Journal of Surgical Research 142, 320–326 (2007)

Angiocidin Inhibitory Peptides Decrease Tumor Burden in a MurineColon Cancer Model1

Catherine Liebig, M.D.,* Neeti Agarwal, Ph.D.,* Gustavo E. Ayala, M.D.,† Gordana Verstovsek, M.D.,†George P. Tuszynski, Ph.D.,‡ and Daniel Albo, M.D., Ph.D.*,2

*Michael E. DeBakey Department of Surgery, Baylor College of Medicine, Houston, Texas; †Department of Pathology, Baylor College ofMedicine, Houston, Texas; and ‡Department of Neuroscience, Center for Neurovirology, Temple University School of Medicine,

Philadelphia, Pennsylvania

Submitted for publication January 9, 2007

doi:10.1016/j.jss.2007.02.036

Introduction. We have recently developed two inhib-itory peptides that target angiocidin, a key mediatorof tumor progression and angiogenesis. In this study,we investigate the expression of angiocidin in humancolon cancer specimens and evaluate the therapeuticefficacy of our angiocidin inhibitory peptides.

Methods. We created a colon cancer tissue array con-taining primary tumor, normal colon, negative and pos-itive lymph nodes, and liver metastases (when available)from 159 consecutive colon cancer specimens. Angioci-din expression was determined by immunohistochemis-try. The efficacy of 6-mer and 25-mer angiocidin inhibi-tory peptides was determined in a murine model ofhuman colon cancer. Treatment efficacy was based onprimary tumor volume and measures of tumor burden,including internal disease score and health score. West-ern blots were used to determine angiocidin expressionin xenografts.

Results. Eighty-nine percent of primary tumors and91% of positive lymph nodes expressed angiocidin.Normal colon was negative in 94% of specimens, andnormal lymph nodes were negative or weakly positivein 79% of specimens. All liver metastases were positivefor angiocidin. Animals in both peptide treatmentgroups showed improvement in health score and in-ternal disease score compared with control animals(P � 0.001). Treatment with 6-mer and 25-mer peptideresulted in 3-fold and 16-fold reductions, respectively,in primary tumor volume (P � 0.001). Angiocidin ex-pression in primary tumors of peptide-treated micecorrelated with tumor burden (P < 0.05).

1 This manuscript was awarded an AAS Resident Research Award.2 To whom correspondence and reprint requests should be ad-

dressed at Michael E. DeBakey VA Medical Center, 2002 Holcombe

Blvd., OCL 112-A, Houston, TX 77030. E-mail: dalbo@bcm.tmc.edu.

3200022-4804/07 $32.00© 2007 Elsevier Inc. All rights reserved.

Conclusions. Angiocidin is overexpressed in humancolon cancer specimens. Angiocidin-inhibitory pep-tides are well tolerated in vivo and effectively reduceprimary tumor volume and tumor burden in humancolon cancer xenografts. © 2007 Elsevier Inc. All rights reserved.

Key Words: colon cancer; angiocidin; CSVTCG; pep-tide therapy; xenograft model; nude mouse model;tumor burden; tissue array; angiocidin-inhibitorypeptide.

INTRODUCTION

Colon cancer causes almost 55,000 deaths per year inthe United States making it the second leading causeof cancer deaths [1]. Almost 60% of patients presentwith regional and/or distant disease (stages 3 and 4) atthe time of diagnosis [2]. Despite toxic adjuvant che-motherapy, nearly one third of the patients with stage3 disease will die from their colon cancer within 5 y ofdiagnosis [1]. The 5-y survival rate for patients withstage 4 disease barely reaches 10%, further underscor-ing the need for more effective and less toxic chemo-therapeutics [2].

Angiocidin, originally referred to as the CSVTCGreceptor, is a tumor-associated molecule that hasemerged as a key mediator of angiogenesis and tumorprogression [3, 4]. Overexpression of angiocidin hasbeen shown in many solid tumors including colorectalcancer, and increased levels of angiocidin have beendemonstrated in the serum of patients with certainepithelial malignancies [5, 6]. Preliminary reports us-ing immunohistochemistry have shown low to negligi-ble expression of angiocidin in normal adult humantissues, making it a key molecular target for tumor-

specific therapy [4, 7, 8].

321LIEBIG ET AL.: ANGIOCIDIN INHIBITORY PEPTIDES IN COLON CANCER

The pro-tumor effects of angiocidin are the result ofmyriad high-affinity binding interactions with matrixproteins, including integrins, Type 1 collagen andthrombospondin-1 (TSP-1) [9–11]. These angiocidin-stromal protein binding interactions lead to up-regulation in gelatinase expression and matrix remod-eling, critical elements in both angiogenesis andmetastasis [4, 10].

We have recently synthesized two angiocidin inhib-itory peptides that block the activity of angiocidinthrough inhibition of its stromal protein binding inter-actions. Preliminary data show that these peptides areexcellent inhibitors of angiogenesis and tumor cell in-vasion in vitro [4, 5, 9]. The CSVTCG peptide is a sixamino acid residue peptide that blocks the binding ofangiocidin to TSP-1. The second peptide is a 25 aminoacid sequence derived from the collagen and integrinbinding sites within the angiocidin molecule.

In the present study, we created a tissue array ofhuman colon adenocarcinomas to elucidate the differ-ential expression of angiocidin in tumor specimensversus matched normal tissues. We demonstrate thatangiocidin is expressed in human colon cancer xeno-grafts and investigate the potential therapeutic effi-cacy of our two angiocidin-inhibitory molecules, the6-mer and 25-mer peptides. Using a nude mousesplenic injection model of human colon cancer, we aimto show that these two peptides are well-tolerated invivo and extremely effective at inhibiting tumorgrowth and decreasing tumor burden.

MATERIALS AND METHODS

Creation of a Human Colorectal Cancer Tissue Array

Original hematoxylin and eosin slides from 159 consecutive re-sected colorectal adenocarcinoma specimens were reviewed by apathologist and mapped. Areas of highest tumor grade and normalcolon were circled. Normal colon specimens came from blocks oftissue distant from the lesion unless the specimen was a polyp, inwhich case the normal colon was immediately adjacent to areas oftumor or dysplasia. Benign and metastatic areas of lymph nodeswere also circled if present in the resection specimen. Biopsied orresected liver metastases were included if present. The tissue arraywas built manually using 2 mm punch biopsy cores that were trans-ferred from mapped donor blocks and placed into recipient paraffinarray blocks. From each mapped and sampled area, cores were takenin duplicate. Internal controls containing up to 10 different types oftissue within each core were also placed at spaced intervals withineach recipient block. A database was built for every block producedand included the coordinates of each core and the area and case oforigin. The final tissue array was composed of 2 to 10 cores perpatient, with a total of 15 blocks containing about 900 cores. Collec-tion of human tissue specimens was performed under a protocolapproved by the Baylor College of Medicine Institutional ReviewBoard and patient identifiers were removed before release of tissueblocks from the pathology department.

Evaluation of Angiocidin Expression by Immunohistochemistry

All reagents were purchased from Vector Laboratories (Burlin-

game, CA) unless otherwise specified and immunohistochemistry was

performed according to the protocol supplied with the Vectastain EliteABC Reagent Kit. Tissue array slides containing 5 �m-thick tissuecores were deparaffinized in xylene and rehydrated through decreasingconcentrations of alcohol ending in phosphate-buffered saline (PBS).Slides were treated with 0.3% hydrogen peroxide for 10 min at roomtemperature and antigen retrieval for angiocidin was performed for 20min in a vegetable steamer in 10 mmol/L citrate buffer (pH 6.0). Slideswere then blocked with blocking serum for 10 min and then incubatedwith monoclonal mouse antihuman angiocidin antibody (Covance, Den-ver, CO) for 1 h at room temperature at a dilution of 1:100. Thesecondary biotinylated universal antibody was applied for 30 min fol-lowed by 30 min incubation with avidin-biotin enzyme. After rinsing,slides were visualized by diaminobenzidine chromogen solution andcounterstained with routine hematoxylin.

Analysis of Angiocidin Immunostaining

Stained slides were digitized using an automated slide scanner(Bacus Laboratories, Lombard, IL). Digital images of each core wereassigned coordinates that corresponded to those in the tissue arraydatabase. Digital images of each core were reviewed by a blindedpathologist and scored for immunoreactivity using a 0 to 3 pointsemiquantitative scoring system for intensity of staining and percentof positive cells, or labeling frequency percentage. Intensity rangedfrom no detectable signal (0) to strong signal (3). Values of 1 and 2corresponded to weak and moderate signal, respectively. Labelingfrequency was scored as 0 (0%), 1 (1 to 33%), 2 (34 to 66%), or 3 (67to 100%). The staining index was obtained by multiplying intensityscore by labeling frequency score, resulting in a possible range of 0 to9. In those instances that duplicate cores from the same specimenreceived different scores, the higher score from each category wasused in calculating the staining index.

Cell Culture and Reagents

All reagents were purchased from Invitrogen (Carlsbad, CA), un-less otherwise specified. The highly metastatic human KM12L4 coloncarcinoma cell line was purchased from Dr. Isaiah J. Fidler at theUniversity of Texas M.D. Anderson Cancer Center. The cell line wasmaintained as a monolayer in Dulbecco’s modified Eagle medium sup-plemented with 10% fetal bovine serum (Sigma Pharmaceutical, SaintLouis, MO), vitamins, sodium pyruvate, L-glutamine, and nonessentialamino acids. Monolayer cultures were maintained at 37°C in a humid-ified atmosphere containing 5% CO2. When cells reached 50% conflu-ence, they were harvest by brief treatment with 0.25% trypsin-0.02%ethylenediamine tetraacetic acid. Only single-cell suspensions with aviability exceeding 90% on trypan blue exclusion were used for the invivo studies.

Peptides

The 6-mer CSVTCG peptide was synthesized by the protein corelab at the Baylor College of Medicine, Houston, TX and was shown tobe �95% pure as assess by high-performance liquid chromatography.The 25-mer peptide was synthesized by Alpha Diagnostic Interna-tional (San Antonio, TX) and was shown to be �80% pure by high-performance liquid chromatography. Sequence derivation of the 25-mer peptide has been previously described [5].

Human Xenograft Splenic Injection Model of Colon Cancer

All experiments were carried out using female BALB/c nu/nu mice(NCI-Charles River, Frederick, MD) at 8 to 10 wk of age (12 to 18 gbody weight). Mice were housed under pathogen-free conditions inlaminar flow rooms and given food and water ad libitum. Experimentalprotocols were approved by the Baylor College of Medicine InstitutionalAnimal Care and Use Committee. All surgical procedures were doneunder sterile conditions. Anesthesia was administered using 2% isoflu-

rane in 100% O2 and maintained with a snout cone. An approximately

322 JOURNAL OF SURGICAL RESEARCH: VOL. 142, NO. 2, OCTOBER 2007

10 mm incision was made in the left flank and the spleen was exteri-orized. KM12L4 cells (1 � 106) suspended in 50 �l serum-free DMEMwere injected into the spleen using a 27 gauge needle. Gentle pressurewith a cotton-tipped applicator was applied for 1 min at the injectionsite to obtain hemostasis and to prevent intraperitoneal leakage oftumor cells. The spleen was returned to the abdominal cavity and theperitoneum and skin were closed in two layers using running 5-0absorbable suture. Animals recovered in warmed, oxygenatedchambers until they resumed upright posture. Mice were weighedweekly and observed for tumor growth and external signs of tumorburden.

Evaluation of Peptide Treatment Efficacy

Twenty-four hours after tumor cell injection, mice were randomizedinto three groups (6 to 9 mice per group): (a) PBS control, 10 mL/kg bodyweight; (b) 6-mer peptide, 10 mL/kg body weight at a dose of 10 mg/kg;and (c) 25-mer peptide, 10 mL/kg body weight at a dose of 10 mg/kg.Peptide dosages were based on previous in vivo studies using murinemodels [5, 12]. Injections were given via the i.p. route and dosed everyother day for 4 wk. All mice underwent necropsy 28 to 30 d after tumorcell injection. Euthanasia was performed using CO2 asphyxiation.

Treatment efficacy was assessed based on volume of the primarytumor nodule and signs of tumor burden. Two semiquantitativescoring methods were used for assessing tumor burden. The first ofthese scores, the internal disease score (IDS), was a measure of grosstumor burden at necropsy and was based on a 4-point scale. An IDSof 1 represented disease localized solely to the spleen. Additionalpoints accrued for each organ containing tumor, whether as a result ofcontiguous spread to the subcutis, pancreas, stomach, kidney, bowel, ordistant metastasis to the liver. Distant metastases were not grosslyevident elsewhere in the abdominal or thoracic cavities. The maximumscore was 4 despite tumor involvement of more than four organs. Byusing the IDS to measure tumor burden, liver metastases are taken intoaccount in the context of overall internal disease.

The second scoring system, the health score (HS), was deter-mined at the end of the 4-wk treatment period and was a measureof external signs of disease based on three criteria: nuchal fat,activity level, and externally appreciable tumor. Development ofascites and extremely large tumors made weight an unreliableindicator for the health of the animal. The HS was also based ona 4-point scale with 1 point being awarded for each of the follow-ing: decreased nuchal fat, absent nuchal fat, decreased activitylevel/weakness, and left upper quadrant tumor on inspection orpalpation. Thus, a score of 1 was given to those animals withminimal evidence of external disease and 4 was given to sickanimals with no nuchal fat, weakness, and a large tumor nodulevisible in the left upper quadrant. Scores of 2 and 3 represent mildand moderate disease, respectively.

Angiocidin Expression in Tumor Xenografts

Angiocidin expression in tumor xenografts was determined usingWestern blot. Fragments of primary tumor from the spleen as well asmetastatic liver nodules from experimental animals were stored inComplete Protease Inhibitor (Roche Diagnostics, Mannheim, Germany)and snap-frozen in liquid nitrogen at the time of necropsy. Sampleswere ground with mortar and pestle in lysis buffer (Cell SignalingTechnology, Danvers, MA) and then homogenized using a polytron (oneburst for 20 s). Samples were incubated for 1 h at 4°C and thencentrifuged. Protein levels of the tissue extracts were determined byBCA protein analysis (Bio-Rad Laboratories, Richmond, CA). Sampleextracts containing equal amounts of total protein (50 �g) were frac-tionated by 12% sodium dodecyl sulfate-polyacrylamide gel electro-phoresis and then transferred to nitrocellulose membranes. Nonspecificbinding was blocked with 5% milk in PBS containing 0.05% Tween 20overnight. The immunoblots were then incubated with 1:500-dilutedpolyclonal rabbit antihuman angiocidin antibody for 3 h at room tem-

perature in PBS with Tween 20. After washing, the immunoblots were

incubated with horseradish peroxidase-conjugated antirabbit IgG (CellSignaling Technologies, Danvers, MA) for 45 min. The bound antibodieswere detected with an enhanced chemiluminescence system (Amer-sham, Arlington Heights, IL).

Statistical Analysis

Treatment efficacy data are expressed as the mean � the stan-dard error of the mean and were tested for statistically significantdifferences between groups using a one-way analysis of variance. Allanalyses were done using GraphPad Prism software (GraphPadPrism Software, Inc., San Diego, CA).

RESULTS

Angiocidin Expression in Human Colorectal Cancer Specimens

A total of 159 human colorectal cancer cases wereevaluated for angiocidin expression using the tissuearray. Angiocidin expression localized to the cell cyto-plasm and the nucleus in both the stromal and epithe-lial components of the specimens. Eighty-nine percent(124/140) of primary tumors expressed angiocidin with40% of those (50/124) showing moderately to stronglypositive expression (staining index: 6 –9). Matchednormal colonic specimens were negative 94% of thetime (101/107) (Fig. 1A and C). Histologically normallymph nodes were negative in 52% (49/95) of speci-mens. Lymph nodes containing tumor metastasesstained positively for angiocidin in 91% of specimens(41/45) (Fig. 1B and C). Specimens of metastatic livernodules were present in 11 cases; all stained posi-tively for angiocidin (Fig. 1B and C).

Effect of Angiocidin Inhibitory Peptides on Tumor Burden

Two semiquantitative scoring systems were used forassessing tumor burden: IDS and animal HS (Figs. 2, 3).Both peptide treatment groups showed improvements inHS and IDS over controls (Fig. 4). Animals treatedwith 6-mer peptide showed a minimum 30% improve-ment in both scores compared with control animals. Inthe 25-mer treatment group, both scores improved byat least 50% compared with control animals. Thesedifferences were statistically significant (P � 0.001).

Effect of Angiocidin Inhibitory Peptides on PrimaryTumor Volume

All PBS-treated control mice (6/6) developed palpa-ble tumor nodules in the left upper quadrant within 14to 20 d after tumor cell injection into the spleen;masses became externally visible in all control miceafter 19 to 23 d. Treatment with either the 6-mer or the25-mer angiocidin-inhibitory peptide significantly re-duced the growth of tumor xenografts. Although allmice in the 6-mer group (8/8) developed palpable tumornodules, these appeared at days 19 to 27, 5 to 7 d laterthan in control mice. Tumor growth was reduced even

more in 25-mer peptide-treated animals where palpa-

323LIEBIG ET AL.: ANGIOCIDIN INHIBITORY PEPTIDES IN COLON CANCER

ble nodules formed in only 44% (4/9) of mice by the endof the 4-wk treatment period.

Six-mer and 25-mer peptide-treated animals showed69% and 94% reductions in primary tumor volume, re-spectively, when compared with controls (Fig. 5). These

FIG. 1. Differential angiocidin expression in tumor versus norexpression in tumor versus normal colonic tissue. A staining indexlabeling frequency score. Frequency distributions of each staining indnormal colonic tissue. Tumor specimens showed angiocidin overexmetastatic versus normal lymph nodes and metastatic liver lesions.benign lymph nodes (LN-N), positive nodes containing colorectal calymph nodes overexpressed angiocidin compared with negative lymImmunohistochemical staining for angiocidin in human colon cancerand positive lymph nodes (if available in the resection specimen) werewas differentially expressed in tumor versus normal specimens. Normof figure is available online.)

differences were statistically significant (P � 0.007).

Angiocidin Expression in Human Colon Cancer Xenografts

Western blots were used to analyze angiocidin expres-sion in xenografts. Angiocidin was expressed in all of theprimary tumors and metastatic liver nodules (Fig. 6A).

l specimens in a human colon cancer tissue array. (A) Angiocidine was obtained for each core by multiplying the intensity score by

value were plotted for colorectal cancer primary tumors and adjacentession compared with normal colon. (B) Angiocidin expression inequency distributions of each staining index value were plotted forr metastases (LN-T), and liver metastases (Liver Mets). Metastaticnodes and all liver metastases showed angiocidin expression. (C)

ue array specimens. Primary tumor, matched normal colon, negativeined for angiocidin. Liver metastases were also included. Angiocidinliver specimens were not included in the tissue array. (Color version

mavaluexprFr

nceph

tissstaal

Primary tumors expressed almost twice as much angio-

ase

324 JOURNAL OF SURGICAL RESEARCH: VOL. 142, NO. 2, OCTOBER 2007

cidin as metastatic liver lesions when blots were com-pared using computerized image analysis (Fig. 6B). Wealso saw a correlation between angiocidin expression andtumor burden in peptide-treated mice. Angiocidin expres-sion in primary tumors was nearly 50% greater in treatedmice with combined HS � IDS of 4 or greater (n � 2)versus those with combined scores of 3 or less (n � 6)(P � 0.05). Angiocidin expression in PBS-treated controlmice did not differ by tumor burden scores.

DISCUSSION

Angiocidin was initially identified in human lungadenocarcinoma cells as a binding protein for theCSVTCG sequence in the TSP-1 molecule [4]. The full-length cDNA of angiocidin was recently cloned in ourlaboratory and the recombinant protein extensivelystudied in in vitro and in vivo models [11]. On itscell-bound form, angiocidin has emerged as a key me-diator of tumor progression and angiogenesis. Thepro-tumor effects of angiocidin are the result of myriadhigh-affinity binding interactions with matrix proteinsincluding integrins, Type 1 collagen, and TSP-1 [4, 5, 11].These angiocidin-stromal protein binding interac-tions lead to an up-regulation in gelatinase expres-sion and matrix remodeling, critical elements of bothangiogenesis and metastasis [3, 10]. In its solubleform, however, angiocidin is a potent inhibitor ofangiogenesis and tumor cell invasion [5].

FIG. 2. IDS. IDS was semiquantitatively measured on a 4-poindistant metastases or local invasion of adjacent organs (A). A score ofas the liver (B). Scores of 3 and 4 are illustrated in (C) and (D), respinvolvement of the pancreas and metastases to the liver (C) or invas(D). Thus, an IDS of 4 was possible without distant metastatic dise

FIG. 3. Animal HS. The HS was also a semiquantitative measuresigns of disease severity. A score of 1 was given to animals with decrefat with a visible tumor was given a score of 2 (B), and if activity le

weakness, and visible tumor was given an HS of 4 (D). (Color version o

Angiocidin overexpression has been shown in anumber of human malignancies, including cancers ofthe breast, lung, prostate, head and neck, and pan-creas [4, 13, 14]. Preliminary studies have also dem-onstrated angiocidin overexpression in colorectalcancer tissues, where higher expression correlatedwith worse tumor phenotypes [15]. Furthermore,high circulating levels of angiocidin can be detectedin the sera of cancer patients, including colon cancerpatients; healthy human subjects have undetectablelevels [16].

In the present study, we investigated the expressionpatterns of angiocidin in human colorectal adenocarci-nomas using a tissue array. We found that angiocidinis overexpressed in colon cancers. We also showed thatangiocidin expression is maintained in colon cancermetastases; positive regional lymph nodes and distantmetastases to the liver overexpressed angiocidin. Al-most 50% of histologically normal lymph nodes ex-pressed angiocidin and the cause of this remains to beelucidated. We know that tumor-associated lympho-cytes within the stroma express angiocidin [15]. Theappearance of these activated lymphocytes may ex-plain why histologically normal regional lymph nodesin cancer patients show angiocidin overexpression. It isalso possible, however, that these nodes containangiocidin-expressing tumor cells without gross histo-logical evidence of tumor.

cale. A score of 1 represents tumor localized to the spleen withoutepresents disease in the spleen as well as one additional organ, suchively, where there is a large primary nodule in the splenic bed withof the subcutaneous tissue and colon (shown in inset) as occurred in. (Color version of figure is available online.)

umor burden based on a 4-point scale that took into account externald nuchal fat but otherwise healthy-appearing (A). Decreased nuchalwas decreased, the score was 3 (C). An animal with no nuchal fat,

t s2 rection

of tasevel

f figure is available online.)

325LIEBIG ET AL.: ANGIOCIDIN INHIBITORY PEPTIDES IN COLON CANCER

We have developed two angiocidin inhibitory pep-tides, and in the second portion of this study we inves-tigated these molecules for their potential therapeuticvalue in colon cancer. The first of these peptides is a6-mer peptide with the sequence CSVTCG. This se-quence is homologous to the angiocidin binding site onthe TSP-1 molecule, a tumor-associated stromal pro-tein known to mediate cell adhesion and tumor pro-gression [17, 18]. Loss of this binding interaction viadeletion mutants missing the CSVTCG binding do-main resulted in decreased tumor cell invasiveness [4].Similarly, the recombinant 6-mer CSVTCG peptide in-hibited tumor cell invasion in vitro and tumor metas-tasis in a mouse melanoma model [12].

The second antiangiocidin peptide we investigated isa 25-mer peptide derived from collagen and integrinbinding sites within the angiocidin molecule. We haverecently shown that angiocidin binds with high affinityand specificity to Type 1 collagen and its cell-surfacereceptors, the �1�1 and �2�1 integrins [5]. Recombi-nant, soluble angiocidin exerts potent antitumor and

FIG. 4. Angiocidin inhibitory peptides decrease tumor burden asmeasured by both HS and IDS. (A) Frequency distribution tableshowing HS and IDS by treatment group. Higher IDS and HS scoresindicate more severe disease and worse tumor burden. Scores wereconsistently higher (worse disease) in control animals. (B) AverageHS and IDS by treatment group. Peptide treatment resulted insignificant reductions in tumor burden scores.

antiangiogenic effects through antagonism of this

collagen-integrin binding interaction, both in vitro andin vivo [5]. We identified a 20 amino acid segment ofangiocidin that mediates its antitumor effects; a dele-tion mutant missing this domain failed to bind collagen1 and �2�1 and lost its antitumor activity [4, 5]. Thesynthetic 25-mer peptide containing this sequencemimics the antitumor and antiangiogenic activity ofrecombinant soluble angiocidin.

Our data shows a potent antitumor effect for both ofour angiocidin inhibitory peptides. Comparative eval-uation of the therapeutic efficacy of our two peptidesreveals that the 25-mer peptide had greater antitumoreffects than an equivalent dose of 6-mer peptide. Micein the 25-mer peptide treatment group had less overalltumor burden as shown by their lower HS and IDS andgreater reduction in primary tumor volume. One ex-planation for this may be the increased binding affinityof angiocidin for Type 1 collagen, which is 50 timesmore favorable than its binding affinity for TSP-1 [5],theoretically making the 25-mer peptide more biologi-cally active. Affinity assays have also demonstratedthat the 25-mer peptide acts as a competitive ligand forcell-surface integrins, effectively competing for cellularadhesion to collagen and preventing �2�1-collagen li-gation [5]. While the 6-mer peptide competitively in-hibits the binding of angiocidin to TSP-1, this bindinginteraction may not be as critical in tumorigenicityand, therefore, inhibition carries less of an effect.

We also observed that during the 4-wk treatment,peptides were well tolerated. In fact, our data showedthat peptide-treated mice had less cachexia, were moreactive, and appeared healthier at the 4-wk end point

FIG. 5. The effect of angiocidin inhibitory peptides on primarytumor volume. Six-mer and 25-mer angiocidin-inhibitory peptidesdecrease the growth of colon cancer xenografts. Twenty-four hoursafter injection of KM12L4 human colon cancer cells into the spleensof nu/nu mice, animals began receiving QOD i.p. injections. Micereceived either 10 mL/kg PBS or 10 mg/kg of one of two angiocidin-inhibitory peptides in an equal volume of PBS. Tumor diameterswere measured at necropsy (28 to 30 d after tumor cell injection) andvolumes were calculated. When compared with controls, animalstreated with either of the angiocidin-inhibitory peptides had de-creased primary tumor volumes. Reduction in tumor volume wasgreatest in animals treated with the 25-mer peptide and results werestatistically significant between all three groups by one-way analysis

of variance (P � 0.007).

326 JOURNAL OF SURGICAL RESEARCH: VOL. 142, NO. 2, OCTOBER 2007

than PBS treated control mice. Our angiocidin inhibi-tory peptides target interactions in the tumor stroma,making them cytostatic, rather than cytotoxic, an at-tractive feature in the development of novel small-molecule therapies [19].

In summary, our data provides strong evidence insupport of angiocidin as a therapeutic target in coloncancer. We have shown that angiocidin is overex-pressed in human colon cancer specimens and thatpeptides inhibitory to the activity of angiocidin signif-icantly decrease tumor growth and tumor burdenin vivo. These data suggest that antiangiocidin therapymay one day be incorporated into current chemother-apeutic regimens to decrease tumor burden whileavoiding increased toxicity. Further research intoangiocidin-targeted peptide therapy is warranted.

ACKNOWLEDGMENTS

This work was supported in part by a South Central VA Health

FIG. 6. Angiocidin expression in vivo correlates with diseaseseverity. (A) Western blot analysis of angiocidin expression in pri-mary splenic tumors (top row) and metastatic liver lesions (bottomrow) in human colon cancer xenografts. Protein was extracted fromtissues at the time of necropsy and subjected to Western blot anal-ysis. All tumors expressed angiocidin. Lane 1, recombinant angioci-din; lanes 2–5, protein extracts from PBS control-treated mice; lanes6–9, protein extracts from 6-mer peptide-treated mice; lanes 10–13,protein extracts from 25-mer peptide-treated mice. (B) Angiocidinexpression in peptide-treated mice correlates with tumor burden.Computerized image analysis of blots revealed increased angiocidinexpression in peptide-treated mice with greater tumor burden (com-bined HS � IDS �4) versus those with less tumor burden (combinedscores �4).

Care Network Research Career Development Grant.

REFERENCES1. Jemal A, Siegel R, Ward E, et al. Cancer statistics, 2006. CA

Cancer J Clin 2006;56:106.2. National Cancer Institute, Surveillance, Epidemiology, and

End Results (SEER) Program Fast Stats: Colon and RectumCancer.

3. Yang X, Rothman VL, L’Heureux DZ, et al. Reduction of angio-cidin expression in human umbilical vein endothelial cells viasiRNA silencing inhibits angiogenesis. Exp Mol Pathol 2006;81:108.

4. Zhou J, Rothman VL, Sargiannidou I, et al. Cloning and char-acterization of angiocidin, a tumor cell binding protein forthrombospondin-1. J Cell Biochem 2004;92:125.

5. Sabherwal Y, Rothman VL, Dimitrov S, et al. Integrin �2�1mediates the antiangiogenic and antitumor activities of angio-cidin, a novel tumor-associated protein. Exp Cell Res 2006;312:2443.

6. Sabherwal Y, Rothman VL, Poon RT, et al. Clinical significanceof serum angiocidin levels in hepatocellular carcinoma. CancerLett 2006.

7. Roth JJ, Buckmire MA, Rolandelli RH, et al. Localization ofthrombospondin-1 and its cysteine-serine-valine-threonine-cysteine-glycine receptor in colonic anastomotic healing tissue.Histol Histopathol 1998;13:967.

8. Roth JJ, Sung JJ, Granick MS, et al. Thrombospondin 1 and itsspecific cysteine-serine-valine-threonine-cysteine-glycine recep-tor in fetal wounds. Ann Plast Surg 1999;42:553.

9. Poon RT, Chung KK, Cheung ST, et al. Clinical significance ofthrombospondin 1 expression in hepatocellular carcinoma. ClinCancer Res 2004;10:4150.

10. Albo D, Shinohara T, Tuszynski GP. Up-regulation of matrixmetalloproteinase 9 by thrombospondin 1 in gastric cancer.J Surg Res 2002;108:51.

11. Tuszynski GP, Rothman VL, Papale M, et al. Identification andcharacterization of a tumor cell receptor for CSVTCG, a throm-bospondin adhesive domain. J Cell Biol 1993;120:513.

12. Tuszynski GP, Rothman VL, Deutch AH, et al. Biological activ-ities of peptides and peptide analogs derived from commonsequences present in thrombospondin, properdin, and malarialproteins. J Cell Biol 1992;116:209.

13. Tuszynski GP, Nicosia RF. Localization of thrombospondin andits cysteine-serine-valine-threonine-cysteine-glycine-specific re-ceptor in human breast carcinoma. Lab Invest 1994;70:228.

14. Arnoletti JP, Albo D, Jhala N, et al. Computer-assisted imageanalysis of tumor sections for a new thrombospondin receptor.Am J Surg 1994;168:433.

15. Wakiyama T, Shinohara T, Shirakusa T, et al. The localizationof thrombospondin-1 (TSP-1), cysteine-serine-valine-threonine-cysteine-glycine (CSVTCG) TSP receptor, and matrix metallo-proteinase-9 (MMP-9) in colorectal cancer. Histol Histopathol2001;16:345.

16. Sabherwal Y, Rothman V, Albo D, et al. Angiocidin, an endog-enous inhibitor of angiogenesis, is increased in the sera ofcancer patients. Proceedings of the American Association forCancer Research 2004;45:Abstract 4950.

17. Sargiannidou I, Qiu C, Tuszynski GP. Mechanisms ofthrombospondin-1-mediated metastasis and angiogenesis. SeminThromb Hemost 2004;30:127.

18. Nicosia RF, Tuszynski GP. Matrix-bound thrombospondin pro-motes angiogenesis in vitro. J Cell Biol 1994;124:183.

19. Parulekar WR, Eisenhauer EA. Phase I trial design for solidtumor studies of targeted, noncytotoxic agents: Theory and

practice. J Natl Cancer Inst 2004;96:990.