Tissue inhibitor of metalloproteinase-2 (TIMP-2) expression during cardiac neural crest cell migration and its role in proMMP-2 activation

Authors

  • V. Cantemir,

    1. Department of Biomedical Science, Creighton University, Omaha, Nebraska
    Search for more papers by this author
    • Drs. Cantemir and Cai contributed equally to this work.

  • D.H. Cai,

    1. Department of Biomedical Science, Creighton University, Omaha, Nebraska
    Current affiliation:
    1. Bunting/Blaustein Cancer Research Center, Johns Hopkins University, Baltimore, MD 21231
    Search for more papers by this author
    • Drs. Cantemir and Cai contributed equally to this work.

  • M.V. Reedy,

    1. Department of Biology, Creighton University, Omaha, Nebraska
    Search for more papers by this author
  • P.R. Brauer

    Corresponding author
    1. Department of Biomedical Science, Creighton University, Omaha, Nebraska
    • Department of Biomedical Sciences, Creighton University School of Medicine, 2500 California Plaza, Omaha, NE 68178
    Search for more papers by this author

Abstract

Matrix metalloproteinases (MMPs) are important mediators of neural crest (NC) cell migration. Here, we examine the distribution of tissue inhibitor of metalloproteinase (TIMP) -2 and TIMP-3 and test whether manipulating TIMP levels alters chicken cardiac NC cell migration. TIMP-2 mRNA is expressed at stage 11 in the neural epithelium and only in migrating cardiac NC cells. TIMP-3 mRNA is expressed only in the notochord at stage 8 and later in the outflow tract myocardium. Exogenous TIMP-2 increases NC motility in vitro at low concentrations but has no effect when concentrations are increased. In vitro, NC cells express membrane type-1 matrix metalloproteinase (MT1-MMP) and TIMP-2 and they secrete and activate proMMP-2. Antisense TIMP-2 oligonucleotides block proMMP-2 activation, decrease NC cell migration from explants, and perturb NC morphogenesis in ovo. Because TIMP-2 is required for activation of proMMP-2 by MT1-MMP, this finding suggests TIMP-2 expression by cardiac NC cells initiates proMMP-2 activation important for their migration. Developmental Dynamics 231:709–719, 2004. © 2004 Wiley-Liss, Inc.

INTRODUCTION

Neural crest (NC) cells are an ectodermally derived mesenchymal population originating at the junction between the neural plate and the ectoderm. As the neural plate closes to form the neural tube, NC cells emigrate from the epithelium and migrate throughout the embryo. NC cells arising between the mid-otic and caudal end of the third somite axial level (often referred to as “cardiac NC cells”) migrate beneath the ectoderm into pharyngeal arches III, IV, and VI, and a subset of these cells then invade the outflow tract of the heart and participate in its septation (Kirby et al., 1983; Waldo and Kirby, 1998). Studies show that severe cardiac defects, such as persistent truncus arteriosus, double-outlet right ventricle, tetralogy of Fallot, and great vessel defects, occur if the migration of these NC cells is perturbed (Kirby, 1987; Conway et al., 2000).

Matrix metalloproteinases (MMPs) play major roles in controlling the migratory capacity of cells, directing embryonic development and tissue remodeling events, and mediating the epithelial–mesenchymal transition of primary epithelial tumors and their subsequent metastatic capacity (Woessner, 1991; Mignatti and Rifkin, 1993; Vassalli and Pepper, 1994; Witty et al., 1995; Chin and Werb, 1997; Pulyaeva et al., 1997; Werb and Chin, 1998; Miettinen et al., 1999; Prontera et al., 1999; Song et al., 2000; Blavier et al., 2001). For example, specifically blocking the enzymatic activity of MMP-2 or MMP-9 blocks tumor and endothelial cell migration (Deryugina et al., 1997, 2001; Koivunen et al., 1999; Makela et al., 1999; Fang et al., 2000). MMPs are secreted into the extracellular matrix (ECM) as inactive proforms. ProMMPs are proteolytically autoactivated, activated by plasmin or other MMPs, or activated by integral membrane proteins called membrane-type MMPs (MT-MMPs; Vassalli and Pepper, 1994; Nagase, 1997).

Four known endogenous inhibitors, called tissue inhibitors of metalloproteinases (TIMPs), regulate MMP enzymatic activity. TIMPs bind active forms of MMPs and block enzymatic activity of all MMPs but to differing degrees. By modulating MMP activity, TIMPs can mediate cell migration and invasion, ECM degradation, growth factor activation and receptor turnover, and tissue remodeling (for reviews, see Blavier et al., 1999; Brew et al., 2000). Paradoxically, studies show TIMPs may also be required for the activation of some proMMPs. For instance, low levels of TIMP-2 promote the formation of a cell-surface proMMP-2/TIMP-2/MT1-MMP complex that activates proMMP-2 (Strongin et al., 1995; Butler et al., 1998; Jo et al., 2000). However, high levels of TIMP-2 inhibit proMMP-2 activation by sequestering inhibitor-free membrane type-1 MMP (MT1-MMP) needed for proMMP-2 activation. Although other mechanisms can activate proMMP-2, most cells in vitro are unable to activate MMP-2 in the absence of TIMP-2 (Butler et al., 1998; Caterina et al., 2000; Hernandez-Barrantes et al., 2000; Toth et al., 2000; Wang et al., 2000). Moreover, active MMP-2 is not detected in TIMP-2 null mice (Caterina et al., 2000; Wang et al., 2000). Therefore, formation of a proMMP-2/TIMP-2/MT1-MMP complex is the predominant mechanism for proMMP-2 activation.

Like tumor cells, NC cells penetrate basement membranes and invade ECM during their emigration and migration and, hence, may use similar invasive mechanisms. During initial formation and early emigration, chicken cardiac NC cells encounter MMP-2 protein in the ECM through which they migrate (Cai et al., 2000). This MMP-2 is synthesized and deposited by the mesoderm. As the NC cells migrate toward the future pharyngeal arches, MMP-2 protein accumulates on the surfaces of cardiac NC cells and there is a concomitant loss of MMP-2 immunostaining in the ectodermal basement membrane as NC cells migrate over this extracellular substrate (Cai et al., 2000). Blocking MMP activity inhibits cardiac NC cell migration both in vitro and in vivo (Cai and Brauer, 2002). Recently, Duong and Erickson (2004) reported that MMP-2 synthesis and MMP activity are also important mediators of trunk NC cell migration. Given that organisms must limit MMP proteolytic activity, TIMPs are likely important mediators of NC cell migration. However, there is little known regarding the temporal and spatial expression of TIMPs or their functional roles during NC cell formation and early migration.

In this study, we examined the expression pattern of TIMP-2 and TIMP-3 during early NC cell migration and began testing the role of TIMPs in cardiac NC cell migration. We found TIMP-2 was uniquely expressed in early migrating cardiac NC cells, and once cardiac NC cells entered the pharyngeal arches, TIMP-2 expression expanded to include other mesenchymal cells. In contrast, TIMP-3 expression at these same stages was only found in the notochord and heart but was not associated with early cardiac NC cell emigration or migration. In ovo electroporation of TIMP-2 antisense morpholino oligonucleotides into cardiac NC cells perturbed NC cell morphogenesis and induced neural tube defects in embryos. In vitro, cardiac NC cells expressed all three components of the proMMP-2/TIMP-2/MT1-MMP trimeric complex and generated active MMP-2. When low levels of exogenous TIMP-2 were added, cardiac NC cell migration increased but returned to basal levels when the concentration of TIMP-2 was increased further. Cardiac NC cells treated with TIMP-2 antisense morpholino oligonucleotides no longer activated proMMP-2 and exhibited impaired cell migration in vitro. These results are consistent with the hypothesis that cardiac NC cells form a proMMP-2/TIMP-2/MT1-MMP complex on their cell surface that activates proMMP-2 and promotes their migration.

RESULTS

TIMP-2 and TIMP-3 Expression Pattern

To determine whether the expression of TIMP-2 and TIMP-3 correlated with MMP-2 expression and early cardiac NC cell migration, in situ hybridization studies for TIMP-2 and TIMP-3 mRNA were performed on Hamburger–Hamilton stage 8 to 14 chicken embryos. TIMP-1 expression was not investigated, because a chicken homolog has not been identified. Before stage 11, some TIMP-2 mRNA transcripts were detected within the midbrain and hindbrain neural tube (not shown). Beginning at stage 11, the expression of TIMP-2 mRNA greatly increased in rhombomeres 5 and 6 of the neural tube (Fig. 1). Migrating cardiac NC cells also began expressing TIMP-2 mRNA at this stage and continued to express it as they entered the pharyngeal arches (Fig. 1A–K). By this time (stage 14), TIMP-2 mRNA expression expanded to include most of the pharyngeal arch mesenchyme and the endocardium of the heart (not shown). Colocalization with HNK-1 immunostaining showed TIMP-2 was only expressed within a subset of cardiac NC cells, particularly those adjacent to the ectoderm (Fig. 1C–H,J,K). In addition, TIMP-2 mRNA levels appeared higher in pioneering (i.e., leading) than in trailing cardiac NC cells (Fig. 1E–H).

Figure 1.

Tissue inhibitor of metalloproteinase (TIMP) -2 and TIMP-3 expression during early cardiac neural crest (NC) cell migration. A: In situ hybridization showing TIMP-2 mRNA in the neural tube, particularly in rhombomeres 5 and 6, and in migrating cardiac NC cells in a stage 12-chicken embryo. Arrowheads indicate the level of cross-sections of this embryo shown in B–H. B: TIMP-2 mRNA expression at the level of rhombomere 5. Cells in the neural tube express TIMP-2 mRNA (arrowheads). C,D: TIMP-2 mRNA expression at the level of rhombomere 6 (C), with D showing colocalization of HNK-1 immunostaining (to identify NC cells, arrowheads). Neural tube cells also express TIMP-2 at this axial level. E,F: TIMP-2 mRNA expression at the level of rhombomere 7 (E), with F showing colocalization of HNK-1. NC cells at the migratory front (arrowheads) appear to have higher levels of TIMP-2 than trailing NC cells (arrows). G,H: TIMP-2 mRNA expression at level of the anterior second somite with H showing colocalization of HNK-1. Transcripts were still detected in the neural tube but levels appeared to be lower than in rhombomeres 5 and 6. I–K: Stage 13 embryo showing TIMP-2 mRNA expression and HNK-1 colocalization. Arrowheads in I indicate the level of the cross-sections shown in J and K of the same embryo. Arrowheads point to TIMP-2–positive cardiac NC cells. Note, not all cardiac NC cells express TIMP-2 (asterisks). L: TIMP-3 mRNA expression in a whole-mount stage 10- embryo. TIMP-3 mRNA was only detected in the notochord (arrowhead). M: TIMP-3 mRNA expression in a cross-section of a stage 11 embryo. TIMP-3 mRNA transcripts were only detected in the notochord (arrowhead) and outflow tract myocardium (not visible at this axial level). nt, neural tube; o, otic placode. Scale bar in M = 255 μm in A, 50 μm in B–H,J,K, 170 μm in I, 125 μm in L, 60 μm in M.

TIMP-3 mRNA transcripts were first detected in the notochord at stage 8 and persisted there through stage 14 (Fig. 1L,M). Beginning at stage 10+/11, TIMP-3 mRNA was also detected in myocardium of the outflow tract as previously described (Brauer and Cai, 2002). By stage 13, TIMP-3 mRNA expression expanded to include the endocardium as well as the endothelium of developing blood vessels. By stage 14, TIMP-3 mRNA transcripts were also detected in the pharyngeal arch mesenchyme. TIMP-3 was not detected in the neural tube, cardiac NC cells, or paraxial mesoderm of embryos before stage 14.

Functional Studies of TIMP-2 on Cardiac NC Cells

To determine whether TIMP-2 expression by cardiac NC cells had a role in NC cell morphogenesis and migration, we electroporated TIMP-2 antisense or control fluorescein isothiocyanate (FITC) -labeled morpholino oligonucleotides into the closing neural folds at stage 9, just cranial to and along the first four somites. Eighteen hours postelectroporation, TIMP-2 antisense morpholino-treated embryos had a significantly lower survival rate (14 of 24 [58%]; two-tailed Fisher's test, P < 0.05), based on the presence of a detectable heart beat, than control morpholino-treated or sham-electroporated embryos (17 of 19 [89%] and 13 of 14 [93%], respectively). TIMP-2 antisense-treated embryos developed several abnormalities, including neural tube defects, otic vesicle defects, and thickening of the ectoderm overlying the neural tube, which were not present in the control morpholino-treated or sham-electroporated embryos (Fig. 2). Moreover, cardiac NC cell migration was significantly impaired (Fig. 2), and we occasionally noted HNK-1–positive cells lying within the lumen of the neural tube. These results show that TIMP-2 plays an important role in neurulation and cardiac NC morphogenesis in vivo.

Figure 2.

Effect of tissue inhibitor of metalloproteinase (TIMP) -2 antisense and control morpholino oligonucleotides on early cardiac neural crest (NC) cell migration and chick development in ovo. Fluorescein isothiocyanate (FITC) -labeled control or TIMP-2 morpholino oligonucleotides were injected into the neural folds of stage 9 chicken embryos anterior to somite 4. Embryos were subjected to electroporation and then reincubated for 18 hr, fixed, and immunostained with HNK-1 to identify NC cells. A: Effect of FITC-labeled control morpholinos. Control embryos usually reached stage 13 and appeared normal (shown here in darkfield illumination). B: Effect of FITC-labeled TIMP-2 antisense morpholinos. These embryos had a higher mortality rate, were smaller in size, exhibited neural tube and cranial defects, and had abnormal otic placodes. C,E,G: Cross-section of an embryo treated with control morpholino showing migrating morpholino/HNK-1–positive NC cells (arrowheads in C and E). G: An image generated by overlaying C and E. D,F,H: Cross-section of an embryo treated with TIMP-2 antisense morpholino showing morpholino/HNK-1–positive NC cells (arrowheads in D and F). Embryos exhibited abnormal neural tubes and delayed cranial development relative to the trunk. Cardiac NC cell migration (arrowheads) was impaired in these embryos relative to control embryos. H: An image generated by overlaying D and F. e, ectoderm; n, notochord; nt, neural tube; s, somite. Scale bar in H = 660 μm in A, 500 μm in B, 50 μm in C–H.

To better elucidate the mechanisms by which TIMP-2 might alter cardiac NC cell migration, we chose to examine the functional role of TIMP-2 using a cardiac NC cell culture system. In vivo, early migrating cardiac NC cells do not synthesize MMP-2 but do so after reaching the pharyngeal arches (stages 14 and older; Cai et al., 2000). However, we did not know whether cardiac NC cells synthesize and secrete MMP-2 under the conditions used in this study to evaluate cardiac NC cell migratory behavior in vitro. Therefore, we immunostained cardiac NC cultures with an MMP-2 antibody and subjected the conditioned medium from these cells to gelatin zymography. We found cultured cardiac NC cells were immunopositive for MMP-2 (Fig. 3A) and that cardiac NC cell conditioned medium contained both proMMP-2 and active MMP-2 (Fig. 3C). TIMP-2 mRNA and protein was also expressed by cardiac NC cells under these conditions (Fig. 3B,D). Low levels of TIMP-2 promote the formation of a cell-surface proMMP-2/TIMP-2/MT1-MMP complex necessary for the activation of proMMP-2. Therefore, we determined whether cultured cardiac NC cells expressed MT1-MMP. MT1-MMP mRNA was detected by reverse transcriptase-polymerase chain reaction (RT-PCR) using polyA mRNA obtained from cultured cardiac NC cells (Fig. 3D). This strategy shows that cardiac NC cells express both MT1-MMP and TIMP-2 and synthesize and activate proMMP-2 in vitro.

Figure 3.

Matrix metalloproteinase (MMP) -2, tissue inhibitor of metalloproteinase (TIMP) -2, and membrane type-1 matrix metalloproteinase (MT1-MMP) in cardiac neural crest (NC) cells cultured under the same conditions used in the migration assays. A: Cultured cardiac NC cells were immunopositive for MMP-2. B: Cultured cardiac NC cells were immunopositive for TIMP-2. C: Zymography of NC cell-conditioned medium showing the presence of proMMP-2 (72 kDa), the active intermediate (68 kDa), and fully active MMP-2 (62–63 kDa) in cardiac NC cell conditioned medium (NCCM). This strategy showed NC cells secreted proMMP-2 and activated it in vitro. D: Ethidium bromide–stained agarose gel showing reverse transcriptase-polymerase chain reaction (RT-PCR) products generated from cultures of cardiac NC cells using primers for chicken TIMP-2 and chicken MT1-MMP. PolyA mRNA from cardiac NC cell cultures was subjected to RT-PCR using either chicken TIMP-2 or MT1-MMP primers designed to generate and amplify 264-bp (black arrowhead) and 185-bp (white arrowhead) products, respectively. mRNA transcripts were detected in cultured cardiac NC cells for both TIMP-2 and MT1-MMP. A 100-bp DNA ladder is also shown (std). No bands were detected if reverse transcriptase was omitted from the RT-PCR reaction (not shown). Scale bar = 15 μm in B (applies to A,B).

Because TIMP-2 expression in vivo was limited to cardiac NC cells, we tested whether blocking TIMP-2 synthesis altered proMMP-2 activation and cardiac NC migratory behavior. Cardiac NC cell cultures were established from neural tube explants of embryos electroporated in ovo. To determine whether proMMP-2 activation required TIMP-2 synthesis, cardiac NC cells were treated with antisense TIMP-2 or control FITC-labeled morpholino oligonucleotides and the relative levels of active and proMMP-2 in the conditioned medium were determined using zymography. In cultured cardiac NC cells from control morpholino-treated embryos, both proMMP-2 and active MMP-2 were detected in the conditioned medium and based on scanning densitometry, approximately 15–20% of the total MMP-2 was in the active form (Fig. 4A). In cultured cardiac NC cells from TIMP-2 antisense morpholino-treated embryos, only proMMP-2 was detected in the conditioned medium (Fig. 4A). We then tested the effect of TIMP-2 antisense control morpholinos on cardiac NC cell migration in vitro by using time-lapse imaging. Although only a subset of NC cells was labeled with morpholinos (Fig. 4C,D), control morpholino-treated NC cells migrated further than TIMP-2 antisense morpholino-treated cells (Fig. 4B). Evaluating TIMP-2 levels using immunostaining in these experiments was compromised by the difficulty in preserving TIMP-2 antigenicity while retaining the FITC-labeled morpholino oligonucleotides in the cells.

Figure 4.

Effect of tissue inhibitor of metalloproteinase (TIMP) -2 antisense morpholino oligonucleotides on cardiac neural crest (NC) cell migration and pro-matrix metalloproteinase-2 (proMMP-2) activation in vitro. Fluorescein isothiocyanate–labeled control and antisense TIMP-2 morpholino oligonucleotides were introduced into neural tubes in ovo, and cardiac NC cell cultures established from these neural tubes. NC cells were subsequently cultured for 18 hr in the absence of any additional morpholinos. Conditioned medium was collected and subjected to zymography. A: Effect of TIMP-2 antisense and control morpholinos on proMMP-2 activation. ProMMP-2 (72 kDa, black arrowhead) and active MMP-2 (62 kDa, white arrowhead) were both detected in the conditioned medium of cardiac NC cells treated with control morpholinos (Control). Antisense TIMP-2 morpholinos (AS-T2) blocked proMMP-2 activation as indicated by the loss of the 62-kDa band. B: Effect of TIMP-2 antisense and control morpholinos on cardiac NC cell migration. TIMP-2 antisense morpholinos (AS-T2) significantly decreased cardiac NC cell migration compared with the controls (*P < 0.05). Error bars represent the mean ± SEM. C: Cultured cardiac NC cells positive for control morpholino incorporation 18 hr after electroporation. D: Cultured cardiac NC cells positive for TIMP-2 antisense morpholino incorporation 18 hr after electroporation. Scale bar = 100 μm in D (applies to C,D).

Low levels of TIMP-2 promote the formation of a cell-surface proMMP-2/TIMP-2/MT1-MMP complex necessary for activation of proMMP-2 while higher levels block activation by sequestering the remaining TIMP-2-free MT1-MMP. Therefore, we next determined if adding exogenous TIMP-2 altered cardiac NC cell migration. We found low levels of exogenous TIMP-2 (250 ng/ml) significantly increased cardiac NC cell motility (Fig. 5). However, when TIMP-2 levels were increased to 1,250 ng/ml, NC cell motility returned to control levels. When TIMP-1 was added, there was no effect on NC cell migratory behavior showing that the effect of TIMP-2 was not a common characteristic of all TIMPs. Collectively, our results show that, in vitro, proMMP-2 activation by cardiac NC cells requires TIMP-2 synthesis and that TIMP-2 is an important mediator of cardiac NC cell migration.

Figure 5.

Effect of tissue inhibitor of metalloproteinases (TIMPs) on cardiac neural crest (NC) cell motility in vitro. TIMP-1 or TIMP-2 were added to the culture medium of explanted neural tubes, and the migratory behavior of emerging cardiac NC cells was monitored over a 5-hr period by time-lapse imaging as described in the Experimental Procedures section. TIMP-2 significantly increased NC motility at 250 ng/ml but had no effect at 1,250 ng/ml. TIMP-1 (250 ng/ml) had no effect on migration. Error bars represent the mean ± SEM. ***P < 0.001.

DISCUSSION

MMP activity, and its regulation by TIMPs, is an important mediator of epithelial–mesenchymal transitions and cell migration during development. NC cells are one of the most extensively migrating and invasive population of cells found in embryos. Here, we show that TIMP-2 is an important mediator of cardiac NC cell morphogenesis because (1) TIMP-2 is uniquely expressed in cardiac NC cells during their early migration in vivo, (2) increasing or decreasing exogenous TIMP-2 levels results in corresponding changes to cardiac NC cell migration in vitro, (3) cardiac NC cells require TIMP-2 to activate proMMP-2, and (4) antisense TIMP-2 morpholino oligonucleotides perturb normal embryonic development and impair cardiac NC cell migration in ovo.

TIMP-2 and TIMP-3 Expression in Cardiac NC Cells

TIMP-2 mRNA transcripts were first detected in cardiac NC cells when pioneering cardiac NC cells began emigrating from the neural tube. Morphological studies show that basement membrane-associated ECM is lost in areas occupied by invading NC cells (Brauer et al., 1983) and that MMP-2 associated with the basement membrane accumulates on NC cell surfaces and is redistributed at the leading edge of the NC migratory front (Cai et al., 2000). Active MMP-2 can degrade many ECM components, including type IV collagen, fibronectin, vitronectin, decorin, tenascin, laminin, and others (Okada et al., 1990; Imai et al., 1995, 1997; Siri et al., 1995). Interestingly, TIMP-2 expression was higher in the leading population of cardiac NC cells compared with the trailing NC cell population, precisely where the bulk of ECM degradation would likely occur as the NC cells invade the ECM. TIMP-2 promotes the formation of a proMMP-2/TIMP-2/MT1-MMP complex on the cell surface (Fig. 5; Nakahara et al., 1997; Butler et al., 1998; Zucker et al., 1998; Hernandez-Barrantes et al., 2000; Wang et al., 2000), where proMMP-2 can then be activated. Hence, TIMP-2 expression by cardiac NC cells may mediate cardiac NC cell migration through the ECM by facilitating the formation of such a complex on the NC cell surface.

TIMP-2 was expressed by cardiac NC cells but not by mesencephalic or trunk NC cells. Why TIMP-2 is expressed only in cardiac NC cells is unknown. Either TIMP-2 expression is intrinsic to this population of NC cells or it is induced/repressed by local signals. Positional and segmental genes along the anterior/posterior body axis dictate regional specificity to the neural tube and NC cells, and control the expression of many other genes. The expression pattern of TIMP-2 is very similar to the homeobox-containing transcription factor mafB/Kr, whose expression pattern is, in turn, regulated by endogenous posteriorizing factors (Grapin-Botton et al., 1998; Dupe et al., 1999; Marin and Charnay, 2000). Ablation/transplantation studies show that mesencephalic and trunk NC cells are incapable of generating mesenchymal cells competent to replace cardiac NC cells that otherwise normally participate in proper cardiac septation (Kirby, 1989). This finding suggests that cardiac NC cells are unique within the overall NC cell population and that TIMP-2 expression in these cells reflects this.

Of the two TIMPs examined, TIMP-3 was the earliest to be detected in chicken embryos, specifically within the notochord beginning at stage 8. Expression of TIMP-3 gradually widened to include the myocardium and endocardium, and by stage 14, also included the endothelium of blood vessels and the pharyngeal arch mesenchyme (Brauer and Cai, 2002). Like other TIMPs, TIMP-3 is capable of blocking the enzymatic activity of most of MMPs (Pavloff et al., 1992; Butler et al., 1999). The temporal and spatial distribution of TIMP-3 mRNA expression makes it unlikely to have a role in cardiac NC emigration or their early migration. But, its unique expression in the notochord suggests that TIMP-3 may participate in the development and remodeling of the floor plate, the adjacent endoderm, and/or sclerotome. NC cell migration is excluded from the notochord region due in part to the presence of extracellular chondroitin sulfate proteoglycans (Pettway et al., 1990, 1996; Perris et al., 1996). TIMP-3 is unique from the other TIMPs in that it is located almost exclusively within the ECM (Pavloff et al., 1992; Leco et al., 1994) and contains a domain that greatly enhances its ability to bind heparan sulfate and chondroitin sulfate proteoglycans (Butler et al., 1999; Yu et al., 2000). By binding to proteoglycans and accumulating in the ECM surrounding the notochord, TIMP-3 could contribute to the inhibition of NC cell migration into this area by restricting MMP activity.

Functional Role of TIMP-2

TIMP-2 null mutant mice develop normally and are fertile, and the only described phenotypic disorder is impaired proMMP-2 activation (Caterina et al., 2000; Wang et al., 2000). Because TIMP family members share a great deal of overlapping functional activities, other family members in these null mice may compensate for the permanent loss of TIMP-2. In our study, TIMP-2 loss-of-function in chicken embryos was specifically directed toward those cells when and where the TIMP-2 was first expressed. Our results show TIMP-2 expression is required for normal early chick development and cardiac NC migration. Blocking TIMP-2 translation using antisense morpholinos or overexpressing TIMP-2 in one- to four-cell stage of zebrafish embryos also leads to anterior–posterior patterning defects (Zhang et al., 2003). Therefore, TIMP-2 plays an important role in early embryonic development, which has not been recognized previously.

TIMP-2 is required for the activation of proMMP-2 by MT1-MMP. Studies show the N-terminal domain of TIMP-2 binds MT1-MMP and inhibits MT1-MMP activity, while the C-terminal end of TIMP-2 recruits proMMP-2 from the ECM. This trimeric complex then interacts with neighboring TIMP-2–free MT1-MMPs, and this MT1-MMP then initiates the activation of the bound proMMP-2 (Fig. 5; Nakahara et al., 1997; Butler et al., 1998; Zucker et al., 1998; Hernandez-Barrantes et al., 2000; Wang et al., 2000). However, excess TIMP-2 interferes with this activation by binding all available MT1-MMP molecules on the cell surface. Therefore, TIMP-2 either facilitates or inhibits activation of proMMP-2, depending on the levels of TIMP-2 relative to MT1-MMP (Strongin et al., 1995; Butler et al., 1998; Jo et al., 2000).

Under basal conditions, cultured cardiac NC cells expressed MT1-MMP, MMP-2, and TIMP-2 and generated and released active MMP-2 into the culture medium. Therefore, cardiac NC cells express all the components needed to generate active MMP-2 using the trimeric complex. Addition of a low concentration of TIMP-2 increased cardiac NC cell migration in vitro, whereas a higher concentration returned migration to control levels. This finding is expected if, at a lower concentration, TIMP-2 was facilitating proMMP-2 activation; but at a higher concentration, it began curtailing the amount of TIMP-2–free MT1-MMP on NC cell surfaces. Moreover, when antisense TIMP-2 morpholino oligonucleotides were introduced into cultured cardiac NC cells, there was a significant reduction in NC cell migration compared with controls. This finding coincided with a loss of active MMP-2. TIMP-1 is incapable of binding proMMP-2 or facilitating its activation (Will et al., 1996). Exogenous TIMP-1 had no effect on cardiac NC cell migration showing that enhanced NC cell migration was specific to TIMP-2. Thus, we conclude that the specific expression of TIMP-2 by cardiac NC cells is required for proMMP-2 activation in vitro and that migration is, at least partly, TIMP-2–dependent.

Once cardiac NC cells have undergone the epithelial–mesenchymal transition, MMPs and their inhibitors likely regulate subsequent cardiac NC cell migration and invasion through embryonic ECM. Active MMP-2 is generated in areas occupied by early migrating cardiac NC cells and blocking MMP activity inhibits cardiac NC cell migration both in vivo and in vitro (Cai et al., 2000; Cai and Brauer, 2002). MMP-2 protein accumulates on the surface of pioneering cardiac NC cells, and there is a concomitant loss of MMP-2 immunostaining within the ectodermal basement membrane as NC cells migrate over this extracellular substrate (Cai et al., 2000). The formation of a proMMP-2/TIMP-2/MT1-MMP complex enables cells to activate and localize MMP-2 activity to specific sites on the cell surface, which is essential for the formation of functional invadopodia and cell invasion (Nakahara et al., 1997; Coopman et al., 1998; Chen and Wang, 1999). TIMP-2 mRNA expression was greater in cardiac NC cells at the leading migratory front. Hence, the timely and localized expression of TIMP-2 may enable NC cells to locally generate active MMP-2 at specific sites, e.g., invadopodia of pioneering NC cells, by means of formation of this complex.

Our results show proMMP-2 activation by cardiac NC cells in vitro requires TIMP-2 and may involve the formation of a proMMP-2/TIMP-2/MT1-MMP ternary complex. An illustration showing our current working model for cardiac NC cells is shown in Figure 6. However, this working model may not be applicable to all NC cell populations. Although the formation of this ternary complex is thought to be the major mechanism for proMMP-2 activation, alternative activation mechanisms have been described that operate in the absence of TIMP-2 (Brooks et al., 1996; Morrison et al., 2001). For instance, αVβ3, an integrin expressed by NC cells (Delannet et al., 1994), binds to proMMP-2 and can dock proMMP-2 to the cell surface for activation by MT1-MMP in the absence of TIMP-2 (Brooks et al., 1996; Deryugina et al., 2001). Studies also suggest TIMP-2 and MT1-MMP may have quite different roles in morphogenesis, depending on whether cells are in two-dimensional or three-dimensional environments (Hotary et al., 2002, 2003; Wolf et al., 2003). Finally, there is compelling evidence that TIMP-2 exhibits functional properties independent of its MMP inhibitory activity through its interaction with cell surface integrins (Fernandez et al., 2003; Seo et al., 2003). Therefore, more detailed studies will be needed to determine the mechanisms various NC cell populations use to activate proMMP-2, to evaluate the possible extent each mechanism is operating at different axial levels, and to evaluate to what degree NC cell formation, migration, proliferation, and apoptosis at each axial level is dependent on MMP-2, TIMP-2, and MT1-MMP in vivo.

Figure 6.

Schematic representation of our current working model for a possible role of tissue inhibitor of metalloproteinase (TIMP) -2 expression during cardiac neural crest (NC) cell migration. Studies indicate that TIMP-2 serves to dock extracellular pro-matrix metalloproteinase-2 (proMMP-2) to cell surface membrane type-1 matrix metalloproteinase (MT1-MMP), thereby enabling MT1-MMP to initiate proMMP-2 activation. Based on our previous studies (Cai et al., 2000), the mesoderm synthesizes and secretes proMMP-2 into the extracellular matrix (ECM) before cardiac NC cell migration (1). As cardiac NC cells emerge from the neural tube, they begin to express TIMP-2 and MT1-MMP (2). The N-terminal domain of secreted TIMP-2 would then bind MT1-MMP while its C-terminal domain would bind extracellular proMMP-2, effectively recruiting proMMP-2 to the cardiac NC cell surface (3). Adjacent TIMP-2–free MT1-MMP would then activate proMMP-2, generating localized MMP-2 activity on the surface of cardiac NC cell invadopodia, resulting in ECM degradation (4).

EXPERIMENTAL PROCEDURES

RT-PCR and In Situ Hybridization

PolyA mRNA was extracted from stage 9–10 chicken embryos by using the Micro-Fast Track mRNA isolation kit (Invitrogen, Carlsbad, CA) and used for RT-PCR. Two sets of nested primers for MT1-MMP were designed based on the partial cDNA sequence for the chicken MT1-MMP homolog entered into the GenBank (accession no. BM489822) by the USDA/IFAFS Animal Genome Project. The first forward primer was 5′-TGCAGCAGTACGGCTATCTG-3′ and the first reverse primer was 5′-TGAAGGCAGAAGGTGATGTC-3′. For reamplification, the forward primer was 5′-ACTTCGGCCCATTCGGTAT-3′ and the reverse primer was 5′-CAATGCGGTAAGTCCCGAG-3′, with the final predicted product size of this RT-PCR being 185 bp. For chicken TIMP-2, the primers were designed based on the chicken TIMP-2 cDNA sequence found in GenBank (accession no. AF004664). The forward primer was 5′-GAACCCCATCAAGCGAATCC-3′ and the reverse primer was 5′-TGGTCGCCATGGTCTACCCG-3′, resulting in a predicted product size of 264 bp. mRNA samples were also subjected to RT-PCR in the absence of reverse transcriptase to ensure that the amplified products were not due to genomic DNA contaminants.

For in situ hybridization studies, digoxigenin-labeled antisense and sense riboprobes for chicken TIMP-2 and TIMP-3 (GenBank accession no. M94531) were prepared using an in vitro transcription method (Genius 4 RNA Labeling Kit; Roche Molecular Biochemicals, Indianapolis, IN). Stage 8–14 chicken embryos were fixed in 4% paraformaldehyde overnight, rinsed in phosphate-buffered saline (PBS), and subjected to whole-mount in situ hybridization as previously described (Brauer and Cai, 2002). In some cases, TIMP-2–hybridized embryos were subsequently immunostained as whole-mounts by using the monoclonal antibody HNK-1 to identify migrating NC cells, post-fixed in 4% paraformaldehyde, and photographed. Embryos were then embedded in freezing compound and sectioned on a cryostat for further examination. Hybridization by sense riboprobes for TIMP-2 and TIMP-3 was not detected in embryos at any stage examined.

Immunostaining

For TIMP-2 immunostaining, cultured cardiac NC cells were fixed for 15 min in cold 80% acetone, rehydrated, and equilibrated in phosphate-buffered saline (PBS) 0.1% Tween-20. After incubation in blocking buffer (PBS-0.1% Tween-20 containing 1% donkey serum) for 15 min, NC cells were incubated in a mouse monoclonal antibody directed against human TIMP-2 (1 μg/ml, Chemicon International, Inc., Temecula, CA). After washing, the cells were incubated in donkey anti-mouse Alexa 568 (Molecular Probes, Inc., Eugene, OR). Cardiac NC cells were then examined on an epifluorescence-equipped microscope and images captured by using a cooled digital CCD camera. Some cardiac NC cell cultures were also immunostained with rabbit anti-chicken MMP-2 as previously described (Cai et al., 2000).

Migration Assays

The migratory behavior of cardiac NC cells was assessed as described previously (Cai and Brauer, 2002). Briefly, neural tube segments from the cardiac NC axial level (mid-otic placode through somite level 3) of stage 10 to 10+ chicken embryos were placed into fibronectin-coated tissue culture dishes containing Medium 199-1% heat-inactivated MMP-depleted fetal bovine serum. In some cases, recombinant human TIMP-1 or TIMP-2 (Chemicon International) was added to the medium when the neural tubes were explanted, and in other experiments, neural tubes were subjected to electroporation just before explantation. Cultures were then maintained in a 5% CO2/95% atmosphere at 37°C using a microincubation chamber located on an inverted microscope. After 1 hr, the migratory behavior of cardiac NC cells emerging from the neural tube explants was assessed by capturing an image every 5 min for the next 5 hr. From these images, the average rate of cell motility and the distance NC cells migrated from their initial starting position was calculated for randomly chosen NC cells. All measurements were made without prior knowledge as to which treatment group they belonged. Statistical analysis was performed by using a Mann–Whitney nonparametric test or a Kruskal–Wallis nonparametric analysis of variance test coupled with Dunn's multiple-comparison posthoc tests using Prism software (GraphPad Software, Inc., San Diego, CA).

Electroporation and Antisense Treatment of Cardiac NC Cells

Eggs were incubated until they reached stage 9 to 10 (before detectable TIMP-2 expression in vivo), windowed, and 9.2 nl of FITC-labeled morpholino oligonucleotides (1 mM, Gene Tools, Inc., Philomath, OR) was microinjected into the lumen of the neural tube between the mid-otic to third somite level. The chicken TIMP-2 antisense morpholino oligonucleotide was designed to hybridize to the 5′ untranslated region and overlap the initiation codon for transcription (5′-CCGGGCATCGTGCGGCGCG GAATT-3′). A standard control sequence, supplied by the manufacturer, was used as a control morpholino (5′-CCTCTTACCTCAGTTACAATTTATA-3′). After injection, the embryos were subjected to electroporation by placing each electrode approximately 0.5 mm on either side of the embryo at the level of the injection site and the embryo given 3 pulses of 100-msec duration at 15–18 V by using a square wave generator (model ECM 830, BTX-Gentronics, Inc., San Diego, CA). The polarity of the electrodes was then reversed and the embryos were subjected to another round of electroporation to introduce labeled morpholino oligonucleotides into both sides of the neural tube. Neural tube segments from the cardiac NC axial level of electroporated embryos were then isolated and used to establish cardiac NC cell cultures. Alternatively, eggs were resealed with surgical tape after electroporation and re-incubated for 18 hr, at which time the embryos were fixed in 4% paraformaldehyde, immunostained with HNK-1 as whole-mounts, photographed, and then embedded, sectioned on a cryostat, and examined; images were captured on an epifluorescent-equipped microscope. Sham-treated embryos were electroporated after injection with 10 mM Fast Green in PBS rather than morpholino oligonucleotides.

Zymography

Cardiac NC cells were cultured for 18 hr, and samples of NC conditioned medium were subjected to gelatin zymography as previously described (Cai et al., 2000). The electrophoretic mobility of proteins with gelatinase activity was revealed by a lack of Coomassie staining at positions reflecting their electromobility within the gel. The relative degree of gelatinolysis by proMMP-2 and active MMP-2 was measured by scanning densitometry.

Acknowledgements

The authors thank Drs. Elizabeth Hahn-Dantona, Ph.D. (Holland Laboratory, Rockville, MD) and James P. Quigley, Ph.D. (Scripps Research Institute, La Jolla, CA) for TIMP-2 cDNA and the MMP-2 antibody. The authors also thank Jenny Balch and Trang Ho for their technical assistance on this project. P.R.B. was sponsored by a grant from American Heart Association-Heartland Affiliate.

Ancillary