBS. C8161 tumors were grown from cells inoculated subcutaneously on the flank. All animal experimentation received approval from the Animal Research Committee of Burnham Institute for Medical Research.
Cells were lysed in cold PBS/200mM n-octyl-β-D glucopyranoside (Calbiochem; San Diego, CA) and supernatant aliquots containing 1mg protein were pre-cleared with 40 μl of streptavidin beads for 2 h at 4°C prior to incubation over night at 4 °C with streptavidin beads loaded with biotinylated peptides (3 μg/10 μl beads). The material eluted from the beads was separated on a 4-20% polyacrylamide gel and visualized by silver staining (Invitrogen). Bands that appeared in the LyP-1 but not control peptide pull down were cut out, digested with trypsin, and the resulting peptides were analyzed by matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) mass spectrometry. Profound software was used to query the information against a protein sequence data.
Microtiter wells (Costar, Corning, NY) were coated with 5 μg/ml of either purified p32 or BSA (Sigma-Aldrich), blocked with Pierce Superblock buffer, and incubated with 108 pfu of LyP-1 or control phage in 100μl of TBS/0.05% Tween-20 for 16 h at 37°C. After 6 washes in TBS/0.05% Tween-20, bound phage were eluted with 200 μl of Tris-HCl 1M pH 7.5/0.5% SDS for 30 min and quantified by plaque assay. To test antibody inhibition of phage binding, 1.5×107 pfu of LyP-1 or insertless phage were allowed to bind for 6 h at 37°C to p32-coated wells coated in the presence of 20 μg/ml of monoclonal anti-p32 or mouse IgG. When the assay was performed with cells, 2×106 Raji cells were resuspended in 500μl of PBS/1% BSA and pre-incubated for 1h at 4°C with 40μg/ml of anti-p32 or mIgG. LyP-1 or insertless phage (108 pfu) were subsequently added to the cells and incubated at 4°C for 3 h.
LyP-1 binding affinity to p32 was quantified by an ELISA-based assay. Microtiter wells coated with 3μg/ml of purified p32 protein were incubated for 1 hour at room temperature with various concentrations of biotinylated LyP-1 peptide in PBS (100μl/well). After washing with Tris-buffered saline containing 1mM CaCl2 and 0.01% Tween-20, streptavidin-conjugated horseradish peroxidase (Zymed, Carlsbad, CA) was added to the wells and incubated for 1 hour at room temperature. Peptide binding to p32 was quantified with 2,2-azino-bis(3-ethylbenzthiazoline-6-sulfonic acid) (Sigma) as a substrate. The background that was determined from wells without p32 coating was subtracted. Kd values were calculated using Prism software.
FACS analysis of cell surface p32 was performed on live cells. Approximately 2.5×105 cells were stained with polyclonal anti-full-length/N-terminus p32 or rabbit IgG (20μg/ml) and secondary antibody, each for 30 min at 4°C, and analyzed without permeabilization and by gating for propidium iodide-negative (live) cells (30).
Immunohistochemical staining of frozen tissue sections was carried out using acetone fixation and reagents from Molecular Probes (Invitrogen). Antibody binding was detected with secondary antibodies, which were: AlexaFluor-594 goat anti-rat or rabbit IgG, AlexaFluor-488 goat anti-rabbit IgG. Hypoxyprobe-1 (pimonidazole hydrochloride; Chemicon) was injected into tumor-bearing mice and hypoxic areas in tumors were detected with a FITC-conjugated anti hypoxyprobe-1 antibody according to the manufacture’s instructions.
Paraffin-embedded normal and malignant human tissue array sections were deparaffinized and treated with Target Retrieval Solution (DakoCytometion, Carpinteria, CA). For sequential staining the sections were stained as described above, except that the sections were treated with DAKO Biotin Blocking system, and p32, EMA, and CD68 were detected with biotinylated anti-mouse IgG and Vectastain ABC kit (Vector Laboratories Inc, Burlingame, CA). We used CD68 as a macrophage marker because an antibody that detects this antigen in paraffin-embedded sections was available. In the double immunohistochemistry labelling, the tumor array was first stained for CD68 as indicated above. A diaminobenzidine chromogen (DakoCytomation) (brown color) was used for antibody detection. Next, the slide was incubated with rabbit anti N-terminus p32 followed by the use of Envision Rabbit-HRP detection system (DakoCytomation) and Vector VIP substrate kit for peroxidase (Vector laboratories), which produces a purple reaction product. Nuclei were counterstained with Vector Methyl Green (Vector laboratories). The slides were scanned on a Scanscope CM-1 scanner and subsequently processed using ImageScope software (Aperio Technology) with color deconvolution and separation algorithms.
The clinical samples were blindly analyzed by a pathologist in the Institute’s core facility. The intensity of p32 staining was visually graded on a scale of 0 to 3. Parallel staining of an epithelial membrane antigen was used to identify tumor cells and an immuno-score (scale 0 to 300) was assigned to each malignant sample based on the percentage of tumor cells within the tissue and their intensity of p32 staining. The tissue arrays used contained a total of 41 normal tissues and 81 tumor samples, each one categorized for histology and malignancy grade. The type of tumors present in the array were: breast carcinomas (ductal, lobular and mucinous), endometroid adenocarcinomas, ovarial adenocarcinomas (serous and mucinous), colon adenocarcinomas (tubular, tubular-villous, and mucinous), adenocarcinomas of the stomach, pancreatic ductal carcinomas, kidney carcinomas, skin melanomas, hepatocellular carcinoma, carcinomas of the testis (seminoma and embryonal), prostate carcinomas, lung squamous cells carcinomas and glioblastomas.
To identify the receptor for the LyP-1 peptide, we incubated biotin-labeled LyP-1 and control peptides with extracts from the MDA-MB-435 cell line, which binds and internalizes LyP-1 (13). LyP-1 bound a specific band in the 30 kDa range that was not seen in the controls (Fig. 1A, left panel). Two independent MALDI-TOF analyses indicated that the specific band represents the mature form of a protein known as p32 or gC1qR (15, 31). LyP-1 affinity isolation also yielded p32 from cultured BT549 breast carcinoma cells and from extracts of MDA-MB-435 xenograft tumors (data not shown).
The identification of the LyP-1-binding protein as p32 was confirmed by immunoblotting and phage binding assays. A monoclonal antibody directed against p32 specifically recognized the same band as p32 (Fig. 1A, right panel). No detectable p32 was pulled down by the control peptides. The LyP-1 phage bound to purified p32 protein on a solid phase an average of 60-fold more than insertless control phage, while only marginal binding of either phage to plates coated with BSA was seen (Fig.1B). LyP-2, a peptide which shares a consensus sequence with LyP-1 but binds a different spectrum of tumor lymphatics (4), did not significantly bind to p32. A monoclonal antibody, mAb 60.11, which binds to p32 near the N-terminus (amino acids 76-93), reduced LyP-1 phage binding to p32 by 90% (Fig. 1C). In contrast, mAb 74.5.2, which recognizes the C-terminal end of p32 (amino acids 204-218), did not inhibit the phage binding. These results indicate that the interaction between LyP-1 and p32 is specific, and that a site contained between amino acids 76 and 93 of p32 plays an important role in the interaction. Saturation binding experiments gave an average binding affinity of Kd=3μM for the LyP-1-p32 interaction (Fig. 1D).
We next studied the effect of forced expression and knockdown of p32 on LyP-1 binding. Transient transfection of C8161 cells with p32 cDNA increased LyP-1 phage binding to 5-fold over control phage (Fig. 2A). A minor increase in binding was obtained upon transfection with the empty vector. Transfection with a p32 siRNA construct markedly reduced expression in MDA-MB-435 cells (Fig. 2B, upper panels), with an accompanying reduction in the binding of FITC-LyP-1 peptide to the cells (Fig. 2B lower left panel). Controls showed that an unrelated siRNA did not affect p32 expression or LyP-1 binding, and neither siRNA changed the expression of β-actin. Also, a FITC-labeled control peptide, which like LyP-1 has three basic residues but does not significantly bind to the MDA-MB-435 cells (3), gave a similar amount of background fluorescence in the p32 knockdown and control cells (Fig. 2B, lower right panel). Finally, blocking p32 with mAb 60.11 in Raji cells (which express high levels of cell surface p32) reduced the binding of LyP-1 phage to these cells in a concentration-dependent manner by up to 50%. LyP-1 phage binding to the Raji cells was unaffected by mAb 74.5.2 (Fig. 2C). These results are consistent with those obtained with purified p32 protein (Fig.1C) and indicate that the level of p32 expression dictates LyP-1 binding to the cells.
For p32 to act as a LyP-1 receptor, it would have to be expressed at the cell surface. While primarily localized in intracellular compartments (mitochondria, nucleus and cytoplasm), p32 is also present at the cell surface (15, 32, 33). Here we confirm the presence of p32 at the cell surface. A polyclonal anti-p32 antibody produced a small but consistent shift in FACS analysis of live MDA-MB-435 cells (Fig. 3A). A greater shift was obtained in an MDA-MB-435 subclone (S35), which binds LyP-1 with higher efficiency than the parental cell line (P. Laakkonen and E. Ruoslahti, unpublished results). Significant levels of cell surface p32 were also detected in several other tumor cell lines, including the MDA-MB-231 and MCF7 human breast cancer lines, 4T1 mouse breast cancer cells and Raji Burkitt lymphoma cells (Fig 3A). Interestingly, the amount of cell surface p32 did not always correlate with the level of total p32 (Fig. 3B), indicating that cell surface expression of p32 is controlled separately from its total level. Cell surface p32 was more strongly expressed in single cell suspensions from MDA-MB-435 tumor xenografts than in cultured MDA-MB-435 cells, whereas C8161 cells were essentially negative for cell surface p32 whether tested as cultured cells (data not shown), or as single cell suspensions from C8161 tumor xenografts (Fig. 3C).
To investigate changes in p32 expression during neoplastic progression, we used the MCF10 model, which consists of progressively more malignant mammary epithelial cell lines obtained by transformation and subsequent cycling as tumors in nude mice (34). These cell lines were derived from the spontaneously immortalized, but otherwise normal, MCF10-A cells. MCF10-AT cells are premalignant in that they are prone to neoplastic transformation (35). The MCF10-CA1a cells are fully malignant and metastatic (36). The premalignant line expressed an increased level of total p32, but significant cell surface expression of p32 was limited to the fully malignant MCF10-CA1a line (Fig. 3D).
The results presented above suggested that cell surface expression of p32 is a particular characteristic of malignant cells. Comparison of p32 expression in tumors and p32-positive normal tissues provided further support for this notion. As already shown, cells suspensions from MDA-MB-435 tumors were strongly positive for anti-p32 binding in FACS. In contrast, cells from the spleen and kidney, which expressed total p32 at a level comparable to the tumor, showed no cell surface expression detectable by FACS (Fig. 4A). These results indicate that cell surface localization of p32 is regulated independently of total p32 expression, and that malignant progression and tumor microenvironment may enhance the expression of p32 at the cell surface.
In agreement with the selective presence of cell surface p32 in tumor-derived cells, intravenously injected anti-p32 antibody accumulated in MDA-MB-435 tumor tissue, but anti-p32 levels in the spleen were no greater than those of control IgG, and no anti-p32 (or control IgG) was detected in the kidneys of the mice (Fig 4C). Similar results were obtained with tumors grown from 4T1 mouse breast cancer cells (Fig 4D).
To investigate the localization of p32 in tumors, sections of MDA-MB-435 tumor xenografts were stained for p32, and LYVE-1 and podoplanin (lymphatic/macrophage markers). Clusters of cells strongly positive for p32 were found in close proximity of podoplanin/LYVE-1-positive structures, whereas there was no association with blood vessels as visualized by staining for CD31 or Meca-32 (Fig. 5A, upper panels). Higher magnification micrographs revealed p32-expressing cells lining vessel-like podoplanin and LYVE-1-positive structures that are presumably tumor lymphatics (Fig. 5A, lower panels). Normal tissues and C8161 tumor xenografts showed much less p32 staining than the MDA-MB-435 tumors (data no shown). Intravenously injected FITC-LyP-1 peptide accumulated in tumor areas with high levels of ex vivo antibody staining for p32 and closely associated with vessel lumens (Fig. 5B-upper panels). Furthermore, p32 was strongly positive in the areas where an injected hypoxia probe localized (Fig 5B-lower panels). As previously noted for the LyP-1 peptide (13), the cells positive for p32 were generally negative for hypoxia probe and vice versa, possible because p32 expression is induced by a low nutrient rather than low oxygen environment (VF and ER, unpublished results).
The p32/LyP-1 positive cells partially co-localized with the macrophage and myeloid cell markers CD11b and Gr-1 (Fig. 5C). Upon quantification, we found a 61±14 % concordance of CD11b staining and FITC-LyP-1 homing, indicating that a major macrophage subpopulation in tumors expresses p32. Similar results for the co-localization of myeloid cell and lymphatic visual markers with p32 were obtained with 4T1 tumors (not shown). The p32-positive cells integrated into the lymphatics in these tumors (Fig. 5D) are likely tumor myeloid cells and/or macrophage-like precursors of lymphatic endothelial cells (37-39).
Next, we studied p32 expression by tumor and myeloid cells in clinical samples of human carcinomas. In agreement with earlier results (28) immunohistochemical staining showed that several tumor types expressed much higher p32 expression levels than the corresponding normal tissues. Results from various tumor types are summarized in Supplementary Table 1, and Figure 6A shows examples of cancer versus normal tissue comparisons. In particular, breast carcinoma (17 out of 20), endometroid adenocarcinoma (5 out of 5), melanoma (3 out of 4), and carcinomas of the colon (5 out of 5) and testis (3 out of 3), as well as squamous cell carcinomas of the lung (3 out of 4), exhibited markedly elevated p32 expression. There was no obvious association of the p32 expression levels with the malignancy grade of the tumors. The 9 prostate carcinomas we examined contained p32 at either very low or undetectable levels. The expression of p32 was also high in cancers of stomach, pancreas and kidney, but the corresponding non-malignant tissues also expressed p32 at readily detectable levels (Fig. 6A and results shown in Figure 3).
We then investigated the expression of p32 in human tumor stromal cells. For this purpose we used two approaches: first, comparison of p32 staining with staining for epithelial (EMA) and macrophage/myeloid (CD68) cell markers in serial sections (Fig 6B); second, color deconvolution of p32/CD68 and p32/EMA double immunohistochemistry (Fig 6C, and data not shown). We found that in 12 out of 47 samples expressing both p32 and CD68, p32 was essentially confined to tumor cells (EMA-positive) (Supplementary Table 1, Fig 6B and C, panels 1-3), while in the remaining 35 samples both the tumor cells and stroma were positive (Supplementary Table 1, Fig 6B and C, panels 4-6). The localization of the macrophage/myeloid cell marker and p32 in the same cells was also confirmed by double immunofluorescence staining in frozen sections of three breast cancers (Supplementary Fig. 1). The presence of p32 in tumor stromal cells has not been previously noted.
We report here that (1) a mitochondrial/cell surface protein, p32/gC1qR, is the receptor for a tumor-homing peptide, LyP-1, which specifically recognizes an epitope in tumor lymphatics and tumor cells in certain cancers; (2) tumor-specificity of the peptide and anti-p32 antibodies is the result of higher expression of total p32 and the propensity of malignant and tumor-associated cells to express p32 at the cell surface and (3) some tumors contain a p32-positive subpopulation of macrophages/myeloid cells that are closely associated with tumor lymphatics.
Several lines of evidence show that the LyP-1 peptide specifically binds a protein known as p32 or the receptor for the C1q component of the complement, gC1qR. First, pull down assays of tumor cell extracts with LyP-1 peptide revealed a specific band identified by mass spectrometry as p32. Second, LyP-1 phage bound purified p32 protein and the interaction was inhibited by antibodies directed against the N-terminus of p32. Third, endogenous expression levels and cell surface localization of p32 correlated with the ability of different cell lines to bind LyP-1. Fourth, overexpression of p32 enhanced, and RNAi silencing decreased, LyP-1 binding to cells. Finally, intravenously injected FITC-LyP-1 peptide homed in vivo to those areas in tumors where p32 expression was high.
The p32 protein is primarily mitochondrial, but it can be found in the cytoplasm, nuclei, and most importantly for the LyP-1 binding, at the cell surface (15, 32, 33). How p32 is targeted to multiple compartments is unclear; however, several other mitochondrial proteins are also found in extra-mitochondrial locations (40). Interestingly, two ubiquitous intracellular proteins, nucleolin (41) and annexin 1 (42) are aberrantly expressed at the cell surface in tumor blood vessels, where they serve as specific markers of angiogenesis. The expression of p32 in tissues is much more restricted than that of nucleolin or annexin 1, but its cell surface expression adds a further degree of tumor specificity, explaining the strikingly specific tumor accumulation of the LyP-1 peptide (3, 13) and anti-p32 (this study) upon systemic administration.
Antibody staining of tissue sections with anti-p32 antibody, and intravenously injected anti-p32, confirmed the previously reported association of LyP-1 with specific areas in tumors. Similar to the LyP-1 peptide (3, 13), the antibody outlined two main locations within tumors: cell clusters in areas that were rich in lymphatics, but sparsely populated with blood vessels, and vessel-like structures that apparently represent lymphatics. We found that a significant number of intensely p32-positive cells within tumors were also positive for macrophage/myeloid cells markers. This is the first report on the presence of p32 in tumor macrophages. Bone marrow-derived macrophages that contribute to lymphangiogenesis have been described by others (37, 39, 43). We hypothesize that the LyP-1/anti-p32-positive cells represent a distinct macrophage/myeloid cells subpopulation that is induced to express p32 in the environment of some tumors and that can serve as a precursor to lymphatic endothelial cells.
The elevated overall expression of p32 in tumors, its localization at the cell surface in tumor cells and in a subpopulation of lymphatic vessel-associated macrophages/myeloid cells, and the absence of cell surface p32 in normal tissues makes cell surface p32 an attractive target molecule for the delivery of diagnostic probes and therapeutics into tumors. The preferential presence of the cells that are strongly positive for p32 in hypoxic/nutrient depleted areas makes it possible to reach parts of tumors that are not readily accessible to passive delivery from blood vessels. Selective targeting of p32-positive tumor cell and macrophage/myeloid cells subpopulations will allow studies on the role of these cells in tumor growth. Monoclonal human antibodies and small molecular weight compounds that bind at the LyP-1 binding site of p32 could be developed to harness this potential.
We thank Drs. Eva Engvall, Douglas Hanahan, Masanobu Komatsu and Elena Pasquale for comments on the manuscript; Dr. Berhane Ghebrehiwet kindly provided the polyclonal anti full-length p32 antibody and Dr. Alex Strongin the p32 expression constructs. Simone Codeluppi helped with imaging, Dr. Edward Monosov gave microscopy advice and Guillermina Garcia and Robbin Newlin assisted with histology.
This work was supported by grants CA115410 and in part by CA104898 and Cancer Center Support Grant CA 30199 from the NCI. VF received support from the Susan Komen Foundation.