Functional Identification of Distinct Sets of Antitumor Activities Mediated by the FKBP Gene Family

Functional identification of distinct sets of antitumor activities mediated by the FKBP gene family

Sylvia Fong*, Leslie Mounkes*, Yong Liu*, Michael Maibaum*, Eric Alonzo*, Pierre-Yves Desprez*, Ann D. Thor†, Mohammed Kashani-Sabet‡, and Robert J. Debs*§

*California Paciﬁc Medical Research Institute, San Francisco, CA 94115; †Department of Pathology, University of Oklahoma School of Medicine, Oklahoma City, OK 73104; and ‡Auerback Melanoma Research Laboratory, Cutaneous Oncology Program, University of California San Francisco Cancer Center and Department of Dermatology, University of California, San Francisco, CA 94115

Edited by James E. Cleaver, University of California, San Francisco, CA, and approved September 29, 2003 (received for review April 17, 2003) Assigning biologic function to the many sequenced but still un- with yet unidentified proteins. Fkbp8 was found to be highly characterized genes remains the greatest obstacle confronting the expressed in various tissues including brain, kidney, liver, and human genome project. Differential gene expression profiling testis, whereas moderate expression was observed in lung, routinely detects uncharacterized genes aberrantly expressed in spleen, heart, ovary (9, 10), and activated T cells (unpublished conditions such as cancer but cannot determine which genes are data). In contrast to Fkbp1a, Fkbp8 does not bind to FK506 or functionally involved in such complex phenotypes. Integrating rapamycin, nor does it possess PPIase activity (9, 10). Fkbp8 has gene expression profiling with specific modulation of gene expres- been shown to be selectively down-regulated in less differenti- sion in relevant disease models can identify complex biologic ated, more aggressive Schwannoma tumor cell lines (10). In functions controlled by currently uncharacterized genes. Here, we addition, we found the Fkbp8 gene to be down-regulated in a used systemic gene transfer in tumor-bearing mice to identify highly metastatic B16-F10-pLUC melanoma cell clone, when novel antiinvasive and antimetastatic functions for Fkbp8, and compared with either the parent cell line, or to the low meta- subsequently for Fkbp1a. Fkbp8 is a previously uncharacterized static B16-F10-p65-R clone (ref. 11 and data not shown). These member of the FK-506-binding protein (FKBP) gene family down- findings suggested that FKBP protein(s) might play a role in regulated in aggressive tumors. Antitumor effects produced by cancer cell progression independent of rapamycin. Fkbp1a gene expression are mediated by cellular pathways entirely Experimental Procedures MEDICAL SCIENCES distinct from those responsible for antitumor effects produced by Fkbp1a binding to its bacterially derived ligand, rapamycin. We Plasmid Construction and Purification. For detailed construction of then used gene expression profiling to identify syndecan 1 (Sdc1) the plasmids used, please refer to Supporting Experimental and matrix metalloproteinase 9 (MMP9) as genes directly regulated Procedures, which is published as supporting information on the by Fkbp1a and Fkbp8. FKBP gene expression coordinately induces PNAS web site. the expression of the antiinvasive Sdc1 gene and suppresses the In Vivo Gene Delivery and Analysis of Antitumor Activity. B16-F10 proinvasive MMP9 gene. Conversely, short interfering RNA-medi- cells and 4T1 cells were freshly thawed and grown in 5% FBS in ated suppression of Fkbp1a increases tumor cell invasion and RPMI medium 1640 or 5% FBS in MEM media for 48 h, MMP9 levels, while down-regulating Sdc1. Thus, syndecan 1 and respectively. On day 0, groups of 10 6-week-old female C57BL͞6 MMP9 appear to mediate the antiinvasive and antimetastatic or BALB͞C mice (Simonsen Laboratories, Gilroy, CA) were effects produced by FKBP gene expression. These studies show injected with 25,000 B16-F10 or 50,000 4T1 cells in 200 ␮lof that uncharacterized genes differentially expressed in metastatic culture media, respectively. Three days after tumor cell injection, cancers can play important functional roles in the metastatic each mouse was injected with 25 ␮g of plasmid DNA complexed phenotype. Furthermore, identifying gene regulatory networks to the pure DOTMA (N-[1-(2,3-dioleyloxy)-propyl]-N,N,N- that function to control tumor progression may permit more trimethylammonium chloride) liposomes at a 1 ␮g DNA:24 nmol accurate modeling of the complex molecular mechanisms of this cationic lipid ratio (12). Mice were killed 25 days after tumor cell disease. injection. For B16-F10-inoculated, C57BL͞6 mice, the lungs were dissected out, weighed, and infused transtracheally with 5% KBP12 (Fkbp1a), an extensively characterized member of the neutral buffered formalin in 1ϫ PBS. For 4T1-inoculated FFK-506-binding protein (FKBP) family, is expressed ubiqui- BALB͞C mice, the lungs were processed as described (13). The tously in cells. It was originally identified as a cytosolic receptor number of black, B16-F10 pulmonary tumors, and counter- for the immunosuppressant drugs, FK506 and rapamycin, and stained 4T1 pulmonary tumors (white) were counted under a later shown to be involved in cell-cycle regulation and in dissecting microscope by an individual blinded to the identity of intracellular calcium homeostasis (1–3). Binding of Fkbp1a to the treatment groups. The potential statistical significance of rapamycin through its FK506͞rapamycin-binding͞peptidyl- differences was assessed by using an unpaired two-tailed Student prolyl isomerase (PPIase) domain mediates the immunosuppres- t test. sive and antitumor effects of rapamycin (1, 4). Rapamycin has previously been documented to exert significant antitumor ef- G1 Cell-Cycle Analysis Using Fluorescence-Activated Cell Sorting fects against several different murine tumors, mediated by its Analysis. For detailed procedures on analyzing percentage of cells in G1 cell cycle, please refer to Supporting Experimental ability to arrest tumor cells in G1 phase of the cell cycle and to reduce tumor angiogenesis (1, 4–8). CCI-779, a soluble ester Procedures. analog of rapamycin, will need to be evaluated against several

different tumor types in human clinical trials (5). This paper was submitted directly (Track II) to the PNAS ofﬁce. Conversely, FKBP-related 38-kDa protein (Fkbp8) has no Abbreviations: FKBP, FK-506-binding protein; RNAi, RNA interference; CLDC, cationic identified function, but has been assigned to the FKBP family, liposome:DNA complex; Sdc1, syndecan 1; MM9, matrix metalloproteinase 9. ͞ based on its extensive amino acid homology to the FK-506 §To whom correspondence should be addressed at: California Paciﬁc Medical Center rapamycin-binding͞PPIase domain of Fkbp1a (9). This protein Research Institute, Stern Building, 2330 Clay Street, San Francisco, CA 94115. E-mail: also contains a three-unit tetratricopeptide repeat and a leucine- [email protected]. zipper repeat, suggesting that it may form multimeric complexes © 2003 by The National Academy of Sciences of the USA

www.pnas.org͞cgi͞doi͞10.1073͞pnas.2332307100 PNAS ͉ November 25, 2003 ͉ vol. 100 ͉ no. 24 ͉ 14253–14258 Downloaded by guest on September 29, 2021 Fig. 1. An i.v. injection of CLDC containing either the human Fkbp8 or human Fkbp1a cDNAs produces signiﬁcant antitumor effects against metastatic murine melanoma and mammary carcinoma lines grown in tumor-bearing mice. (A and B) Comparison of lung weights (e, left y axis) and number of lung tumors (f, right y axis) in groups of 10 tumor-bearing C57BL͞6(A) or BALB͞C mice (B), respectively, injected with luciferase pLuc (control), pFkbp8, or pCC3 plasmid DNA complexed to cationic liposomes. (A) *, P Ͻ 0.05 versus pLuc lung weight; **, P Ͻ 0.01 versus pLuc lung weight; &, P Ͻ 0.005 versus pLuc lung tumor number; #, P Ͻ 0.0005 versus pLuc lung tumor number. (B) *, P Ͻ 0.05 versus pLuc lung weight; **, P ϭ 9.43E-06 versus pLuc lung tumor number; &, P ϭ 3.24E-07 versus pLuc lung tumor number. (C) Western blot analysis of Fkbp1a protein levels in lung extracts (three from each treatment group) isolated 24-hour post i.v. injection of CLDC. (D) Comparison of lung weights (e, left y axis) and number of lung tumors (f, right y axis) in pLuc-, pFkbp1a-, or pCC3-injected BALB͞C mice. *, P Ͻ 0.01 versus pLuc lung weight; **, P Ͻ 0.05 versus pLuc lung weight; &, P Ͻ 0.05 versus pLuc lung tumor number. At least two independent experiments were carried out for each study, yielding comparable results.

Western Analysis. For detailed conditions used in Western anal- Results ysis, please refer to Supporting Experimental Procedures. Because FKBP genes are differentially expressed in aggressive tumors, we used systemic, cationic liposome:DNA complex Real-Time Quantitative PCR. Because the antibody against Fkbp8 is (CLDC)-based gene transfer to assess the role of FKBP gene not available, expression levels were documented by real-time expression in regulating the progression of metastatic tumors in quantitative PCR. Detailed procedure is available in Supporting mice. Systemic CLDC-based gene transfer can produce biolog- Experimental Procedures. ically active levels of delivered gene products in a significant percentage of both tumor cells and normal cells present in the Measurement of IL-2 Production. For detailed protocol on IL-2 lungs of tumor-bearing mice (11, 12). The ability to transfect measurement after transfection or rapamycin treatment of Ju- normal cells, as well as tumor cells, is important, because tumors rkat cells, please refer to Supporting Experimental Procedures. metastasize not simply due to alterations of gene expression within tumor cells, but, rather, from the complex interplay of RNA Interference (RNAi) Construction and Transfection. Fkbp1a altered gene expression patterns within critical normal cell types RNAi was synthesized in vitro by using the Silencer short and tumor cells (14, 15). Cationic liposomes were complexed to interfering RNA (siRNA) construction kit (Ambion, Austin, an Epstein–Barr virus-based expression plasmid that can express TX), per the manufacturer’s instructions. The sense DNA oligo delivered genes at therapeutic levels for prolonged periods in used in in vitro transcription was: 5Ј-AATAGGCATAGTCT- immunocompetent mice (16). GAGGAGACCTGTCTC-3Ј. A negative control RNAi sequence was designed by randomizing the above oligonucleotide FKBP Gene Expression Reduces Tumor Progression Functions in Tumor- sequence to obtain 5Ј-CCTGGGTAGTCTAAATAGAAGG- Bearing Mice. First, we i.v. injected groups of 10 C57BL͞6 mice TAGCCTC-3Ј and the negative control RNAi was synthesized by bearing murine B16-F10 melanoma tumors with CLDC contain- following the same protocol as the Fkbp1a RNAi. A total of 25 ing a human cytomegalovirus-driven expression plasmid linked nM RNAi was transfected by using siPORT lipid transfection to either the human Fkbp8 cDNA, the human CC3 cDNA (a reagent (Ambion) per the manufacturer’s instructions. potent antitumor gene; refs. 12 and 17), or the luciferase cDNA (mock-treated controls). [Mock-treated and untreated tumor- In Vitro Invasion Assay. For detailed methods on assessing invasive bearing mice show comparable numbers of metastatic B16-F10 potential of tumor cells in vitro, see Supporting Experimental tumors (12).] Systemic delivery of the pFkbp8 or pCC3 plasmids Procedures. each significantly reduced the metastatic progression of B16-F10 tumors in syngeneic C57BL͞6 mice, as determined by both whole Microarray Analysis. The cDNA microarray analysis was carried lung weights and total number of lung tumors (Fig. 1A). To out by following protocols available at http:͞͞derisilab.ucsf.edu͞ ascertain whether Fkbp8 gene expression could also control the microarray. For detailed protocol, see Supporting Experimental metastatic progression of 4T1 tumor cells, an aggressive mouse Procedures. breast cancer line, we then i.v. injected CLDC containing the

14254 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.2332307100 Fong et al. Downloaded by guest on September 29, 2021 MEDICAL SCIENCES

Fig. 2. Unlike rapamycin, neither Fkbp1a nor Fkbp8 overexpression arrests tumor cells in G1, nor does Fkbp1a overexpression suppress IL-2 production in activated Jurkat cells. Fluorescence-activated cell sorting analysis of transfected 4T1 (A) and B16-F10 (B) cells in G1 phase of cell cycle at 24, 48, and 72 h. Luciferase-transfected control cells treated with 1 ␮g͞ml rapamycin are shown in the far right column. (C) IL-2 secretion in the media of transfected Jurkat cells in the presence or absence of 50 nM rapamycin. &, P ϭ 6.89924E-07 versus luciferase control (no rapamycin); *, P ϭ 5.13644E-09 versus luciferase control (no rapamycin). (D) Western blot analysis of Fkbp1a expression in transfected Jurkat cells in the presence or absence of rapamycin.

Fkbp8, CC3, or luciferase cDNAs into groups of 10 BALB͞c B16-F10 melanoma cells, stably transfected with either Fkbp1a, mice bearing 4T1 mammary carcinoma tumors (12, 18). As seen Fkbp8, or luciferase were inconclusive, because overexpression with B16-F10 tumors, systemic delivery of the pFkbp8 or CC3 of Fkbp1a or Fkbp8 was rapidly lost in tumor cells after their plasmids significantly reduced the metastatic progression of 4T1 i.v. injection into mice (data not shown). Thus, we found tumors in syngeneic mice (Fig. 1B). Thus, the expression of that expression of either Fkbp8 or Fkbp1a, two quite different Fkbp8, a previously uncharacterized gene, can play a significant members of the FKBP gene family, can regulate metastatic functional role in controlling tumor metastasis. progression in the absence of rapamycin. Because Fkbp8 does Because systemic delivery of the Fkbp8 gene reduced the not bind rapamycin (9, 10), the antimetastatic effects produced metastatic progression of two different highly aggressive mouse by increased expression of the Fkbp8 appear independent of tumor cell types (Fig. 1 A and B), we then tested whether those mediated by Fkbp1a-rapamycin binding. Conversely, by up-regulation of the human Fkbp1a gene, the best characterized using the same dose, schedule, and parenteral route of rapamy- member of the FKBP gene family, could produce similar anti- cin administration that has previously been shown to produce tumor progression effects in mice. Although Fkbp1a has been antitumor effects in vivo (7), we found that rapamycin failed to shown to mediate antitumor effects through binding to rapamy- significantly (P Ͼ 0.05 in all cases) reduce either the total lung cin, its immunosuppressive, bacterial-macrolide ligand, regula- weights or lung tumor numbers of B16-F10 or 4T1 tumors in tion of Fkbp1a gene expression per se has not previously been mice (data not shown). Thus, unlike FKBP gene expression, shown to alter the malignant phenotype. We first documented rapamycin treatment is not significantly active against either 4T1 that CLDC-based systemic delivery of plasmid pFkbp1a in- or B16-F10 tumors in mice. creased Fkbp1a protein levels in the lungs of tumor-bearing mice, by using Western analysis (Fig. 1C). We then showed that Cellular Pathways Mediating the Antitumor Effects of FKBP Gene CLDC-based systemic delivery of plasmid pFkbp1a significantly Expression Are Distinct from Rapamycin-Induced Pathways. The final reduced the metastatic spread of 4T1 mammary carcinoma common mechanism through which the rapamycin:Fkbp1a com- tumors in BALB͞c mice (Fig. 1D). In addition, systemic CLDC- plex produces its antiproliferative effects is by arresting tumor based delivery of pFkbp1a reduced the progression of B16-F10 cells in G1 (1). Fluorescence-activated cell sorting analysis melanoma metastases in C57BL͞6 mice, as demonstrated by documented that rapamycin treatment (1 ␮M) of 4T1 mammary significant reductions of lung weights (P Ͻ 0.05) and of tumor carcinoma (Fig. 2A) and B16-F10 melanoma cells (Fig. 2B) Ͻ numbers (P 0.05; data not shown). Preliminary studies com- induced significant G1 arrest. The same levels of G1 cell-cycle paring the number of lung metastases after i.v. injection of arrest were achieved with lower doses of rapamycin (0.01 and 0.1

Fong et al. PNAS ͉ November 25, 2003 ͉ vol. 100 ͉ no. 24 ͉ 14255 Downloaded by guest on September 29, 2021 Table 1. Quantitation of Fkbp1a or Fkbp8 overexpression in Fkbp1a was originally identified as the major cellular receptor transfected cells mediating the potent suppression of T lymphocyte function Genes Fkbp1a, Fkbp8, ␤-actin, produced by FK-506 or rapamycin treatment (1, 4). Therefore,

Cell line transfected mean CT* mean CT mean CT we compared the effects on T lymphocyte function of Fkbp1a overexpression versus rapamycin treatment. Rapamycin treat- B16 Luciferase 32.45 Ϯ 0.13 36.69 Ϯ 0.35 17.31 Ϯ 0.30 ment significantly reduced IL-2 secretion from Jurkat human T Fkbp1a 19.02 Ϯ 0.19 ND 17.24 Ϯ 0.24 lymphocytes (Fig. 2C). In contrast, overexpression of Fkbp1a Fkbp8 ND 19.98 Ϯ 0.10 17.07 Ϯ 0.13 (Fig. 2D) significantly increased IL-2 secretion from Jurkat cells 4T1 Luciferase 26.28 Ϯ 0.06 32.18 Ϯ 0.48 17.48 Ϯ 0.19 (Fig. 2C). Fkbp1a 19.25 Ϯ 0.03 ND 16.91 Ϯ 0.08 We attempted to identify the cellular mechanisms by which Fkbp8 ND 19.28 Ϯ 0.02 17.50 Ϯ 0.09 Fkbp1a gene expression produces its antitumor effects. FKBP

*CT values were expressed as the mean CT Ϯ SE of the mean. CT, the threshold gene expression does not alter tumor apoptosis or mitosis, cycle number. ND, not determined. CT is the point at which the ﬂuorescence because tumor cell apoptosis and proliferation in lung tumors signal rises above the base line ﬂuorescence and begins to increase expo- obtained from B16-F10 tumor-bearing mice receiving systemic nentially. The CT value is in logarithmic inverse relationship with the abun- CLDC-Fkbp1a did not differ from control tumor-bearing mice dance of the transcripts, based on the assumption that CT values increase (data not shown). Furthermore, there was no difference in tumor Ϸ by 1 for each 2-fold dilution. Total RNA from the transfected cells were angiogenesis between tumor-bearing mice receiving systemic isolated and subjected to real-time PCR as described in Experimental Proce- CLDC-Fkbp1a versus those receiving systemic CLDC-Luc (data dures. Using Fkbp1a and Fkbp8 primer sets, we determined the steady-state amounts of these transcripts present in the samples. We used ␤-actin as the not shown). Intravenous CLDC-based injection of genes with endogenous reference to indicate similar amounts of total RNA were used in known antiangiogenic activities, including angiostatin and p53, each reverse transcription and PCR. each significantly reduced tumor angiogenesis in the same B16-F10 tumor model in which i.v. CLDC-based injection of the Fkbp1a gene failed to do so (12). Thus, neither tumor cell ␮M) (data not shown). In contrast, neither 4T1 nor B16-F10 cells proliferation, apoptosis, nor tumor angiogenesis were altered by transfected with, and documented to overexpress high levels of overexpression of Fkbp1a. either Fkbp1a or Fkbp8 (Table 1), exhibited any evidence of G1 cell-cycle arrest (Fig. 2 A and B). Furthermore, [3H]thymidine FKBP Gene Expression Reduces Tumor Cell Invasion. We then as- uptake studies showed that 4T1 and B16-F10 cells overexpress- sessed whether Fkbp1a overexpression altered the ability of ing Fkbp1a or Fkbp8 remained in exponential growth, whereas B16-F10 melanoma cells, 4T1 cells, human MDA-MB-231 mam- rapamycin-treated cells were growth-arrested (data not shown). mary carcinoma cells, or human PPC-1 prostate cancer cells to Thus, unlike Fkbp1a-rapamycin binding (1), overexpression of invade the extracellular matrix (ECM). Overexpression of either Fkbp1a or Fkbp8 does not induce cell-cycle arrest in tumor Fkbp1a (Fig. 3A) reduced tumor cell invasion into matrigel by cells. Ն50% for B16-F10, 4T1, and PPC-1 cells, and by Ͼ35% for

Fig. 3. Increased expression of Fkbp1a or Fkbp8, but not rapamycin treatment, reduces tumor invasiveness.(A) Western blot analysis of Fkbp1a overexpression in each transfected cancer cell line. Lanes 1, 3, 5, and 7 show luciferase-transfected cells, and lanes 2, 4, 6, and 8 show Fkbp1a-transfected cells. (B) Boyden chamber invasion assay showing B16-F10, 4T1, MDA-MB231, and PPC-1 cells transfected with luciferase control (f) or Fkbp1a (e) plasmid DNA. Data are presented as relative invasiveness, where the respective controls are set as 100% (f). *, P Ͻ 0.05 versus luciferase control; **, P ϭ 0.06 versus luciferase control. (C) Western blot analysis of Fkbp1a protein levels in RNAi-transfected B16-F10 cells. Lane 1, control RNAi-transfected; lane 2, Fkbp1a-targeting RNAi-transfected. (D) Boyden chamber invasion assay with B16-F10 cells transfected with control RNAi and Fkbp1a-targeting RNAi. *, P Ͻ 0.0001 versus control RNAi.

14256 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.2332307100 Fong et al. Downloaded by guest on September 29, 2021 MDA-MB-231 cells (Fig. 3B). Fkbp8 overexpression inhibited invasion (P Ͻ 0.01) into matrigel at a level comparable to that produced by Fkbp1a (data not shown). Conversely, biologically active concentrations of rapamycin (0.01–1 ␮g͞ml; ref. 7) failed to inhibit tumor cell invasion (data not shown). We then used anti-Fkbp1a-specific siRNA oligonucleotides to assess the effects of suppressing Fkbp1a gene expression on tumor invasion. The anti-Fkbp1a, but not the control siRNA, substantially reduced Fkbp1a protein levels (Fig. 3C). RNAi-mediated suppression of Fkbp1a in B16-F10 tumor cells increased their ability to invade into matrigel by 2.5-fold (P Ͻ 0.0001; Fig. 3D). Taken together, our results show that increasing Fkbp1a gene expression reduces tumor cell invasion, while suppressing Fkbp1a enhances invasion.

FKBP Genes Regulate a Network of Adhesion͞Invasion-Related Genes. We then performed cDNA microarray analysis to characterize genes differentially expressed in B16-F10 cells by rapamycin treatment versus Fkbp1a overexpression. These studies revealed that the two interventions produced widely divergent patterns of gene expression (P ϭ 0.1E-62; data not shown), which was consistent with the distinct antitumor pathways they mediate. We then used this analysis to identify specific genes regulated by Fkbp8 and Fkbp1a in tumor cells. Because FKBP genes reduce tumor cell invasion (Fig. 3A), we identified strongly FKBP- regulated genes that were also known to play important functional roles in cellular invasion. Syndecan 1 (Sdc1) mRNA was consistently up-regulated 4–5 fold in B16-F10 cells by either Fig. 4. Regulation of Sdc1 and MMP9 serves as potential molecular mechanism of the antiinvasion effects produced by the FKBP proteins. (A) Regula- MEDICAL SCIENCES Fkbp8 or Fkbp1a gene expression, but not by rapamycin treat- tion of Sdc1 and MMP9 by the level of Fkbp1a protein in transfected B16-F10 ment (data not shown). Sdc1 protein levels were coordinately cells. Lane 1, no rapamycin treatment; lane 2, 1 ␮g͞ml rapamycin treatment; increased in Fkbp1a-transfected B16-F10 cells, whereas anti- lane 3, pLuc control-transfected; lane 4, pFkbp1a-transfected; lane 5, control Fkbp1a RNAi significantly reduced Sdc1 protein levels (Fig. 4A). RNAi-transfected; lane 6, Fkbp1a-targeting RNAi-transfected. Arrow indicates Sdc1 blocks tumor cell invasion directly, by increasing cell Fkbp1a protein with 5ЈHA tag. (B) Regulation of MMP9 by FKBP proteins. adhesion to ECM (19, 20), and by inhibiting the expression of B16-F10-pooled populations of clones stably overexpressing Fkbp1a or Fkbp8 matrix metalloproteinase 9 (MMP9), which degrades ECM were compared with cells stably transduced with vector control for MMP9 substrates (21, 22). protein expression. Lane 1, vector control pooled population; lane 2, Fkbp1a B16-F10 cells overexpressing Fkbp1a demonstrated signifi- overexpressing pooled population; lane 3, Fkbp8 overexpressing pooled population. cantly reduced MMP9 protein levels, whereas RNAi-mediated suppression of Fkbp1a significantly increased MMP9 (Fig. 4A). We further confirmed FKBP mediated up-regulation of Sdc1 gene, Fkbp8 neither possesses PPIase activity nor binds rapa- mRNA levels (Table 2) and down-regulation of MMP9 by stably mycin (9, 10). However, based on Fkbp8’s partial sequence overexpressing Fkbp1a or Fkbp8 in B16-F10 cells (Fig. 4B). homology with Fkbp1a, we used this same approach to show that Blocking MMP9 gene expression significantly reduces the met- Fkbp1a gene expression can also inhibit metastatic progression. astatic spread of B16 melanoma tumors in mice (23). Several Furthermore, antitumor activities produced by FKBP gene observations support the hypothesis that Fkbp8- or Fkbp1a- expression are mediated by mechanisms distinct from those induced antitumor effects are mediated by inhibition of tumor induced by Fkbp1a-rapamycin binding. cell invasion. The lack of effect on tumor apoptosis, mitosis, or Specifically, antitumor effects produced by Fkbp1a binding to angiogenesis in mice successfully treated with Fkbp1a or Fkbp8 rapamycin occur through the induction of cell-cycle arrest (25) gene transfer provides indirect evidence. Furthermore, the ͞ induction of Sdc1 and suppression of MMP9 by FKBP gene and or the inhibition of tumor angiogenesis (7), and can be expression further supports the hypothesis that the antitumor associated with potent T cell-mediated immunosuppression, effects produced are mediated, at least in part, by suppressing depending on the dosing schedule of rapamycin used (26). tumor invasion. Conversely, antitumor effects produced by Fkbp1a gene expression appear unrelated to cell cycle or angiogenic regulation, are Discussion mediated, at least in part, by the inhibition of invasion-related Gene expression profiling studies have shown that large numbers pathways, and are associated with stimulation of T cell function. of genes, including (i) genes known to be functionally involved Binding of FAP48, a Fkbp1a-interacting protein, to Fkbp1a, also in the malignant phenotype, (ii) genes with known biologic increases IL-2 production (27), documenting Fkbp1a-mediated activities not previously linked to cancer, and (iii) genes with no T cell stimulation. Finally, whereas a single systemic injection of identified function, are all aberrantly expressed in cancer (24). CLDC containing either the Fkbp1a or Fkbp8 gene significantly However, it has been difficult to determine which of these reduces the metastatic progression of 4T1 or B16-F10 tumors in aberrantly expressed genes, particularly those with no currently mice, high-dose i.p. injections of rapamycin administered daily identified function, actually control the metastatic spread of for 18 days failed to significantly reduce the progression of either cancer. We combined gene expression profiling with the systemic 4T1 or B16-F10 tumors. Thus, FKBP gene expression and transfer of selected cDNA microarray-identified genes to assign Fkbp1a-rapamycin binding appear to act by different pathways antimetastatic function to Fkbp8, a previously uncharacterized in tumor-bearing mice as well. member of the FKBP gene family that is down-regulated in Although a number of specific genes and cellular pathways aggressive tumors. In contrast to the well characterized Fkbp1a regulated by binding of Fkbp1a to rapamycin have already been

Fong et al. PNAS ͉ November 25, 2003 ͉ vol. 100 ͉ no. 24 ͉ 14257 Downloaded by guest on September 29, 2021 Table 2. Quantitation of Sdc1 up-regulation in B16-F10 cells stably overexpressing Fkbp1a or Fkbp8 genes

† B16-F10 stable pool Genes transfected Fkbp1a, mean CT* Fkbp8, mean CT ␤-Actin, mean CT Sdc1, mean CT Folds up-regulation of Sdc1

Vector None 32.5 Ϯ 0.13 36.7 Ϯ 0.35 18.3 Ϯ c1.14 25.5 Ϯ 0.11 1.0 Fkbp1a Fkbp1a 19.0 Ϯ 0.19 ND 18.9 Ϯ 0.24 24.7 Ϯ 0.20 2.6 Fkbp8 Fkbp8 ND 20.0 Ϯ 0.10 19.1 Ϯ 0.13 25.3 Ϯ 0.29 2.0

*CT values were expressed as the mean CT Ϯ SEM. † ⌬⌬C Sdc1 up-regulation was calculated as 2 T. ⌬⌬CT values were calculated as ⌬CT of Vector-⌬CT of Fkbp1a or Fkbp8. ⌬CT of each pool sample was calculated as Sdc1 Mean CT-␤-actin mean CT of each respective sample.

identified, the downstream effects of Fkbp1a gene expression are different human cancers (30–32), whereas the loss of Sdc1 largely uncharacterized. We initially found that Fkbp8, a mem- expression is associated with increased aggressiveness and me- ber of the FKBP gene family, was significantly down-regulated tastases in a number of human cancers (30, 33, 34). Therefore, by ribozyme-based suppression of NF-␬B expression in a mel- the coordinate up-regulation of Sdc1 and down-regulation of anoma tumor cell line. Because targeting NF-␬B expression MMP9 produced by Fkbp8 or Fkbp1a gene expression may play significantly decreased the metastatic spread of this tumor (11), significant roles in mediating their antiinvasive and antimeta- we then investigated whether overexpressing the Fkbp8 gene or static effects. the Fkbp1a gene in mice bearing this tumor line altered its Aberrant expression in aggressive tumors led us to identify metastatic progression. We found that overexpressing either antimetastatic and antiinvasive effects for Fkbp8, a gene with no Fkbp8 or the Fkbp1a significantly reduced the progression of previously identified function (9), and subsequently for Fkbp1a. B16-F10, as well as of 4T1 tumors in mice (Fig. 1). Overall these FKBP gene expression concurrently induces Sdc1 and suppresses results indicate that the regulation of FKBP gene expression can MMP9, suggesting that a complex network of adhesion and play a significant role in mediating tumor progression in vivo. matrix-remodeling-related genes mediate the abilities of FKBP We then used expression profiling to identify Sdc1 and MMP9 genes to control tumor invasion and metastasis. Genome-wide as FKBP-regulated genes that may mediate the antiinvasive expression profiling studies have revealed the genetic determi- effects produced by Fkbp8 or Fkbp1a gene expression. Fkbp1a nants of tumor metastasis to be extremely complex. The iden- gene expression strongly up-regulates Sdc1, down-regulates tification of gene regulatory networks that function to control MMP9, and decreases tumor invasion, whereas RNAi-mediated tumor metastasis may permit more accurate modeling of the suppression of Fkbp1a significantly down-regulates Sdc1, up- complex molecular mechanisms of this disease. regulates MMP9, and increases tumor invasion. Sdc1, as well as MMP9, have each been shown to be functionally involved in We thank Drs. Nick Lemoine and Martin Matzuk for helpful comments, regulating tumor invasion and tumor metastasis (28, 29). Spe- and Drs. Christina Jamieson and Caroline Mrejen of the Mouse Mi- cifically, targeted disruption of the MMP9 gene has been doc- croarray Consortium at the University of California, San Francisco, for their support on microarray experiments. This work was supported by umented to reduce the metastatic spread of B16 melanoma funds from the California Pacific Medical Center Research Institute, tumors, as well as other cancers in MMP9 knockout mice (23). National Institutes of Health-R01 Grant CA 82575 (to R.J.D.), Califor- Furthermore, increased levels of Sdc1 expression in tumor nia Breast Cancer Research Program Grant 8WB-0614 (to R.J.D.), and biopsies is associated with a favorable prognosis in a variety of funds from the Human Frontier of Sciences Program.

1. Hidalgo, M. & Rowinsky, E. K. (2000) Oncogene 19, 6680–6686. 20. Liu, W., Litwack, E. D., Stanley, M. J., Langford, J. K., Lander, A. D. & 2. Liu, J., Farmer, J. D., Jr., Lane, W. S., Friedman, J., Weissman, I. & Schreiber, Sanderson, R. D. (1998) J. Biol. Chem. 273, 22825–22832. S. L. (1991) Cell 66, 807–815. 21. Stetler-Stevenson, W. G., Aznavoorian, S. & Liotta, L. A. (1993) Annu. Rev. 3. Shou, W., Aghdasi, B., Armstrong, D. L., Guo, Q., Bao, S., Charng, M. J., Cell Biol. 9, 541–573. Mathews, L. M., Schneider, M. D., Hamilton, S. L. & Matzuk, M. M. (1998) 22. MacDougall, J. R., Bani, M. R., Lin, Y., Rak, J. & Kerbel, R. S. (1995) Cancer Nature 391, 489–492. Res. 55, 4174–4181. 4. Sehgal, S. N. (1995) Ther. Drug Monit. 17, 660–665. 23. Itoh, T., Tanioka, M., Matsuda, H., Nishimoto, H., Yoshioka, T., Suzuki, R. & 5. Elit, L. (2002) Curr. Opin. Investig. Drugs 3, 1249–1253. Uehira, M. (1999) Clin. Exp. Metastasis 17, 177–181. 6. Mita, M. M., Mita, A. & Rowinsky, E. K. (2003) Clin. Breast Cancer 4, 126–137. 24. Perou, C. M., Jeffrey, S. S., van de Rijn, M., Rees, C. A., Eisen, M. B., Ross, 7. Guba, M., von Breitenbuch, P., Steinbauer, M., Koehl, G., Flegel, S., Hornung, D. T., Pergamenschikov, A., Williams, C. F., Zhu, S. X., Lee, J. C., et al. (1999) M., Bruns, C. J., Zuelke, C., Farkas, S., Anthuber, M., et al. (2002) Nat. Med. Proc. Natl. Acad. Sci. USA 96, 9212–9217. 8, 128–135. 25. Aghdasi, B., Ye, K., Resnick, A., Huang, A., Ha, H. C., Guo, X., Dawson, T. M., 8. Luan, F. L., Ding, R., Sharma, V. K., Chon, W. J., Lagman, M. & Suthanthiran, Dawson, V. L. & Snyder, S. H. (2001) Proc. Natl. Acad. Sci. USA 98, 2425–2430. M. (2003) Kidney Int. 63, 917–926. 26. Saunders, R. N., Metcalfe, M. S. & Nicholson, M. L. (2001) Kidney Int. 59, 3–16. 9. Lam, E., Martin, M. & Wiederrecht, G. (1995) Gene 160, 297–302. 27. Krummrei, U., Baulieu, E.-E. & Chambraud, B. (2003) Proc. Natl. Acad. Sci. 10. Pedersen, K. M., Finsen, B., Celis, J. E. & Jensen, N. A. (1999) Electrophoresis USA 100, 2444–2449. 20, 249–255. 28. Benassi, M. S., Gamberi, G., Magagnoli, G., Molendini, L., Ragazzini, P., Merli, 11. Kashani-Sabet, M., Liu, Y., Fong, S., Desprez, P. Y., Liu, S., Tu, G., Nosrati, M., Chiesa, F., Balladelli, A., Manfrini, M., Bertoni, F., et al.(2001) Ann. Oncol. M., Handumrongkul, C., Liggitt, D., Thor, A. D. & Debs, R. J. (2002) Proc. 12, 75–80. Natl. Acad. Sci. USA 99, 3878–3883. 171, 12. Liu, Y., Thor, A., Shtivelman, E., Cao, Y., Tu, G., Heath, T. D. & Debs, R. J. 29. Langford, J. K. & Sanderson, R. D. (2001) Methods Mol. Biol. 495–503. (1999) J. Biol. Chem. 274, 13338–13344. 30. Anttonen, A., Heikkila, P., Kajanti, M., Jalkanen, M. & Joensuu, H. (2001) 13. Wexler, H. (1966) J. Natl. Cancer Inst. 36, 641–645. Lung Cancer 32, 297–305. 14. Berns, A. (2001) Nature 410, 1043–1044. 31. Numa, F., Hirabayashi, K., Kawasaki, K., Sakaguchi, Y., Sugino, N., Suehiro, 15. Bissell, M. J. & Radisky, D. (2001) Nat. Rev. Cancer 1, 46–54. Y., Suminami, Y., Hirakawa, H., Umayahara, K., Nawata, S., et al. (2002) Int. 16. Tu, G., Kirchmaier, A. L., Liggitt, D., Liu, Y., Liu, S., Yu, W. H., Heath, T. D., J. Oncol. 20, 39–43. Thor, A. & Debs, R. J. (2000) J. Biol. Chem. 275, 30408–30416. 32. Wiksten, J. P., Lundin, J., Nordling, S., Kokkola, A. & Haglund, C. (2000) 17. Shtivelman, E. (1997) Oncogene 14, 2167–2173. Anticancer Res. 20, 4905–4907. 18. Lin, P., Buxton, J. A., Acheson, A., Radziejewski, C., Maisonpierre, P. C., 33. Fujiya, M., Watari, J., Ashida, T., Honda, M., Tanabe, H., Fujiki, T., Saitoh, Yancopoulos, G. D., Channon, K. M., Hale, L. P., Dewhirst, M. W., George, Y. & Kohgo, Y. (2001) Jpn. J. Cancer Res. 92, 1074–1081. S. E. & Peters, K. G. (1998) Proc. Natl. Acad. Sci. USA 95, 8829–8834. 34. Wiksten, J. P., Lundin, J., Nordling, S., Lundin, M., Kokkola, A., von 19. Liebersbach, B. F. & Sanderson, R. D. (1994) J. Biol. Chem. 269, 20013–20019. Boguslawski, K. & Haglund, C. (2001) Int. J. Cancer 95, 1–6.

14258 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.2332307100 Fong et al. Downloaded by guest on September 29, 2021