Hypoxia, which characterizes ischemia, trauma, inflammation, and solid tumors, recruits monocytes, immobilizes them, and alters their function, leading to an anti-inflammatory and proangiogenic phenotype. Monocyte extravasation from the circulation and their migration in tissues are partially mediated by the balance between matrix metalloproteinases (MMPs) and their tissue inhibitors (TIMPs). The mechanisms evoked by hypoxia that regulate monocyte migration and activation are not entirely clear. Specifically, the effect of hypoxia on TIMPs in these cells has hardly been investigated. We show that hypoxia reduces TIMP-2 secretion from human primary monocytes and from the monocyte-like cell lines U937 and THP-1 by three- to fourfold (P < 0.01), by inhibiting TIMP-2 transcription through mechanisms that involve the transcription factor SP-1. Hypoxia also lowers TIMP-2 protein secretion from human endothelial cells (by 2-fold, P < 0.05). TIMP-2 levels do not influence the reduced migration of THP-1 cells in hypoxia; however, low TIMP-2 levels enhance endothelial cell migration/proliferation, their ability to form tubelike structures in vitro, and the appearance of mature blood vessels in a Matrigel plug assay in vivo. Thus we conclude that reduced TIMP-2 levels secreted from both hypoxic monocytes and endothelial cells are proangiogenic.
- low oxygen tension
low oxygen tensions (hypoxia) are a common denominator in different physiological and pathological conditions (e.g., ischemia, infection, inflammation, and solid tumors). Monocytes are recruited in large numbers from the circulation to the inflicted tissues, where they accumulate within hypoxic sites (21), mature into macrophages (2), and change their morphology, survival, and functions (18, 19, 22). Generally, hypoxia polarizes macrophages toward an alternative activation, primarily by upregulating the transcription factors hypoxia-inducible factor (HIF)-1 and -2, which in turn activate many migratory, inflammatory, and proangiogenic genes (17, 20). Monocytes/macrophages are particularly important and are considered orchestrators of hypoxia-induced angiogenesis, as they secrete a myriad of proangiogenic mediators, including vascular endothelial growth factor (VEGF) and fibroblast growth factors (FGFs) (5, 17). Most studies are focused on characterization of hypoxia-induced genes in macrophages, particularly in tumor-associated macrophages (TAMs), and reveal overlap in the expression of such genes, together with some differences between mononuclear phagocytes at various maturation stages (4, 19, 48). However, only limited information is available on the response of monocytes to local hypoxia.
Hypoxia-induced migration of monocytes into tissues and their trapping within the hypoxic sites are regulated by many chemoattractants, including cytokines, chemokines, and their receptors (4, 16). Matrix metalloproteinases (MMPs), a group of secreted and membranal proteases that can collectively degrade all components of the extracellular matrix (ECM), play a critical role in the migration of these cells. In addition, MMPs are important mediators in other cellular processes, including angiogenesis (7), as they can degrade diverse substrates (e.g., growth factors, kinases, adhesion molecules, cytokines, other MMPs) (6, 24, 27). To avoid excessive ECM degradation and tissue damage, MMPs are tightly regulated at several levels, including transcription, secretion as inactive zymogens that require activation, and inhibition of their activity by the family of four tissue inhibitors of MMPs (TIMPs). Like MMPs, TIMPs are also tightly regulated, as they are involved in cell migration, apoptosis, proliferation, and angiogenesis, either as inhibitors of MMPs or independently (9).
Hypoxia was shown to differently affect the expression of MMPs in monocytes/macrophages (e.g., MMP-9 and -25 are downregulated and MMP-1, -7, -12, -16, and -19 are upregulated) (4, 5, 22, 32, 48). However, the effects of hypoxia on TIMPs and their role in migration and angiogenesis in these cells were not investigated.
Expression of all four mammalian TIMPs has been found in most normal tissues (25, 39), although TIMP-1, TIMP-3, and TIMP-4 demonstrate a more selective pattern than TIMP-2. MMP-2, a constitutively expressed MMP with exceptional importance in inflammation and cancer (3, 44), is specifically inhibited by TIMP-2, which is also constitutively expressed in all normal tissues but demonstrates reduced expression in human cancer cells (11). In addition to its ability to inhibit MMP-2 in high concentrations, TIMP-2 is unique in that its low levels activate MMP-2 by preventing autodegradation of MMP-14, the main activator of MMP-2 (14). Thus the exact stoichiometric ratio between TIMP-2 and MMP-2 is critical to the net activity of MMP-2. When TIMP-2 is not in complex with MMP-2 or MMP-14 it has additional functions that are independent of MMPs. In particular, TIMP-2 can inhibit proliferation of endothelial cells, promote their differentiation into a quiescent state, enhance the expression of reversion-inducing cysteine-rich protein with Kazal motifs (RECK), the membrane-associated inhibitor of MMPs, and inhibit endothelial cell migration, collectively acting to inhibit angiogenesis (40, 41). The regulation of TIMP-2 expression remains not fully elucidated, although several SP-1 elements in its promoter (10, 13, 50), which are characteristic of housekeeping genes, may explain its constitutive expression. Despite its apparent importance, the effect of hypoxia on TIMP-2 expression, particularly in monocytes, has not been investigated.
In this study we evaluated the effects of hypoxia on TIMP-2 secretion in primary human monocytes and in the two monocyte-like cell lines THP-1 and U937, which are widely accepted as monocyte models (15, 45). We explored the mechanisms regulating these effects and the biological relevance of changes in TIMP-2 levels, both on monocyte migration and on angiogenic parameters in endothelial cells. We found that the proangiogenic effects of hypoxia are not limited to transcriptional induction of proangiogenic factors via HIF transcription factors but also extend to transcriptional suppression of the regulatory protein TIMP-2.
MATERIALS AND METHODS
Reagents and antibodies.
Buffy coats from 500 ml of peripheral blood of healthy volunteers were purchased from the Blood Services Center, Magen David Adom (Tel-Hashomer, Israel). All media, fetal calf serum (FCS), cell culture reagents, and XTT kits were obtained from Biological Industries (Kibbutz Beit-haemek, Israel). Endothelial cell growth factor (ECGF) was from Biomedical Technologies (Stoughton, MA). BD Matrigel basement membrane matrix growth factor reduced was from BD Biosciences (Bedford, MA). Bovine serum albumin (BSA), PKH26 reference beads, and deoxynucleotides were from Sigma (St. Louis, MO). Random hexamers, RNAguard, and [γ-32P]ATP were from Amersham Pharmacia Biotech (Piscataway, NJ). TriReagent was from Molecular Research Center (Cincinnati, OH). Amplification was carried out with Moloney murine leukemia virus (MMLV) reverse transcriptase from US Biochemicals (Cleveland, OH) and with AmpliTaq Gold Taq Polymerase from Roche Molecular Systems (Branchburg, NJ). Monocyte chemoattractant protein (MCP)-1 and human DuoSet TIMP-2 ELISA kits were purchased from R&D Systems (Minneapolis, MN). Recombinant TIMP-2 was from Calbiochem (Darmstadt, Germany). Mouse monoclonal antibodies for nuclear factor Y subunit A (NF-YA), SP-1, SP-3, AP-2, c-Fos, and CCAAT/enhancer binding protein (C/EBP)β were from Santa Cruz Biotechnology (Santa Cruz, CA). Mouse monoclonal anti-TIMP-2 was from Oncogene (Boston, MA). Horseradish peroxidase (HRP)-conjugated donkey anti-mouse IgG was from Jackson ImmunoResearch Laboratories (West Grove, PA).
Human monocytes were separated from buffy coats by density centrifugation on Ficoll-Hypaque, washed with PBS, and adhered to 24-well plates (107 cells/well) with RPMI 1640 medium and 20% FCS for 2 h. Monocyte purity was routinely ≥82 ± 0.5% as evaluated by flow cytometry of cells labeled with anti-CD14 and anti-CD11b. The human monocyte-like cell lines U937 and THP-1 were cultured in RPMI 1640 medium with 10% FCS and antibiotics. To avoid masking of signals by exogenous stimuli, untreated cells were serum starved during their exposure to normoxia or hypoxia. The human endothelial cell line EaHy926 (gift of Dr. C. J. Edgell, University of North Carolina, Chapel Hill, NC) was cultured in DMEM with 2% glutamine, 10% FCS, 2% hypoxanthine-aminopterin-thymidine (HAT), and 1% antibiotics. Before the beginning of the experiments the medium was replaced with serum-free medium with 0.1% BSA. Human umbilical vein endothelial cells (HUVEC) isolated from different donors were cultured in M-199 medium with 15% FCS, 2% glutamine, 1% ECGF, 0.06% heparin, and 1% antibiotics on 10 μg/ml fibronectin (FN)-coated dishes and used at passages 3–5. Before the beginning of the experiments, the medium was replaced with serum-free Bio-MPM1 medium with 0.1% BSA, 1% glutamine, and 1% ECGF. This part of the study was approved by the Carmel Medical Center Helsinki Committee. Viability of all cells was determined with XTT kits.
Normoxic and hypoxic conditions.
For normoxia, cells were incubated in a regular incubator (21% O2, 5% CO2, 74% N2). Hypoxic incubation was performed in a sealed anaerobic workstation (Concept 400, Ruskin Technologies, Leeds, UK), where the hypoxic environment (<0.3% O2, 5% CO2, 95% N2), the temperature (37°C), and the humidity (>90%) were kept constant. Supernatant samples were taken at the end of the exposure to hypoxia, and the partial pressures of O2 (mean of 25 ± 0.8 mmHg), and CO2 (mean of 32.9 ± 0.3) as well as pH (mean values of 7.3 ± 0.04) were determined with an ABL510 blood gas analyzer (Radiometer).
Human TIMP-2 ELISA kits were used according to the manufacturer's instructions.
Semiquantitative and real-time PCR analyses.
Total RNA was extracted from 107 U937 or THP-1 cells with TriReagent, and 5 μg was transcribed to cDNA and used to semiquantitatively amplify TIMP-2 mRNA. Amplification was performed only within the linear range (27 cycles with annealing temperature of 63°C for 30 s), and relative comparison between the samples was allowed after normalization for 18S rRNA, which does not change in hypoxia, in contrast to other commonly used housekeeping genes (49), and was also amplified within its linear range (18 cycles with annealing temperature of 58°C for 30 s). TIMP-2 mRNA expression was also quantified by real-time PCR using the TaqMan assay on demand kit with the ABI-PRISM 7000 (Applied Biosystems, Foster City, CA), as described previously (32), with the normoxic nonstimulated sample used as a calibrator to allow comparison of relative quantity between the samples.
Electrophoretic mobility shift assay.
Nuclear extracts were prepared from 2 × 107 THP-1 cells as described previously (31). The cloned human TIMP-2 promoter (gift of Prof. Y. DeClerck, University of Southern California, Los Angeles, CA) was amplified to yield a 347-bp product of the proximal promoter, and 2 ng was 5′-end labeled with [γ-32P]ATP and incubated with 6 μg of nuclear extracts in BRM buffer (10 mM HEPES pH 7.9, 36 mM KCl, 3.6% glycerol, 1.8 mM DTT) with 0.4 μg/μl poly(dI-dC) for 30 min at room temperature. In several experiments unlabeled oligonucleotides containing consensus sequences for transcription factors and flanking sequences (Table 1) were added in a 100-fold molar excess or nuclear extracts were preincubated with antibodies for 30 min at 4°C before addition of the reaction mixture. DNA-protein complexes were separated on 4% PAGE, and the gel was dried and autoradiographed.
THP-1 monocytes (6 × 107) were fixed (11% formaldehyde, 0.1 M NaCl, 1 mM EDTA, 500 μM EGTA, 500 mM HEPES, pH 8) and consecutively washed with cold PBS, buffer I (0.25% Triton X-100, 10 mM EDTA, 0.5 mM EGTA, 10 mM HEPES, pH 6.5), and buffer II (200 mM NaCl, 1 mM EDTA, 0.5 mM EGTA, 10 mM HEPES, pH 6.5). Next, cells were resuspended in lysis buffer (1% SDS, 10 mM EDTA, 50 mM Tris·HCl, pH 8.1, 1 mM PMSF, 25 μg/ml aprotinin, 25 μg/ml leupeptin, 10 mM iodoacetamide), and pelleted nuclei were sonicated eight times for 10 s to receive 700- to 800-bp-long DNA fragments, diluted in 1 ml of dilution buffer (1% Triton X-100, 2 mM EDTA, 150 mM NaCl, 20 mM Tris·HCl, pH 8.1), divided into aliquots, and stored at −80°C. One hour of preclearing with protein G beads blocked with salmon sperm DNA preceded overnight incubation at 4°C of the nuclear extracts with specific antibodies (10 μg each antibody). One sample from each treatment was left with no antibody and served as control (no IP). Immunocomplexes were precipitated for 1 h at 4°C by incubation with new blocked beads. Precipitates were consecutively washed with cold buffer TSE I (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 150 mM NaCl, 20 mM Tris·HCl, pH 8.1), buffer TSE II (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 500 mM NaCl, 20 mM Tris·HCl, pH 8.1), and buffer III (0.25 M LiCl, 1% NP-40, 1% deoxycholate, 1 mM EDTA, 10 mM Tris·HCl, pH 8.1) and twice with Tris-EDTA (TE) buffer. Pellets were incubated with freshly prepared extraction buffer (1% SDS, 0.1 NaHCO3) overnight at 65°C to reverse formaldehyde cross-linking. DNA was extracted with phenol-chloroform and ethanol precipitated with 3 M sodium acetate pH 5.2 and 20 μg of glycogen. DNA from the control tubes and from the samples was amplified with the same primers used for electrophoretic mobility shift assay (EMSA) and nested primers to yield a 320-bp fragment of TIMP-2 proximal promoter.
Western blot analyses.
Cellular extracts of THP-1 cells were lysed in RIPA buffer. Western blot analyses of equal amounts of protein or equal volumes were performed as described (32), with rabbit polyclonal anti-SP-1 and anti-NF-YA (diluted 1:1,000) and HRP-conjugated donkey anti-rabbit IgG (diluted 1:5,000). To evaluate the SP-1 phosphorylation state, anti-SP-1 and anti-phospho-SP-1 (Active Motif, Carlsbad, CA) were also used.
Expression and secretion of MMP-2 were analyzed by zymography to detect both the zymogen and the activated forms of the enzyme, as described previously (32), and were compared with the molecular standards (Chemicon, Temecula, CA) that were run on each gel.
Migration assay for monocytes.
Migration assays were performed in Transwells (5-μm pores, Costar, Corning, NY) coated with the ECM protein FN (100 μg/well), as described previously (32), with 5 × 105 THP-1 cells added to the upper chamber and chemokine (C-C motif)2 (CCL2)/MCP-1 (120 ng/ml) added to the lower chamber. In some wells anti-TIMP-2 (200 ng/well) or recombinant TIMP-2 (10 or 100 ng/ml) was added to the upper chamber. MCP-1-directed migration was determined after 72 h by adding a fixed volume of fluorescent beads (200 μl of PKH26 reference beads) to the cells that migrated to the lower chamber and determining by flow cytometry their ratio in the sample relative to the total number of cells.
In vitro “wound assay.”
EaHy926 monolayers (2 × 106 cells) in six-well dishes were wounded with a wooden toothpick after overnight incubation, and the line of injury was marked. Detached cells were washed away with medium, and cells were incubated in serum-free medium in normoxia or hypoxia with or without recombinant TIMP-2 (20 ng/ml) or anti-TIMP-2 (20 ng/ml). Images of the field of injury were acquired at the beginning of the experiment and after 48 h and were analyzed with ImagePro plus 4.5 software.
In vitro tube formation assay.
The tube formation assay was carried out by coating 24-well plates with 200 μl/well of Matrigel basement membrane extract at 4°C and then allowing it to polymerize at 37°C for 2 h. EaHy926 cells (0.5 × 106) were then seeded in triplicate in 1 ml of DMEM with 2% FCS and incubated for 48 h in normoxia or hypoxia with or without recombinant TIMP-2 (20 ng/ml) or anti-TIMP-2 (20 ng/ml). Formation of tubelike structures was observed under a phase-contrast microscope, and the number of closed lumens per microscopic field (at ×40 magnification) was counted in three separate fields.
In vivo plug assay.
Liquid Matrigel (0.4 ml) was mixed with 20 ng/ml of recombinant TIMP-2 and injected subcutaneously into the flank of BALB/c mice. As a control, Matrigel was mixed with serum-free DMEM and injected as above. Matrigel plugs were surgically removed after 7 days and photographed to give visual assessment of angiogenesis.
All values are presented as means ± SE. Significance between two groups was determined by two-tailed unpaired t-test. Differences between three or more experimental groups were analyzed by analysis of variance (ANOVA) and the Student-Newman-Keuls multiple comparisons test. P values exceeding 0.05 were not considered significant.
Hypoxia reduces TIMP-2 secretion from monocytes but does not affect their viability.
Preliminary studies revealed that incubation in hypoxia for 24, 48, and 72 h did not cause significant cell death relative to the normoxic controls in primary monocytes or U937 or THP-1 cells. After 72 h of incubation in hypoxia, cell viability was evaluated by XTT assay to be 1.05 ± 0.18-, 0.85 ± 0.12-, and 0.82 ± 0.04-fold relative to the normoxic control, respectively. Only after 96 h of hypoxic exposure did ∼50% of all cells die (P < 0.05, data not shown). These results were also confirmed by Trypan blue exclusion test (data not shown). To examine the effects of hypoxia on TIMP-2, and to minimize the number of primary monocytes required, the experiments were conducted at 72 h, unless otherwise stated. The constitutive expression of TIMP-2 resulted in its gradual accumulation in normoxic supernatants. Despite the differences in the ability of normoxic monocytes to secrete TIMP-2 to the supernatants (∼10- to 100-fold less than U937 and THP-1 cells, respectively), in hypoxia TIMP-2 levels were reduced in all cells tested by about fivefold (P < 0.001, P < 0.01, and P < 0.01, respectively; Fig. 1).
Hypoxia reduces TIMP-2 mRNA accumulation in monocytes.
To examine whether hypoxia-induced reduction of TIMP-2 occurs at the transcriptional level, we determined the relative TIMP-2 mRNA levels in different time periods by a semiquantitative approach and then confirmed our results by real-time PCR at 72 h (Fig. 2). In normoxia TIMP-2 was constitutively transcribed in both U937 and THP-1 monocyte-like cells, resulting in a constant level of TIMP-2 mRNA throughout the duration of the experiment. Hypoxia gradually inhibited its accumulation in the monocyte-like cells, so that after 72 h of incubation TIMP-2 mRNA levels were reduced by fivefold for U937 cells (P < 0.05; Fig. 2, A and C) and threefold for THP-1 cells (P < 0.01; Fig. 2, B and D). Real-time PCR analysis confirmed these results, showing that hypoxia decreased the steady-state levels of TIMP-2 mRNA by 3.8-fold in U937 cells (P < 0.001; Fig. 2E) and by 4.5-fold in THP-1 cells (P < 0.01; Fig. 2F).
Hypoxia transcriptionally regulates TIMP-2 in monocytes.
Next, we used EMSA to examine the effects of hypoxia on the binding of transcription factors to TIMP-2 proximal promoter in THP-1 monocyte-like cells. Figure 3B shows constitutive binding of three complexes that we named heavy (H), light (L), and very light (VL), that did not change in normoxia over time. However, in hypoxia the binding of all three complexes was gradually reduced, leaving the promoter bare after 48 h [reduction of 35-fold (P < 0.01), 4-fold (P < 0.01), and 9-fold (P < 0.01) for the H, L, and VL complexes, respectively]. To identify the composition of transcription factors in each complex, we competed the binding to the labeled promoter with 100× molar excess of unlabeled boxes, which include flanking sequences and putative binding sites for several transcription factors (depicted in Table 1 and Fig. 3A). The three complexes were not affected by the excess of the unlabeled AP-1/AP-2, CAAT, or NF-κB/NF-IL-6 boxes (Fig. 3C), but the SP-1/SP-3 box simultaneously competed away the binding of the H (by 43%, P < 0.001), L (by 44%, P < 0.001), and VL (by 47%, P < 0.001) complexes, suggesting a pivotal role for either SP-1 or SP-3. Direct binding to each box, in the presence of specific antibodies or unlabeled boxes, showed that anti-SP-1, but not anti-SP-3, supershifted some of the complex binding to the SP-1/3 box (Fig. 4A, arrow), implicating SP-1 as a binding transcription factor for TIMP-2 promoter. Hypoxia and the unlabeled SP-1/SP-3 box completely abolished binding, whereas competition with unlabeled irrelevant sequence of similar length (a STAT-1 box) did not change the binding, demonstrating binding specificity. Similar supershifting was evident for the binding of NF-YA to the CAAT box, and again the CAAT box, but not the irrelevant box, competed with this binding. No binding complexes or only weak binding were observed for the AP-1/AP-2 and NF-κB/NF-IL-6 boxes (data not shown).
These findings were further corroborated by the chromatin immunoprecipitation (ChIP) approach, which reflects the in vivo binding of transcription factors to a larger part of TIMP-2 promoter in fixed cells. ChIP analysis (Fig. 4B) demonstrated that in normoxia only SP-1 directly bound to the TIMP-2 promoter in THP-1 cells, whereas in hypoxia the promoter was bare, and none of the other transcription factors examined was bound to it.
To explain the hypoxia-induced dissociation of SP-1 from the TIMP-2 promoter, we looked at the phosphorylation state of both SP-1 and NF-YA. Two bands were observed for each protein, where the higher band represented the phosphorylated form. Figure 4, C, E, and F, shows that hypoxia did not change the amount of phosphorylated NF-YA but increased that of SP-1 by 46% (P < 0.01). To further highlight the SP-1 state of phosphorylation, we used different antibodies that recognize the total amount of SP-1 or specifically SP-1 phosphorylated on Ser101. The total amount of SP-1 remained unchanged with a slight shifting up in hypoxia, indicating again its increased phosphorylation, and this is clearly shown in the Ser101 phosphorylated SP-1 form that appears in hypoxia (Fig. 4D).
Hypoxia inhibits secretion of MMP-2.
Since many of the effects of TIMP-2 are MMP-2 dependent, we evaluated the accumulation of proMMP-2 and partially activated MMP-2, which is an additional activated form of MMP-2 (37), in THP-1 cell culture supernatants (Fig. 5A). ProMMP-2 and partially activated MMP-2 were constitutively expressed, but hypoxia inhibited proMMP-2 (by 49% P < 0.001) and abolished the expression of the partially activated MMP-2.
Hypoxia reduces migration of monocytes by mechanisms unrelated to TIMP-2.
We next examined whether changes in TIMP-2 levels or in the MMP-2-to-TIMP-2 ratio affected cell migration toward CCL2/MCP-1. THP-1 cells were incubated in FN-coated Transwells and subjected to normoxia or hypoxia, with addition of recombinant TIMP-2 (10 or 100 ng/ml) or anti-TIMP-2 (200 ng). In normoxia almost all monocyte-like cells migrated (95 ± 1.4%; Fig. 5B), and addition of either the antibody or the recombinant protein had no effect. Hypoxia decreased migration by 27% (P < 0.001), and addition of either anti-TIMP-2 or recombinant TIMP-2 had no effect.
Hypoxia-induced reduction of TIMP-2 enhances proangiogenic properties of endothelial cells.
To examine the effects of TIMP-2 on endothelial cell migration we first considered its endogenous production. Primary HUVEC or EaHy926 cells were incubated in normoxia and hypoxia for 48 h. Hypoxia did not change endothelial cell viability (0.97 ± 0.02 and 0.95 ± 0.05 relative to their normoxic counterparts) but reduced TIMP-2 (Fig. 6A) by 2.4-fold (P < 0.05) in both HUVEC and EaHy926 cells. In contrast, hypoxia increased the secretion of the active form of MMP-2 (Fig. 6B) by 42% in HUVEC (P < 0.05) and by 54% in EaHy926 cells (P < 0.05). Since endothelial cells migrate as one sheet, migration and proliferation are inseparable. The wound assay performed in EaHy926 cells (Fig. 6, C and D) revealed that hypoxia increased migration/proliferation of endothelial cells as assessed by the 96% increase in the mean length of the area to which the cells migrated (P < 0.05). Addition of TIMP-2 antibody in normoxia elevated the mean length of migration by 87% (P < 0.05), while recombinant TIMP-2 did not influence migration. In contrast, in hypoxia addition of anti-TIMP-2 did not affect the enhanced migration/proliferation of endothelial cells, while recombinant TIMP-2 inhibited their migration to the level observed in normoxic conditions.
To evaluate the effects of TIMP-2 on another step of the angiogenic process we measured the ability of EaHy926 cells to spontaneously form a tubular network in vitro on Matrigel-coated wells. Closed lumens were counted after 48 h of incubation, as EaHy926 cells form tubes in a later kinetics than other endothelial cells (1). Morphologically, normoxic tubes consisted of three or four rows of cells, generating large and closed lumens, whereas in hypoxia more sprouting cells were observed and cells formed smaller lumens with only one or two rows of cells (Fig. 6E). Addition of TIMP-2 antibody to EaHy926 cells in normoxia elevated the number of tubes by 36% (P < 0.001), whereas recombinant TIMP-2 had no effect. In contrast, in hypoxia addition of anti-TIMP-2 did not change the number of tubes relative to normoxia, but recombinant TIMP-2 decreased it back to normoxic levels (P < 0.001 relative to hypoxia). No effect of anti-TIMP-2 or recombinant TIMP-2 on morphology of endothelial cells was detected (data not shown).
To evaluate the effects of TIMP-2 on in vivo angiogenesis, we used the Matrigel plug assay, in which liquid Matrigel was injected subcutaneously, with or without recombinant TIMP-2 in a concentration similar to that measured in the hypoxic microenvironment, and harvested plugs after 7 days. Gross morphology of the plugs showed transparent control plugs with minimal blood vessels, whereas addition of recombinant TIMP-2 resulted in opaque and reddish plugs with more visible blood vessels, suggesting an increase in angiogenesis (Fig. 6G).
The detailed regulation of TIMP-2 in the monocytic lineage and the effects of hypoxia on its expression have not yet been studied. Here we demonstrate that hypoxia reduces TIMP-2 in monocytes and monocyte-like cells by inhibiting its transcription. As hypoxia also reduces secretion of TIMP-2 from endothelial cells, the overall reduced TIMP-2 levels in the microenvironment, achieved by the effects of hypoxia on both monocytes and endothelial cells, do not play a role in the decreased migration of monocytes. However, they elevate the ability of endothelial cells to migrate/proliferate, to form tubes in vitro, and to generate blood vessels in vivo, thus enhancing angiogenesis.
We demonstrate a constitutive expression of TIMP-2 mRNA in normoxic monocyte-like cells, as is common in most cell types. In our study, this is mediated by the binding of the transcription factor SP-1, which was mainly considered as a constitutive activator of housekeeping genes but is now known to modulate the expression of genes implicated in the control of essential cellular processes (29, 42). SP-1 was shown to bind the TIMP-2 promoter in fibroblasts (13, 50). SP-1 phosphorylation, which may occur on several sites, determines its DNA-binding and transactivation abilities (35, 42, 46). Hypoxia was shown to both enhance (38) and reduce (8) SP-1 activity; however, the effect of hypoxia on SP-1 phosphorylation was not yet demonstrated.
We found three complexes that bound the TIMP-2 proximal promoter but migrated differently, suggesting different protein composition, and competition experiments demonstrated the pivotal role that SP-1 plays in each of these. Both EMSA and ChIP analyses showed that SP-1, but not other putative transcription factors (e.g., NF-κB, NF-IL-6, AP-1, or AP-2), was directly bound to TIMP-2 promoter. SP-1 can be phosphorylated on many residues, and its state of phosphorylation was both negatively and positively associated with DNA binding. Here we show that hypoxia caused complete dissociation of the three complexes and increased SP-1 phosphorylation, which has been linked with SP-1 dissociation from DNA, specifically on residue Ser101 (26, 35, 42, 46).
These results together with the supershifting of NF-Y, lead us to propose a cooperative model for the binding of transcription factors to TIMP-2 promoter. In this model, SP-1 binds first and is responsible for recruiting other transcription factors, such as NF-Y and possibly NF-κB or NF-IL-6, generating large and stable complexes. These structures deny access of competitive boxes or specific antibodies, explaining the lack of competition in the EMSA assay or additional proteins precipitating TIMP-2 promoter in the ChIP assay. Increased SP-1 phosphorylation in hypoxia causes its dissociation from TIMP-2 promoter, thereby destabilizing the binding of other transcription factors and leaving the promoter unoccupied. This model resembles the cooperative binding of regulatory factor X (RFX) to major histocompatibility complex (MHC) class II promoters, which is required for subsequent recruitment of NF-Y and cAMP-response element binding protein (CREB), where any mutation in the RFX subunits prevents binding of all transcription factors, leading to a “bare promoter” (33, 47). In this model, the three bands observed represent intermediate products in the process of the assembly of the larger complex.
Additional studies are required to determine the identity of the proteins binding to the TIMP-2 promoter and the process of their assembly into a complex. Likewise, the effects of hypoxia that trigger signaling events that lead to SP-1 phosphorylation and its dissociation from TIMP-2 promoter also merit further investigation. In this context, high Ca2+ concentrations, which characterize the hypoxic cytoplasm and were shown to inhibit NF-Y (34, 43), or hypermethylation, which was shown to silence TIMP-2 promoter (11, 30), may also be relevant in TIMP-2 regulation, although an association between hypoxia and hypermethylation has not yet been established.
MMPs, including MMP-14 and MMP-2, are pivotal in the regulation of monocyte/macrophage motility (23, 27). The effect of TIMP-2 on motility may be complex because of its opposite influences on MMP-14/MMP-2, where high and low concentrations inhibit or activate them, respectively. Moreover, MMP-independent functions of TIMP-2, which are carried out by its free form, also depend on the amounts of MMP-14 and MMP-2 available (40, 41). Thus the net balance between MMP-14, MMP-2, and TIMP-2 may determine the MMP-dependent and -independent biological activities of TIMP-2. Here, no correlation was observed between the simultaneous reduction of both TIMP-2 and MMP-2 in hypoxic monocytes and their reduced, yet high, mobility, and exogenous anti-TIMP-2 or recombinant TIMP-2 had no additional effect, suggesting that reduced TIMP-2 secretion from monocytes in hypoxia may have paracrine, rather than autocrine, effects on cell migration. Thus we conclude that mechanisms other than the MMP-2-to-TIMP-2 ratio are at work. Such mechanisms that attenuate migration could involve other MMPs (32), chemokines (21), MAPK phosphatase 1 (MKP-1) (12), macrophage migration inhibitory factor (MIF) (21), or angiostatin (28).
Although TIMP-2 has no influence on monocyte mobility, it seems to influence endothelial cells profoundly. In our study, addition of anti-TIMP-2 to the balanced MMP-2/TIMP-2 environment of normoxic endothelial cell cultures enhanced migration/proliferation and tube formation, while it did not affect these already augmented angiogenic parameters in hypoxic cultures (Fig. 6, D and F), which contain reduced levels of TIMP-2 (Fig. 6A). Addition of recombinant TIMP-2 restored TIMP-2 levels and brought the hypoxia-induced elevation of migration/proliferation and tube formation back to normoxic values. In this respect, the mechanisms and biological importance of the hypoxia-induced changes in the in vitro tube morphology, which were not affected by TIMP-2, are not clear to us.
Our findings showing that reduced levels of TIMP-2 enhance migration/proliferation of endothelial cells are in agreement with the consensus as to the inhibitory role of homeostatic TIMP-2 in angiogenesis (39, 41). Our in vivo results, which demonstrate a proangiogenic effect of TIMP-2, would seem contradictory to the in vitro results. However, the amounts of TIMP-2 mixed with the Matrigel were low and similar to those found in the hypoxic cultures, actually supporting our in vitro results that demonstrate that TIMP-2 is proangiogenic in a limited range. Furthermore, our Matrigel plugs did not include additional proangiogenic triggers (e.g., FGF-2, VEGF) used in previous in vivo research that demonstrated an antiangiogenic effect of high concentrations of TIMP-2 (16-fold higher than in our experiment) (36). In our study we cannot determine whether TIMP-2 influences are MMP-2 dependent or -independent. In addition, since the in vivo environment includes different cell types, TIMP-2 may have direct influences on endothelial cells and/or other cells, such as smooth muscle cells, which are recruited to newly formed blood vessels and stabilize them. However, these results further emphasize the context- and cell-dependent effects of TIMP-2 and its role during specific steps of angiogenesis.
In conclusion, we have shown that in the hypoxic microenvironment transcriptionally reduced levels of TIMP-2 are proangiogenic but do not support monocyte migration. Therefore, the proangiogenic effects of hypoxia are not limited to transcriptional induction of proangiogenic molecules via the involvement of the HIF family of transcription factors (the TIMP-2 promoter has no HIF binding sites), but also extend to transcriptional suppression.
This work was supported by the Rappaport Family Institute for Research in the Medical Sciences and by Research Grant 5343 from the Chief Scientist of the Israeli Ministry of Health.
No conflicts of interest, financial or otherwise, are declared by the author(s).
- Copyright © 2011 the American Physiological Society