Translocating a High-Affinity Designer TIMP-1 to the Cell Membrane for Total Renal Carcinoma Inhibition: Putting the Prion Protein to Good Use

Membrane type 1-matrix metalloproteinase (MT1-MMP) and tumor necrosis factor α (TNF-α)-converting enzyme (TACE) are prominent membrane-anchored metalloproteinases that regulate the turnover of extracellular matrix (ECM) components and bioactive molecules required for cancer proliferation. In this study, we describe a novel approach that would allow tissue inhibitor of metalloproteinase 1 (TIMP-1), the endogenous inhibitor of the matrix metalloproteinases (MMPs), to be translocated to the cell membrane for simultaneous MT1-MMP/TACE inhibition.

ABSTRACT

Membrane type 1-matrix metalloproteinase (MT1-MMP) and tumor necrosis factor α (TNF-α)-converting enzyme (TACE) are prominent membrane-anchored metalloproteinases that regulate the turnover of extracellular matrix (ECM) components and bioactive molecules required for cancer proliferation. In this study, we describe a novel approach that would allow tissue inhibitor of metalloproteinase 1 (TIMP-1), the endogenous inhibitor of the matrix metalloproteinases (MMPs), to be translocated to the cell membrane for simultaneous MT1-MMP/TACE inhibition. We achieve this by fusing T1^TACE, a designer TIMP-1 with superb affinities for MT1-MMP and TACE, to the glycosyl-phosphatidyl inositol anchor of prions to create a membrane-tethered, broad-spectrum inhibitor, named T1^{Pr αTACE}, that colocalizes with MT1-MMP and TACE on the cell surface. Transduction of T1^{Pr αTACE} in human fibrosarcoma cells results not only in a substantial reduction in gelatinolytic and TNF-α/heparin binding epithelial growth factor shedding activities but also in a loss of tubulogenic capability in three-dimensional matrices. In renal carcinoma, T1^{Pr αTACE} triggers cellular senescence and disrupts MMP-mediated proteolysis of ECM components such as fibronectin and collagen I, leading to an impairment in cell motility and survival under both in vitro and in vivo conditions. Taken together, our findings may provide a new perspective in TIMP targeting that could be exploited to halt metastatic renal carcinoma progression.

INTRODUCTION

Metastatic cancers rely on endoproteinases of the matrix metalloproteinase (MMP) and a disintegrin and metalloproteinase (ADAM) families to modulate the surrounding microenvironment and support cellular processes, including invasion and proliferation, differentiation, signaling, cross talk, and angiogenesis (1–3). Among the 25 MMPs and 13 functional ADAMs characterized, membrane type 1-matrix metalloproteinase (MT1-MMP and MMP-14) and tumor necrosis factor alpha (TNF-α)-converting enzyme (TACE and ADAM-17) are unquestionably the best investigated to date. MT1-MMP is the first of the six membrane-anchored MMPs to be discovered and linked to tumor malignancy (4). Like the majority of the MMPs, MT1-MMP is a multidomain proteinase consisting of a prodomain, a conserved catalytic domain in which the canonical zinc-binding motif HEXXHXXGXXH resides, a disclike hemopexin domain, and finally a transmembrane anchor and a short cytoplasmic tail comprised of only 20 amino acids. Substrate-wise, MT1-MMP cleaves a wide range of extracellular matrix (ECM) and basement membrane macromolecules, such as collagens I, II, and III, fibronectin, vitronectin, laminins, fibrinogens, apolipoproteins, and fibrillar amyloid β protein (5, 6). Apart from carrying out pericellular proteolysis, MT1-MMP also shed some of the best-known adhesion and signaling molecules involved in cancer dissemination, such as CD44, syndecans, RANKL, and MUC1 (7–10). Since its first appearance in the literature in 1994, there has been overwhelming evidence linking MT1-MMP to adverse prognosis in many cancer types, including renal carcinoma (8, 11, 12).

Of the 13 ADAMs that encode functional proteinases, TACE is the best described and studied owing to its versatility in the ectodomain shedding of a large repertoire of cell surface ligands required for growth and proliferation. Among the long list of bioactive molecules shed by TACE are proinflammatory cytokines and growth factors (TNF-α, tumor growth factor alpha [TGF-α], and heparin binding epidermal growth factor [HB-EGF]), receptors (tumor necrosis factor receptor II [TNFR II], EGF receptor [EGFR], interleukin-6 receptor [IL-6R], and Notch-1), adhesion molecules (I-CAM-1, V-CAM-1, and l-selectin), and even the prion protein, which causes the fatal brain disorder Creutzfeldt-Jakob disease (13–15). In terms of domain organization, TACE is comprised of a prodomain, a catalytic domain, a disintegrin domain, and an EGF-like domain, followed by a transmembrane anchor and a relatively large cytoplasmic tail of approximately 15 kDa.

Under physiological conditions, the enzymatic activities of the MMPs and ADAMs are regulated by the endogenous inhibitors termed the tissue inhibitors of metalloproteinases (TIMPs). There are four human TIMPs (TIMP-1 to -4), and they are all relatively small proteins between 21 and 24 kDa in mass. TIMPs inhibit the metalloproteinases by inserting their wedge-like N-terminal domains into the catalytic clefts of the target enzymes to form 1:1 stoichiometric complexes that are essentially nondissociable (Protein Data Bank entries 4ILW, 3V96, and 3CKI). With the exception of TIMP-3, the TIMPs are all highly soluble. TIMP-3 is unique in that it has a propensity to associate with the ECM through a congregation of surface lysine and arginine residues (16) and is able to inhibit the ADAM proteinases with C-terminal thrombospondin repeats (ADAM-TSes), such as ADAM-TS4 and -5 (also known as aggrecanases 1 and 2), that would otherwise be intractable to TIMP inhibition (17).

Despite sharing a similar tertiary configuration, TIMPs differ considerably in their metalloproteinase selectivity. TIMP-1, for instance, can inhibit the majority of soluble MMPs, including collagenases (MMP-1 and -13), stromelysins (MMP-3 and -10), gelatinases (MMP-2 and -9), and ADAM-10 but not MT1-MMP and TACE (18–22). TIMP-3, in contrast, is a broad-spectrum inhibitor active against all soluble MMPs as well as MT1-MMP and TACE (23, 24). Collectively, TIMPs modulate the activities of the MMPs and the majority of the ADAMs and ADAM-TSes with only a few exceptions, such as ADAM-TS13 and -15 (25).

Through systematic and extensive site-directed mutations of the MMP-binding ridge, we have previously created a panel of designer TIMP-1s that display superb inhibitory activities against MT1-MMP and TACE (24, 26–28). In this report, we describe how a soluble designer TIMP-1, named T1^TACE, that has dual potencies for MT1-MMP and TACE can be targeted to the cell membrane for renal carcinoma (CaKi-1) inhibition by fusion with the prion protein (named T1^{Pr αTACE} here). Renal carcinomas are highly metastatic and refractive cancers for which there is no effective chemotherapy or radiation treatment (29, 30). Immunohistochemistry and mRNA analysis on clinical samples indicate that MT1-MMP and TACE are often overexpressed in renal carcinoma tissues, and inactivation of the proteinases may provide an effective means of blocking interleukin-1β- and Notch-mediated cancer cell invasion (31–34). Here, we show not only that the membrane-anchored designer TIMP T1^{Pr αTACE} could abrogate the activities of cellular MT1-MMP and TACE but also that it is highly potent in the prevention of cell migration, tubulogenesis, and tumorigenesis under both in vitro and in vivo settings. We are optimistic that the approach reported here can be exploited as a potential intervention strategy for renal carcinoma therapy.

RESULTS

T1^{Pr αTACE}: a membrane-anchored, broad-spectrum designer TIMP-1 created specifically for dual MT1-MMP/TACE inhibition.

Listed in Fig. 1A are the amino acid sequences of the wild-type TIMP-1 (T1^WT), a glycosyl-phosphatidylinositol (GPI)-anchored TIMP-1 (T1^Pr), and a GPI-anchored designer TIMP-1, T1^{Pr αTACE}, developed specifically for dual MT1-MMP/TACE inhibition. The predecessor of T1^{Pr αTACE}, a soluble TIMP-1 mutant named “T1^TACE” created by our group, was a TIMP-1 derivative that had undergone extensive mutagenesis at the MMP-binding ridge (27). Originally designed with TACE as the intended target, T1^TACE contained a complex V4S/TIMP-3-AB-loop/V69L/T98L quadruple mutation tailored specifically to fit the catalytic pocket of TACE (K_i^app of 0.14 nM) (Fig. 1A and B). In this project, we subjected the TIMP to further kinetic analysis and showed that it too displayed significant anti-MT1-MMP activity (K_i^app of 7.70 nM) (Fig. 1B, lower).

FIG 1.

Translocation of a designer TIMP-1 to the cell membrane for dual MT1-MMP/TACE inhibition. (A) Amino acid sequences of the wild-type and designer TIMP-1s tailored for MT1-MMP and TACE inhibition. First sequence, wild-type TIMP-1; second sequence, GPI-anchored TIMP-1; third sequence, GPI-anchored designer TIMP-1 tailored for MT1-MMP and TACE inhibition; fourth sequence, GPI signal peptide of the prion protein. Residues not native to TIMP-1 are boxed. (B, top) Inhibitory constants, K_i^app, of T1^TACE against MT1-MMP and TACE. While T1^WT was a poor inhibitor for MT1-MMP and TACE (K_i^app of >170 nM), its mutant, T1^TACE, displayed superb affinities for the proteinases. Superscripts a and b indicate results that have been previously published. (Lower) Inhibition of MT1-MMP by T1^TACE, the TIMP-1 mutant originally developed for TACE inhibition. (C) Sequestration of T1^Pr and T1^{Pr αTACE} to the cell membrane of CaKi-1 as revealed by reverse zymography. In contrast to T1^WT, T1^Pr and T1^{Pr αTACE} were sequestered exclusively to the cell membrane. Note that T1^{Pr αTACE} had a slightly lower molecular mass (ca. 26 kDa) due to the abolition of a glycosylation site (EVN*QT) as a result of site-directed mutagenesis. The positions of the GPI-anchored TIMPs are highlighted by asterisks. Note that the upper image is a merged image of the same blot after the omission of another designer TIMP (not included in this paper) originally positioned between T1^Pr and T1^{Pr αTACE}. (D) Nonpermeabilized immunostaining with a TIMP-1 antibody confirmed successful translocation of T1^Pr to the HT1080 cell surface, as viewed under a fluorescence microscope. (E) Similar patterns of T1^Pr localization in HT1080 fibrosarcoma, CaKi-1 renal carcinoma, and A549 lung adenocarcinoma cells (images captured with a confocal microscope). (F) Cross-sectional views of a T1^Pr-transduced CaKi-1 cell highlighting the immobilization of the GPI-TIMP on the cell surface (left) and membrane ruffles (right).

To target T1^TACE to the cellular membranes, we chose the GPI anchor of the prion protein because of the similarities in the way PrP^Sc, the scrapie form of the protein, and MT1-MMP are localized on the cell surface (35). Following stable transduction in HT1080 fibrosarcoma, CaKi-1 renal carcinoma, and A549 lung adenocarcinoma cells, the sequestration profiles of all the TIMPs were analyzed by reverse zymography. As shown, while much of the T1^WT (ca. 25 kDa) was secreted to the conditioned medium, the membrane-anchored TIMP-1s T1^Pr and T1^{Pr αTACE} were found to be sequestered exclusively to the membrane fraction (Fig. 1C, highlighted by an asterisk; see also Fig. S1 in the supplemental material). There was no detectable level of these TIMPs in the medium. The molecular mass of T1^{Pr αTACE} (ca. 26 kDa) was noticeably lower than that of T1^Pr (ca. 28 kDa), owing to the elimination of one of its N-linked glycosylation sites (residue N³⁰) during the process of site-directed mutagenesis (T1^WT sequence EVN³⁰QT, T1^TACE mutant sequence EVN³⁰QG [residues 30 and 32 are boldface] [boxed in Fig. 1A]) (16).

Subsequent immunostaining on nonpermeabilized T1^Pr-transduced HT1080 cells revealed the abundance of the membrane-anchored TIMPs on the cell surface (Fig. 1D). Instead of being confined to specific locations, T1^Pr appeared to cover almost the entire cell surface and cell edges where membrane ruffling is at its thickest (Fig. 1D and E, highlighted by arrowheads). In addition to that of HT1080, a similar pattern of localization was observed in renal carcinoma (CaKi-1) and lung adenocarcinoma (A549) cells (Fig. 1E). Enclosed in Fig. 1F are three cross-sectional views of a typical CaKi-1 transductant that demonstrate how T1^Pr is immobilized on the cell surface as well as membrane ruffles.

Immunostaining confirms colocalization of T1^{Pr αTACE} with MT1-MMP and TACE in CaKi-1 and HT1080 cells.

Immunostaining under nonpermeabilizing conditions showed that, like its T1^Pr predecessor, T1^{Pr αTACE} was also localized predominantly on the cell surface. As shown in Fig. 2A, the designer TIMP colocalized with MT1-MMP throughout the cell perimeter as well as much of the cell surface in both CaKi-1 and HT1080 cells (Fig. 2A and Fig. S2). Presumably because of its inherent affinity for MT1-MMP, the degree of colocalization observed in T1^{Pr αTACE} was far more intense than that in T1^Pr, which showed only a limited colocalization at the cell edges and cell-cell junction.

FIG 2.

Nonpermeabilized immunostaining reveals colocalization of T1^{Pr αTACE} with MT1-MMP and TACE in CaKi-1 cells. Stably transduced CaKi-1 cells were coimmunostained with antibodies against TIMP-1 and MT1-MMP or TACE under nonpermeabilizing conditions. (A) T1^{Pr αTACE} was detected in abundance throughout the cell surface, particularly the cell edges, and in excellent colocalization with MT1-MMP. While the signal of T1^Pr was relatively weak and limited to the membrane ruffles and cell-cell junctions, T1^{Pr αTACE} colocalized with MT1-MMP extensively throughout the cell edges. (B) Colocalization of T1^{Pr αTACE} and TACE at the membrane ruffles as well as the cell-cell junction. (C) Colocalization of T1^{Pr αTACE} with MT1-MMP (upper) and TACE (lower) at the membrane ruffles as well as the cell-cell junction. Arrowheads A to E denote the same cell sites as those viewed from orthogonal (left) and nonorthogonal (right) perspectives.

A similar colocalization pattern was noticed in TACE. Besides the cell edges, colocalization was most evident along the cell-cell junction. As with the previous case, T1^{Pr αTACE} demonstrated a higher degree of colocalization than T1^Pr (Fig. 2B and Fig. S2). Figure 2C are two independent panels of orthogonal views that illustrate how colocalization occurred between T1^{Pr αTACE} and MT1-MMP (top) or TACE (lower) along the cell edges as well as much of the membrane ruffles in stably transduced CaKi-1 cells.

Under permeabilizing conditions, a significant proportion of T1^{Pr αTACE} was found to colocalize with MT1-MMP at the perinuclear regions (Fig. S3 and S4). Overall, the profile of TACE staining was reminiscent of that of MT1-MMP. Of particular note (Fig. S3 and S4) is the superb colocalization of TACE with T1^{Pr αTACE} at the membrane ruffles, a finding that corroborates the observation from the nonpermeabilized staining condition.

T1^{Pr αTACE} disrupts MT1-MMP-dependent HT1080 tubulogenicity in 3D Matrigel.

Like the majority of the metastatic fibrosarcomas and cutaneous melanomas, HT1080 cells form vascular networks with lumen-like features in Matrigel suspension as part of the neovascularization process required for survival (36). As previously shown, this program of branching morphogenesis relies on proteolytic rearrangement of the ECM and is largely MT1-MMP dependent (37). Figure 3A shows that while the empty vector, T1^WT, and T1^Pr transductants spontaneously consolidated into a robust network of vasculatures within 24 h of seeding in Matrigel, T1^{Pr αTACE}-transduced cells proliferated only as loose aggregates (highlighted by arrowheads; enlarged views below) with no sign of vasculature formation (*, P < 0.01 versus other TIMPs; quantified by total cell junctions).

FIG 3.

T1^{Pr αTACE} disrupts tubulogenesis and inhibits native MT1-MMP/TACE in HT1080 cells. (A) In contrast to the empty vector, T1^WT, and T1^Pr transductants that form a network of cord-like vasculatures in the Matrigel, T1^{Pr αTACE}-transduced HT1080 cells proliferated only as loose aggregates. The total number of junctions per field was compared between groups. Data represent the means ± standard errors of the means (SEM) from four independent samples. (B) Near-complete abrogation of the gelatinolytic activity of HT1080 cells by T1^{Pr αTACE}. The adjacent panels show the same cells stained with an anti-MT1-MMP antibody. To calculate gelatin degradation, areas without green fluorescence were measured (n = 4), and the mean values were compared with that of the empty vector control (set at 100%). (C) In both HB-EGF and TNF-α shedding assays, T1^{Pr αTACE}-transduced HT1080 cells demonstrated the lowest shedding activities upon induction with PMA (*, P < 0.05 versus T1^Pr in both instances). (D) Similar patterns of MMP-2 and -9 processing in HT1080 cells stably expressing the designer TIMP-1s as revealed by zymography. (E) Relative amount of TACE, MT1-MMP, and BRCA1 in the whole-cell lysates of stably transduced HT1080 cells as shown by Western blotting. In all, the designer TIMP-1s had no discernible effect on the expression levels of all the proteins tested. Note that the figures are merged images of the same blot after the omission of another designer TIMP (not included in this paper) originally positioned between T1^Pr and T1^{Pr αTACE}.

The relative potency of the designer TIMP-1s against native MT1-MMP is best illustrated by the images of degraded gelatin in Fig. 3B. In marked contrast to the large, irregular trails of imprints seen in the empty vector control chamber, there was a near-complete lack of gelatin degradation in the chamber occupied by T1^{Pr αTACE} transductants (P < 0.01 versus other TIMPs, as measured by total areas of degraded gelatin). Judging from the generally smudgy and often fainter imprints, T1^Pr appeared to be slightly more effective than T1^WT. As shown, the chamber once occupied by T1^WT-expressing HT1080 was littered with numerous dark splotches that directly testified to the lack of inhibitory efficacy by the wild-type TIMP on MT1-MMP.

T1^{Pr αTACE} suppresses HB-EGF and TNF-α shedding in HT1080 cells.

Two independent heparin binding epithelial growth factor (HB-EGF) and TNF-α shedding assays were performed to assess the potency of the designer TIMP-1s against cellular TACE. In both assays, T1^{Pr αTACE} transductants demonstrated the lowest shedding ability followed closely by T1^Pr (P < 0.05) (Fig. 3C). Indeed, the levels of shed HB-EGF/TNF-α in T1^{Pr αTACE} samples amounted to only 35% of those of the empty vector controls.

No alteration in MMP-2, MMP-9, TACE, MT1-MMP, and BRCA1 processing as a result of designer TIMP-1s’ expression.

To find out if the GPI-TIMP-1s had a direct impact on MMP processing, especially MMP-2 and -9, a gelatin zymography was set up to analyze the conditioned medium of HT1080 cells stably transduced with the TIMPs. Apart from a slight reduction in the 90-kDa pro-MMP-9 species in T1^Pr and T1^{Pr αTACE}, there was no indication to suggest that the processing pattern of the MMPs had been dramatically altered (Fig. 3D). In all, three gelatinolytic bands corresponding to the proform (90 kDa), intermediate form (85 kDa), and mature form (80 kDa) of MMP-9 and two further closely associated 70-kDa and 68-kDa MMP-2 bands could be discerned in similar intensities across the lanes.

Likewise, immunoblotting on the total cell lysates indicated no dramatic change in the levels of TACE, MT1-MMP, and BRCA1 among the transductants (Fig. 3E).

T1^{Pr αTACE} induces CaKi-1 cell hypertrophy and ECM accumulation.

An interesting observation that we noticed during the course of this study was the morphological transformation that CaKi-1 cells underwent as a result of T1^{Pr αTACE} expression. Instead of the typical phenotypic characteristics of an epithelial cell, T1^{Pr αTACE} transductants appeared stretched without a defined shape or boundary. Overall, the cells are thinner but larger in size (Fig. 4A). Suspecting a change in the cytoskeletal meshwork, we stained the cells under nonpermeabilizing conditions and found an overwhelming abundance of cell surface actin filaments that markedly resembled stress fibers (Fig. 4B). Further staining on the ECM also revealed a strikingly large quantity of collagen I and fibronectin at the pericellular matrices of the cells transduced with T1^{Pr αTACE} but not T1^Pr and T1^WT (Fig. 5A and B). In contrast, the level of integrin αv detected in the same sample was noticeably lower under similar staining conditions (Fig. 5C).

FIG 4.

T1^{Pr αTACE} induces cell hypertrophy and stress fibers in CaKi-1 cells. (A) Instead of the typical phenotypic appearance of an epithelial cell, CaKi-1 cells stably expressing T1^{Pr αTACE} appeared larger, without a defined shape or boundary. (B) Immunostaining under nonpermeabilized condition reveals a large amount of cell surface actin filaments that markedly resembled stress fibers in T1^{Pr αTACE} transductants. Nuclear staining was performed with 4’,6-diamidino-2-phenylindole (DAPI).

FIG 5.

Accumulation of collagen I and fibronectin in the ECM of T1^{Pr αTACE}-expressing CaKi-1 cells. (A and B) Immunostaining on the ECM shows an accumulation of collagen I (A) and fibronectin (B) in the ECM of CaKi-1 cells stably transduced with T1^{Pr αTACE} but not T1^WT or T1^Pr. The adjacent panels show the same cells stained with DAPI. (C) In contrast, the level of integrin αv was found to be downregulated (images captured with a Nikon Eclipse Ni fluorescence microscope).

T1^{Pr αTACE} suppresses CaKi-1 cell motility and proliferation in 2D and 3D culture.

In a parallel study, the impact of the TIMPs on CaKi-1 cell motility was investigated by a transwell migration assay. As shown in Fig. 6A, irrespective of wild-type or designer origin, all the TIMPs had an adverse effect on cell migration. Among the designer TIMPs, T1^{Pr αTACE} was again the most effective (P < 0.05 versus its closest runner-up, T1^WT). Besides motility inhibition, T1^{Pr αTACE} also negatively impacted cell proliferation under two-dimensional (2D) culture conditions. Indeed, the proliferation rate of T1^{Pr αTACE} transductants was no higher than 24% of that of the empty vector-transduced cells (P < 0.01) (Fig. 6B). In the confines of three-dimensional (3D) Matrigel, the antiproliferative effect of T1^{Pr αTACE} was even more evident. Figure 6C is a collage of snapshots that provide an overview on the effect of the TIMPs on CaKi-1 clonogenic development after 25 days of incubation in Matrigel suspension. While a significant proportion of the empty vector- and T1^WT-transduced cells had grown into clumpy, irregularly shaped spheroids in excess of 100 μm in diameter, colonies in the T1^{Pr αTACE} wells were much smaller. As a matter of fact, there were hardly any spheroids that exceeded 50 μm within the T1^{Pr αTACE} population by the end of the study (P < 0.001 versus empty vector control).

FIG 6.

T1^{Pr αTACE} suppresses CaKi-1 motility and has no growth potentiation activity in vivo. Instead, the designer TIMP inhibits CaKi-1 proliferation in a 3D Matrigel model and induces cell senescence. (A) T1^{Pr αTACE}-transduced CaKi-1 cells demonstrated the lowest migration rate across a transwell filter, closely followed by T1^WT and T1^Pr (*, P < 0.05 for T1^{Pr αTACE} versus the closest runner-up, T1^WT). (B) CaKi-1 expressing T1^{Pr αTACE} proliferated at a substantially lower rate than the empty vector control cells under 2D culture conditions (*, P < 0.01 versus empty vector). Results in the bar chart represent the averages from three technical repeats ± SD. (C) T1^{Pr αTACE} suppressed CaKi-1 cell proliferation in 3D Matrigel model. While many of the empty vector control cells developed into colonies in excess of 100 μm within 25 days of incubation, T1^{Pr αTACE} transductants grew at a significantly lower pace (*, P < 0.001 for number of colonies of >100 μm versus empty vector control). (D) High incidence of aged cells in T1^{Pr αTACE}-transduced CaKi-1 population as revealed by X-gal staining (*, P < 0.001 versus empty vector control). (E) Membrane-anchored TIMP-1s have no growth-potentiating activity in vivo. Under the harsh conditions stipulated in this study, no tumor growth was detected in the control, T1^Pr, and T1^{Pr αTACE} mice by the end of the 35-day incubation period. In contrast, six of the inocula in the T1^WT group developed into solid tumors ranging between 25 and 240 mm³ in volume (*, P < 0.05 for tumor development incidence versus other groups).

T1^{Pr αTACE} induces cellular senescence in CaKi-1 renal carcinoma cells.

Believed to be a tumor-suppressive mechanism, senescence represents an arrested state in which the cells remain viable and metabolically active but incapable of dividing despite stimulation by passaging. Typical manifestations of cellular senescence include cell hypertrophy and an increase in senescence-associated β-galactosidase (SA-β-Gal) activity (38). Apart from an alteration in morphology (elaborated in Fig. 4), another distinctive feature of CaKi-1 cells transduced with T1^{Pr αTACE} was their abnormally high level of SA-β-Gal activities compared to that of the empty vector-, T1^WT-, and T1^Pr-transduced cells (Fig. 6D). While the formation of local blue precipitates in the control groups was sparse and occasional, staining in T1^{Pr αTACE} transductants appeared far more intense and clustered. With an average percentage of senescing cells exceeding 19%, the incidence of aged cells in the T1^{Pr αTACE} population was significantly higher than those of the empty vector, T1^WT, and T1^Pr cells (between 2 and 4%; P < 0.001).

Unlike T1^WT, T1^{Pr αTACE} lacks tumor potentiation activity.

A major issue concerning the use of TIMP-1 in cancer therapy is its growth potentiation activity. Apart from its preeminent role as a proteinase inhibitor, TIMP-1 also has the undesirable side effect of promoting cell proliferation (39). To assess if this stimulatory effect on cell growth is retained in the GPI-anchored form of TIMP-1s, CaKi-1 cells transduced with the designer TIMPs were inoculated in nonobese diabetic/severe combined immunodeficient (NOD/SCID) mice and observed for a period 35 days. To harshen growth conditions, Matrigel was removed from the suspension medium prior to inoculation and the cell number was intentionally kept low (2 × 10⁵ per injection site) to render the potentiating effects of the TIMPs, if there are any, in greater contrast.

Figure 6E is a bar chart detailing the individual tumor masses at the conclusion of the experiment. As shown, not a single tumor was detected in the designer TIMP groups, be it T1^Pr or T1^{Pr αTACE}. Under the conditions adopted in this study, even the empty vector control cells were unable to develop into tumors. As expected, six of the eight inocula in the T1^WT group grew into solid tumors ranging between 25 and 245 mm³ in volume (P < 0.05 for tumor development incidence versus that of other groups). The results confirmed that, unlike wild-type TIMP-1, none of the membrane-anchored TIMP-1s had growth potentiation activity.

Total inhibition of CaKi-1 xenograft in NOD/SCID mice by T1^{Pr αTACE}.

So far, all the in vitro results strongly supported the efficacy of T1^{Pr αTACE}, as opposed to T1^WT and T1^Pr, as the most effective tumor suppressor. To assess the effectiveness of the designer TIMP under in vivo settings, T1^{Pr αTACE}-transduced CaKi-1 cells were grafted onto NOD/SCID mice for further observation (n = 10). Figure 7 is a summary of the findings on day 57, when the experiment reached a humane endpoint.

FIG 7.

T1^{Pr αTACE} completely abrogates CaKi-1 proliferation in NOD/SCID mice. (A, left) CaKi-1 tumor growth curves measured at different time points over a 57-day period. Without interference from the TIMP, control (empty vector-transduced) CaKi-1 implants rapidly developed into solid tumors upon inoculation into NOD/SCID mice. (Right) Circles highlighting the contrast between the control and T1^{Pr αTACE} tumors before the experiment was terminated (n = 10). (B) Surgically removed tumors at the end of the study. Note that not a single T1^{Pr αTACE} inoculum developed into a tumor by the time the experiment was concluded. Circles represent T1^{Pr αTACE} inocula that failed to develop into tumors. (C) Scatter chart showing individual tumor masses for the control tumors (average, 2,531 mg; P < 0.001 versus T1^{Pr αTACE}). The mean for each group is indicated by a bar.

Enclosed in Fig. 7A are the growth curves charting the development of the control (i.e., empty vector transduced) and T1^{Pr αTACE} tumors for the entire course of this study. Without interference from the TIMPs, CaKi-1 tumors started to appear in all the control mice within 10 days of inoculation. The average tumor volume in the group at the end of the 2 weeks was just over 165 mm³. Rapid expansion was recorded in all the control mice from day 14 to 57, when the tumors reached a very substantial average of 3,910 mm³. In complete contrast, not a single tumor was observed in all the T1^{Pr αTACE} inocula throughout the period of this experiment (Fig. 7A, right, and B).

Figure 7C is a scattered chart detailing the individual tumor masses following postmortem dissection. While all 10 inocula in the control group developed into solid tumors between 1,298 and 4,066 mg in mass (average, 2,531 mg), not a single T1^{Pr αTACE} inoculum was able to proliferate into solid tumors by the time the experiment was terminated (P < 0.001 for tumor development incidence versus that of control mice).

T1^{Pr αTACE} prevents CaKi-1 cell proliferation and migration in vivo, as demonstrated by bioluminescence imaging.

To allow the impact of T1^{Pr αTACE} to be visualized in greater detail in vivo, we further set up a bioluminescence study in BALB/c nude mice (n = 8). Figure 8 represents images taken 2 weeks after cell inoculation. In contrast to the control mice, in which extensive cancer cell resettlement was found to have occurred in the colons (7/8) and testicular regions (6/8), not a single subject in the T1^{Pr αTACE} group exhibited signs of wide-spread migration. As a matter of fact, the overall signal intensity detected in the T1^{Pr αTACE} group was far weaker than those of the control mice. Of the eight mice in the T1^{Pr αTACE} group, only three exhibited weak signs of luciferase activity, mostly at the testicular regions (mice 1, 3, and 7).

FIG 8.

Inability of T1^{Pr αTACE}-transduced CaKi-1 to proliferate and metastasize in vivo, as shown by bioluminescence imaging. CaKi-1 cells stably transduced with T1^{Pr αTACE} were labeled with luciferase and administered via intrahepatic injection into BALB/c nude mice to allow for cell growth and migration for 14 days prior to imaging (n = 8). While CaKi-1 cells in the control group were found to have metastasized extensively to the colon (7/8) and testicular regions (6/8), not a single mouse in the T1^{Pr αTACE} group showed signs of widespread migration. Of the eight mice in the T1^{Pr αTACE} group, only three exhibited weak luciferase activity in the testicles (mice 1, 3, and 7). Note that the figure is composed of merged images from three scanning sessions performed in the same afternoon, as the Kodak FX Pro in vivo imaging system imaging station can accommodate a maximum of six mice at any one time.

DISCUSSION

We have previously shown that the GPI anchor of prions can be used for MT1-MMP targeting due to the similarities in which the scrapie form of the prion protein PrP^Sc and MT1-MMP are localized at the membrane ruffles (40). Here, we demonstrate that the coverage of the anchor can be further extended to TACE by the use of a broad-spectrum TIMP variant such as T1^{Pr αTACE} with even more impressive outcomes. MT1-MMP and TACE are both attractive therapeutic targets due to the critical roles they play in cancer cell invasion and dissemination (6, 14, 41). Among the four endogenous inhibitors of the MMPs, only TIMP-3 exhibits dual activities against MT1-MMP and TACE (23, 24). Our attempt to translocate TIMP-3 to the cell membrane has been largely unsuccessful due to the inability of the protein to fold correctly upon fusion with prion (unpublished results). T1^{Pr αTACE}, the leading designer TIMP-1 in this report, is a membrane-tethered “TIMP-3 mimic” developed specifically to meet the requirement for dual MT1-MMP/TACE inhibition.

Collective findings from all the biochemical and cell-based experiments performed in this study are unanimous in supporting the deleterious effect that T1^{Pr αTACE} has on MT1-MMP and TACE. As far as the consequences of MT1-MMP/TACE activity abrogation are concerned, we have provided clear evidence of fibronectin and collagen I deposition around T1^{Pr αTACE}-expressing cells. Maintenance of normal ECM homeostasis is integral to numerous processes, including cell proliferation, survival, migration, and differentiation. As documented by Hotary et al., tumor cells incapable of exercising control over their surroundings would likely be trapped in a compact configuration unable to undergo cytoskeletal reorganization and a change in cell geometry required for proliferation (42). This proteolytic proficiency, as demonstrated here, is severely curtailed in T1^{Pr αTACE} transductants. In our view, this deprivation of metalloproteinase-conferred 3D-specific growth advantage is the principal reason for the failure of the CaKi-1 inoculum to thrive in the newly introduced physiological environment.

In addition to pericellular stroma disruption, the high incidence of aged cells in the T1^{Pr αTACE} population could be another factor that compounded the TIMP’s efficacy. T1^{Pr αTACE} triggered ageing in CaKi-1 cells, causing them to manifest all three hallmarks of senescence, namely, cell hypertrophy, a slowdown in proliferation, and an upsurge in SA-β-Gal activity. The molecular basis behind the phenomenon, in all probability, could again be attributed to MT1-MMP activity disruption. MT1-MMP is crucial in preventing cellular senescence by maintaining an uninterrupted signal transmission between the nucleus and extracellular compartment through a framework of cytoskeletal/nucleoskeletal proteins in a process termed mechanotransduction (43). Studies carried out with MMP-resistant collagen and MT1-MMP knockout mice have independently shown that a downregulation in MT1-MMP function leads to an acceleration in senescence response (44, 45). The high volumes of senescing cells and stress fibers as evidenced in the T1^{Pr αTACE} population, we surmise, were the direct outcomes of a failed mechanotransduction brought about by MT1-MMP inactivation (46).

The impact of T1^{Pr αTACE} on TACE, on the other hand, is best summarized by our findings on integrin αv. The adhesion molecule is of special interest to cancer therapists because it can combine with multiple integrin β chains to form integrin variants, such as αvβ₆, which is of great significance in tumor malignancy (47). Considered a good prognostic indicator, integrin αvβ₆ is upregulated in various cancer types in inverse correlation with the patients’ survival time (47). As of now, antagonists to integrin αv in the form of monoclonal antibodies have been successfully employed to halt melanoma and prostate cancer progression in vivo (48–50). Cilengitide, a cyclic RGD pentapeptide that selectively binds to and inhibits αv integrins, has even made it to phase III clinical trials (51). Here, we show that integrin αv is downregulated in T1^{Pr αTACE} transductants. The finding is not entirely surprising, as it has been known that TNF-α promotes cancer migration by stimulating integrin αvβ₆ production in a direct and coordinate manner (52). Deprived of TNF-α shedding capability, T1^{Pr αTACE} transductants had indeed been expected to be less able to make integrin αv for migratory purposes. An in-depth profiling is under way to ascertain if other integrin α/β subunits have been similarly affected.

We have thus far demonstrated the effectiveness of the current chimeric approach in TIMP engineering. Of the numerous benefits this strategy could potentially provide, we find the broad-spectrum potency of T1^{Pr αTACE} most impressive, as it is a capability that only TIMP-3 can match (24, 53). Besides the dual advantages of being in close proximity to MT1-MMP/TACE and having a wide spectrum of inhibitory activities, T1^{Pr αTACE} has another competitive edge over TIMP-3 that could make it a more favorable therapeutic agent. Unlike TIMP-3, T1^{Pr αTACE} has no propensity to adhere to the ECM. Wastage through diversion to the interstitial compartment therefore can be reduced to a minimum. Moreover, abolishment of the undesirable growth-stimulating effect of wild-type TIMP-1 by the GPI anchor has further enhanced its appeal as a possible therapeutic agent.

In conclusion, we have provided here clear and concrete evidence to support the use of the prion protein as a targeting device in TIMP delivery. Not only have our findings demonstrated the practical feasibility of the fusion approach but the results are also a proof of principle that could pave the way for a more advanced conceptual exploration. We are currently assessing the effectiveness of adenoviral and recombinant protein forms of T1^{Pr αTACE} in animal subjects. The results will be made public upon complete satisfaction of all the patenting criteria. Our unique approach of TIMP anchorage, we believe, may provide an exciting new perspective for the development of experimental therapeutics aiming at renal carcinoma eradication.

MATERIALS AND METHODS

Materials.

Unless otherwise stated, all the reagents used in this study were purchased from ThermoScientific USA. Antibodies against TIMP-1 (Ab1827; Abcam), MT1-MMP (Ab38970; Abcam), TACE (Ab28233; Abcam), actin (A2066; Sigma), fibronectin (MAB 19181; R&D Systems), integrin αv (MAB12191; R&D Systems), and collagen I (Ab34710; Abcam) were obtained either from R&D Systems (Minneapolis, MN), Sigma-Aldrich (Dorset, UK), or Abcam (Cambridge, MA). The senescence detection kit (number K-320-250), Matrigel (number 354234), and potassium d-luciferin (number 10101-2) were products of Biovision (Milpitas, CA), BD Biosciences (Franklin Lakes, NJ), and Biotium (Fremont, CA), respectively. HT1080, A549, and CaKi-1 cell lines were acquired from the Shanghai Cell Bank, Chinese Academy of Science, Shanghai, China, where authentication was performed by short-tandem repeat profiling, and subjected to regular mycoplasma screening in our laboratory to ensure no contamination in the cell stocks.

Transduction, zymography, reverse zymography, and transwell migration assay.

The procedures for lentiviral transduction, zymography, reverse zymography, and transwell migration assay have all been described in detail elsewhere (54). For the Transwell migration assay, the experiment had been performed three times and the readouts were expressed as the average of triplicate measurements ± standard deviations (SD).

Immunofluorescence microscopy.

Cells seeded in Nunc Lab-Tek II chamber slides were fixed in 4% paraformaldehyde before blocking and permeabilization with 5% bovine serum albumin–0.3% Triton X-100 in phosphate-buffered saline (PBS) for 2 h. Incubation in primary antibodies was usually performed overnight at 4°C. Following a brief rinse with PBS, Alexa Fluor 488 (or 555)-conjugated anti-mouse and/or anti-rabbit secondary antibodies were added to the samples for 2 h prior to visualization with a Zeiss LSM880 Airyscan confocal microscope. The same procedure was employed for nonpermeabilized immunostaining, except that no Triton X-100 was added to the blocking and incubation buffers during the experiment.

Gelatin degradation assay.

HT1080 cells stably transduced with the TIMPs (approximately 1,000 cells) were cultured overnight in Nunc Lab-Tek II chamber slides precoated with 0.5 mg/ml porcine Oregon Green 488-conjugated gelatin. Following fixation in 4% paraformaldehyde, the slides were probed with an anti-MT1-MMP antibody and examined under a Nikon Eclipse Ni fluorescence microscope for degraded gelatin patches, which should have lost fluorescence and appeared black.

HB-EGF and TNF-α shedding assays.

Heparin binding epithelial growth factor (HB-EGF) and TNF-α enzyme-linked immunosorbent assays (ELISAs) were performed essentially as described previously (55). Two days after transfection, the cells were stimulated with 200 ng/ml of phorbol 12-myristate 13-acetate (PMA), and the amount of alkaline phosphatase–TNF-α released into the medium after 3 h was quantified either by colorimetric assay or ELISA. Data on the bar chart were plotted as the means from four technical repeats ± SD. To ensure the reproducibility of the findings, the assays were performed no fewer than four times.

Cellular senescence assay.

Caki-1 cells stably transduced with the TIMPs were grown in a 12-well plate for 1 week before staining with 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-gal) according to the manufacturer’s instructions. Following overnight incubation at 37°C, the cells were washed with PBS and visualized under a Nikon Eclipse Ti inverted microscope for development of blue precipitates.

Clonogenic assay in Matrigel.

The procedure for the assay is essentially identical to the one described in reference 55, with the exception that Matrigel (3 mg/ml) was used in place of agarose. The assay was performed in duplicate, and the numbers of colonies larger than 100 μm after 25 days of incubation were summed and averaged for analysis.

Tumor development study in NOD/SCID mice.

All experiments involving the use of animals were conducted by GenePharma, Co., Ltd. (animal license registration number SYXK [Su] 2014-0054), Singapore-Suzhou Industrial Park, by strictly following the regulations outlined in the National Guidance for Animal Care, China (56), with respect to veterinary care and experimental procedures. To determine the growth potentiation activity of the TIMPs, 2 × 10⁵ freshly prepared CaKi-1 cells in Dulbecco’s modified Eagle’s medium (DMEM) (without Matrigel) were inoculated subcutaneously into the left or right flanks of 6-week-old female mice (n = 8 per group) for 35 days. To evaluate the efficacy of the TIMPs in CaKi-1 cell suppression, a higher number of cells (4 × 10⁶) in 30% Matrigel–DMEM suspension was administered per site of injection (n = 10 per group). Tumor growth was measured with digital calipers twice a week for 57 days. To calculate the tumor volume, the following formula was employed: volume = length × (width)₂ × π/6, where the width is defined as the smaller of the two perpendicular diameters. To ensure reproducibility, two independent animal tests were performed. On both occasions, no tumor growth was recorded in the subjects injected with CaKi-1 cells transduced with T1^{Pr αTACE}.

Bioluminescence imaging in BALB/c nude mice.

CaKi-1 cells with or without T1^{Pr αTACE} (5 × 10⁶ cells in 50 μl DMEM suspension) stably transduced with a luciferase reporter gene (Genomeditech, Shanghai, China) were introduced into BALB/c nude mice (male, 5 to 6 weeks old) intrahepatically 2 weeks prior to in vivo imaging (n = 8). Ten minutes before images were scheduled to be recorded, d-luciferin (potassium salt, 150 mg/kg of body weight) was injected via intraperitoneal administration. Bioluminescence images were taken with sequential 5-s exposures for a total of 5 min with a Kodak FX Pro in vivo imaging system imaging station at the nearby School of Biological Sciences, Suzhou University.

Statistical analysis.

Statistical analysis was performed with the online calculator at the www.socscistatistics.com website. Statistical significance between two independent samples was determined by Student's t test, usually under a two-tailed hypothesis unless stated otherwise.

Data availability and materials.

This study does not require data or materials linked to external sources. All clones are available upon request.