In the brain, angiotensinogen (AGT) is primarily expressed in astrocytes; brain ANG II derived from locally produced AGT has been shown to influence blood pressure. To better understand the molecular basis of AGT expression in the brain, we identified a human astrocytoma cell line, CCF-STTG1, that expresses endogenous AGT mRNA and produces AGT protein. Studies examining CCF-STTG1 cell AGT after N- and O-glycosidase suggest that AGT may not be posttranslationally modified by glycosylation in these cells as it is in plasma. Small amounts of AGT (5% of HepG2) were detected in the culture medium, suggesting a low rate of AGT secretion. Immunocytochemical examination of AGT in CCF-STTG1 cells revealed mainly nuclear localization. Although this has not been previously reported, it is consistent with nuclear localization of other serpin family members. To examine this further, we generated a fusion protein consisting of green fluorescent protein (GFP) and human AGT and examined subcellular localization by confocal microscopy after confirming expression of the fusion protein by Western blot. In CCF-STTG1 cells, a control GFP construct lacking AGT was mainly localized in the cytoplasm, whereas the GFP-AGT fusion protein was primarily localized in the nucleus. To map the location of a potential nuclear localization signal, overlapping 500-bp fragments of human AGT cDNA were fused in frame downstream of GFP. Although four of the fusion proteins exhibited either perinuclear or cytoplasmic localization, one fusion protein encoding the COOH terminus of AGT was localized in the nucleus. Importantly, nuclear localization of human AGT was confirmed in primary cultures of glial cells isolated from transgenic mice expressing the human AGT under the control of its own endogenous promoter. Our results suggest that AGT may have a novel intracellular role in the brain apart from its predicted endocrine function.
- renin-angiotensin system
- transgenic animal
- central nervous system
the role of the renin-angiotensin system (RAS) in blood pressure regulation has long been established. The vasoactive component angiotensin II (ANG II) is produced from serial cleavage of angiotensinogen (AGT) by renin, and ANG I by angiotensin-converting enzyme (ACE). The source of circulating AGT is the liver, and increased circulating AGT has been correlated with increased blood pressure in experimental models and in humans (13, 17). Tissue RASs are also important in blood pressure regulation and have been implicated in hypertension (reviewed in Refs. 22, 30). For example, increased activity of the RAS in the brain has been implicated as a mechanism causing or maintaining elevated arterial pressure in both genetic and experimental models of hypertension (19, 31, 47). The importance of the brain RAS in hypertension is supported by numerous studies that showed reduction of blood pressure in spontaneously hypertensive rats after intracerebroventricular injection of RAS inhibitors, antagonists, and antisense oligonucleotides (7, 11, 29, 42).
In the adult, AGT is synthesized in most regions of the brain, with its most abundant production in the medulla and hypothalamus, areas with demonstrated control of cardiovascular function (40). The most abundant sources of AGT in the brain are astrocytes, which have been reported to constitutively secrete AGT into the extracellular and cerebrospinal fluids (9, 35, 40). In addition to the AGT-positive astrocytes that are found throughout the brain, AGT is also present in a limited population of neurons in regions of the brain dedicated to cardiovascular control (45). Although the relative importance of glial and neuronal AGT remains poorly defined, the importance of glial cells as a source of AGT participating in arterial pressure regulation is becoming better appreciated. Transgenic mice exhibiting glial-specific overexpression of human AGT (hAGT) and human renin have a moderate increase in blood pressure (25). Transgenic rats carrying a glial-targeted antisense construct to AGT exhibited a 90% decrease in brain AGT levels and a significant reduction in blood pressure (37). Astrocyte AGT has also been reported as important for the maintenance of the blood-brain barrier in response to cold-induced brain injury (15).
Despite the growing awareness for the importance of glial AGT, no in vitro models of glial cells endogenously producing AGT were previously identified. This stands in contrast to HepG2, cells that are a well-characterized model of hepatic AGT synthesis. Indeed, much of our understanding of AGT expression and regulation was learned with HepG2 cells (10, 41). Given the importance of glial AGT in arterial pressure regulation, the lack of a glial cell culture model expressing AGT has become a serious limitation. Herein, we report the identification of a human astrocytoma cell culture model that endogenously expresses AGT and that retains robust expression of glial fibrillary acidic protein (GFAP), a marker of glial cells. Whereas low levels of AGT are secreted from these cells, the bulk of the AGT remains intracellular and is localized in the nucleus. AGT-enhanced green fluorescent protein (eGFP) fusion proteins were used to identify a region of the AGT protein targeting the astrocyte nucleus.
MATERIALS AND METHODS
All cell lines were grown at 37°C in 5% CO2 to 90% confluence. HepG2 cells were propagated in Eagle's MEM supplemented with 10% FBS, 1.0 mM sodium pyruvate, and 0.1 mM nonessential amino acids. CCF-STTG1 cells were cultured in RPMI 1640 supplemented with 10% FBS, HEPES, penicillin (100 U/ml), and streptomycin (100 mg/ml). SVG p12 cells were cultured in Eagle's MEM containing 10% FBS, 2% penicillin and streptomycin, and 0.002 μg/ml fungizone. Transfection of HepG2 and CCF-STTG1 cells was performed with Lipotaxi (Stratagene) according to the protocol supplied by the manufacturer with slight modifications. For transfection, 115 μl of serum-free medium were mixed with 35 μl of Lipotaxi and 7 μg of DNA. For the eGFP-C2 control vector, only 2 μg of DNA were used. After incubation at room temperature for 30 min, 500 μl of serum-free medium were added, and the mixture was added to cells plated on coverslips in a six-well dish. After 5 h, 700 μl of medium containing 20% serum were added to the cells overnight before replacement with normal growth medium for another 24 h.
Primary cells were obtained from hAGT transgenic mouse brains and cultured as previously described (14). Cultured cells were plated on coverslips coated with poly-l-lysine (Sigma) containing prewarmed medium and serum. After overnight recovery, one-half of the medium was removed, replaced with fresh medium, and grown overnight. Primary cultures were grown for 14 days, and the medium was replaced every 3 days.
RNA (20 μg) was DNase treated, and RT-PCR was performed. For PCR amplification, 10% of the RT reaction was used with the following synthetic oligonucleotides: ACE, 5′-ACGTGCTGGCGCTCTCGGTCTC-3′ and 5′-GCTATGAGGGCCCACTGCACCA-3′ (402 bp); angiotensin AT1 receptor, 5′-CAGAAAGTCGGCACCAGGT-3′ and 5′-TCTTTTGAATTTAGCACTGGC-3′ (282 bp). The hAGT cDNA was amplified from CCF-STTG1 total RNA by RT-PCR as stated above with the following primers: forward primer 5′-ACTAAGCTTAGACCGGGTGTACATACAC-3′ and reverse primer 5′-CGGGTACCTCATGCTGTGCTCAGCGG-3′ amplifying a 1.3-kb segment of DNA. The primer contained upstream KpnI and downstream HindIII restriction sites. The hAGT PCR product, lacking the 33-amino acid signal peptide, was cloned in frame downstream of eGFP, generating the fusion protein eGFP-hAGT. For deletion analysis, ∼500-bp overlapping fragments of hAGT were amplified and cloned in frame downstream of eGFP with the following primer sets: fragment A (nucleotides 40–516): 5′-TATAAGCTTACGGAAACGAGCACCCCAGTCTGAG-3′ and 5′-TATGGTACCCTACTGTAGCCTGTCAGCTGTGTGGTC-3′; fragment B (nucleotides 517–1,026): 5′-TATAAGCTTAGCAATCCTGGGTGTTCCTTGGAAG-3′ and 5′-TATGGTACCCTAGAAGTTGTCCTGGATGTCACTCCA-3′; fragment C (nucleotides 1,027–1,497): 5′-TATAAGCTTATCGGTGACTCAAGTGCCCTTCACT-3′ and 5′-CGGGTACCTCATGCTGTGCTCAGCGG-3′; fragment D (nucleotides 268–735): 5′-TATAAGCTTATCCCCTGTGGATGAAAAGGCCCTA-3′ and 5′-TATGGTACCCTAAGGGGTATAGAGAGCCAGGCCCTG-3′; and fragment E (nucleotides 817–1,257): 5′-TATAAGCTTAGGATGGAAGACTGGCTGCTCCCTG-3′ and 5′-TATGGTACCCTAGTGCAGAATGGCGGGCAGCTCAGC-3′. These generated fusion polypeptides ranged from 147 to 170 amino acids in length.
Cells were washed and lysed using TRI-Reagent (Molecular Research Center) to isolate total cellular RNA, according to the manufacturer's protocol. A [α-32P]UTP antisense RNA probe generated by in vitro transcription from an hAGT cDNA sequence spanning nucleotides 302–931 was used as probe for Northern and RNase protection assay analyses as described previously (46). The RNase protection assay III kit (Ambion) was used according to the manufacturer's protocol with 28S serving as a loading control. Protected fragments were 350 (hAGT) and 150 (28S) nucleotides, respectively.
Total cellular protein was obtained after washing and lysing in a buffer containing 50 mM Tris (pH 8), 10 mM EDTA, 10 μg/ml leupeptin, and 1% Triton X-100. After centrifugation, cell lysates were collected and stored at −20°C. To remove N-linked glycosylation, 20 μg of total cellular lysate were incubated with Peptide: N-glycosidase F (New England Biolabs) according to the manufacturer's instructions. To remove both N- and O-linked glycosylation, PROzyme (Glyko) was used according to the manufacturer's protocol. Briefly, 20 μg of protein were denatured, and deglycosylation was performed with N-glycanase, sialidase A, and O-glycanase enzymes, all provided in the kit. To ensure removal of complex Core 2 O-linked carbohydrates, Pro-Link extender enzyme [β(1–4) galactosidase and glucosaminidase] was added. Protein was separated and analyzed with primary polyclonal antibodies to hAGT [1:10,000 dilution, a generous gift provided by Dr Duane Tewksbury (Marshfield Medical Research Foundation, Marshfield, WI)] or eGFP (1:1,000 dilution, Chemicon). Anti-rabbit horseradish peroxidase-conjugated secondary antibodies (1:1,000 dilution, Amersham Pharmacia Biotech) were used in addition to enhanced chemiluminescence (Amersham Pharmacia Biotech) for visualization.
Cells were grown overnight on coverslips in six-well dishes. Samples were fixed for 15 min with 4% paraformaldehyde in PBS containing 0.1% Triton X-100 and incubated for 2 h in a blocking solution (PBS, 5% donkey serum, 1% BSA, 0.5% sodium azide). For primary staining, we used anti-rabbit polyclonal antibodies to hAGT (room temperature, 1:100 dilution) and mouse monoclonal GFAP (1:100 dilution, Chemicon International) or mouse p115 (1:100, Transduction Laboratories) antibodies in blocking solution. Secondary incubations were performed with rhodamine-conjugated anti-rabbit IgG (1:100, Chemicon International) and fluorescein-conjugated anti-mouse IgG (1:100 Chemicon International) in blocking serum. As a nuclear control, 4′,6-diamidino-2-phenylindole (DAPI; Sigma) was used. For preabsorption experiments, antibodies were incubated for 4 h with 500-fold molar excess of purified hAGT (room temperature, Scripps Laboratories) before proceeding as described above.
The absence of an hAGT-expressing glial cell line from the brain prompted us to screen human astrocytoma cell lines available from ATCC. One of these, CCF-STTG1, expressed both isoforms of hAGT mRNA derived from differential polyadenylation (Fig. 1A). No expression of hAGT was detected in another glial cell line, SVG p12, or in renin-expressing cells derived from the lung (Calu-6). Expression of hAGT was confirmed by RNase protection, which revealed significantly lower expression than in HepG2 cells, a well-defined model of hAGT-expressing cells derived from the liver (Fig. 1B). Because all RAS components have been identified in the brain, we assayed CCF-STTG1 cells for expression of other RAS components by RT-PCR. Expressions of ACE and the AT1 receptor were evident in CCF-STTG1 cells (Fig. 2). This defines a difference with HepG2 cells, which do not express ACE. No expression of renin or AT2 receptor mRNA was detected (data not shown).
We next assayed for production of hAGT protein by Western blot. CCF-STTG1 cells produce two predominate isoforms of hAGT protein (arrows in Fig. 3A), although much lower amounts of a higher molecular weight form can also be detected (asterisk in Fig. 3B). Consistent with previous reports, HepG2 cells express multiple isoforms of hAGT due to differential glycosylation. This is also consistent with studies showing that AGT contains four asparagine-linked glycosylation sites, and as many as five different isoforms are found in the systemic circulation (10). We therefore examined the extent of glycosylation of hAGT protein in CCF-STTG1 cells. As expected, the multiple isoforms of hAGT in HepG2 cells were reduced to primarily one protein band after N-glycosidase or combined N- and O-glycosidase treatment (Fig. 3B). On the contrary, there was no reduction in size of the major isoforms of hAGT from CCF-STTG1 cells after glycosidase treatment. We consistently observed the reduction in size of one minor higher molecular weight isoform of hAGT from CCF-STTG1 cells (asterisk in Fig. 3B). Bovine fetuin was used as an experimental control to verify the fidelity of the glycosidase treatment.
Inhibition of glycosylation has been reported to inhibit AGT secretion from astrocyte cultures (38). We therefore measured the secretion of hAGT by collecting samples of medium and assaying the conversion of hAGT to ANG I by radioimmunoassay in the presence or absence of excess purified human renin (40). Small amounts of AGT (equivalent to 4.5 ng·ml−1·24 h−1 ANG I) were secreted from CCF-STTG1 cells, compared with nearly 400 ng·ml−1·24 h−1 released from HepG2 cells. No ANG I was detected when the media was incubated without human renin. These results suggest that CCF-STTG1 cells have the capacity to secrete small amounts of hAGT.
AGT is constitutively released from the liver into the systemic circulation, and the high levels of AGT secreted from HepG2 cells is consistent with this. Given the small amounts of AGT released from CCF-STTG1 cells, we asked whether the protein was being retained in the cell. Indeed, we previously reported abundant intracellular hAGT in glial cells expressing hAGT under the control of the glial-specific GFAP promoter in transgenic mice (24). We therefore used double labeling and confocal microscopy to identify the subcellular localization of hAGT in CCF-STTG1 and HepG2 cells (Fig. 4). Although AGT exhibited perinuclear localization in HepG2 cells, we made the surprising finding that hAGT was localized in the nucleus of CCF-STTG1 cells. As expected, there was no staining of HepG2 cells with GFAP. Confocal microscopy revealed that there was essentially no overlap between the staining of the cytoplasmic marker of glial cells, GFAP, and hAGT in CCF-STTG1 cells. Nuclear staining was confirmed in cells labeled with DAPI. The absence of perinuclear staining in CCF-STTG1 cells was confirmed by staining with hAGT and p115, a cis-to-medial Golgi apparatus marker (data not shown). As an important control, we wanted to experimentally verify the specificity of the hAGT antibody. Consequently, we preabsorbed the antisera with excess purified recombinant hAGT. The specificity of the antisera was demonstrated by the loss of overlapping fluorescence between hAGT and DAPI (from purple to blue) in the nucleus of CCF-STTG1 cells.
We next tested whether a fusion protein consisting of eGFP and hAGT also transited to the nucleus. We generated a COOH-terminal fusion between eGFP and the mature form of hAGT protein lacking the 33-amino acid signal peptide. COOH-terminal fusions have been used extensively to assay for the presence of nuclear localization signals (NLSs) in many proteins, including the related serpin proteinase inhibitor 9 (2). CCF-STTG1 and HepG2 cells were either left untransfected or transfected with 1) the control vector containing eGFP but lacking hAGT (eGFP-C2) or 2) the eGFP-hAGT fusion protein vector. First, we assayed total cellular lysates by Western blot analysis to verify production of the fusion proteins (Fig. 5). Only the fusion or control proteins were detected with antisera directed against eGFP (Fig. 5A), whereas both the fusion proteins and endogenous hAGT were detected with hAGT antisera (Fig. 5B). To visualize subcellular localization, immunohistochemistry and confocal microscopy were performed (Fig. 6). For these experiments, antibodies to eGFP (green) and GFAP (red) were used. The control vector eGFP-C2 had a diffuse pattern of cellular localization in CCF-STTG1 and HepG2 cells (Fig. 6A). In HepG2 cells, eGFP-hAGT was mainly cytoplasmic. However, in CCF-STTG1 cells, the eGFP-hAGT fusion protein was primarily localized in the nucleus and did not exhibit an overlapping pattern of fluorescence with GFAP (Fig. 6B).
Sequence analysis of AGT with ProSite (http://us.expasy.org) did not reveal any obvious NLS. We therefore constructed five overlapping hAGT deletion fragments each ∼500 bp and encoding polypeptides 147 to 170 amino acids in length and then fused them downstream of eGFP (Fig. 7A). CCF-STTG1 cells were transfected, and immunohistochemistry and confocal microscopy were performed with antibodies to eGFP and GFAP (Fig. 7B). In these experiments, the A, B, and E fusion proteins displayed diffuse cytoplasmic localization, whereas the D fusion protein sometimes exhibited perinuclear staining. Interestingly, however, the C fusion protein appeared primarily in the nucleus. This suggests that the COOH-terminal end of the hAGT protein may contain a signal that is recognized for nuclear targeting of hAGT in CCF-STTG1 cells.
Finally, to determine whether this finding is physiologically relevant, we analyzed primary glial cell cultures derived from hAGT transgenic mice. These mice express a 13-kb genomic transgene containing the hAGT gene under control of its own endogenous promoter, which we previously demonstrated exhibits appropriate cell-specific expression in the brain (45, 46). The cultures were prepared from the brains of 4-mo-old adult hAGT transgenic mice and contained both neurons and glial cells. Immunohistochemistry was performed on cultures grown for 3, 9, and 14 days. After 9 days of growth, hAGT localization was detectable in GFAP-positive cells and appeared nuclear (data not shown). After 14 days, the glial cells were very distinct and hAGT localization was observed in both the cytoplasm and nucleus (Fig. 8). The nuclear staining was often quite prominent (arrows in Fig. 8), thus confirming our observations in CCF-STTG1 cells.
There is a growing awareness for the importance of tissue RAS and its involvement in the local production and action of ANG II in a number of organs, including the kidney and brain (reviewed in Ref. 22). Local production of ANG II in these tissues plays an important role in the regulation of arterial pressure, sodium reabsorption, sympathetic outflow, vasopressin release, and drinking behavior. Studies in transgenic mice, where ANG II production has been specifically targeted to kidney or brain, have shown an increase in arterial pressure and, depending on the model, increased drinking and salt preference as well (6, 25). At the cellular level, the generation of ANG II in tissues occurs via the same mechanism as in the systemic circulation where secreted AGT is cleaved by secreted renin in the extracellular space. Our findings reported herein coupled with other observations suggest that there is 1) an additional pathway for the generation and action of ANG II intracellularly, particularly in the brain, and 2) a second less appreciated intracellular function for AGT that may be independent of its role as a substrate for ANG II. That nuclear AGT was confirmed in primary glial cells derived from transgenic mice expressing an appropriately regularly hAGT transgene suggests that this is a physiologically relevant finding and that its role as an intracellular and nuclear protein needs to be considered.
AGT as an intracrine.
Emerging evidence of intracellular ANG II and AT1 receptors and the recent finding of an intracellular unsecreted form of renin have supported the possibility of an intracrine RAS (33, 34, 39). To be classified as an intracrine, a factor must first have the potential of being found in the extracellular space with the ability to affect biological activity in the presence of a target cell and second must also be found inside its cell of synthesis in association with one or more intracellular organelles not of the secretory or degratory pathway (34). Multiple components of RAS, including AGT, can be considered intracrines, and evidence supporting intracellular synthesis and action of ANG II has been reported. For example, intracellular ANG II immunoreactivity was reported to be associated with chromatin of hepatocytes, adrenal cells, and cerebellar neurons, and binding of ANG II to nuclear angiotensin receptors results in increased transcription of renin, AGT, and platelet-derived growth factor (4, 8). Although its precise role remains unclear, it is possible that nuclear localization of AGT in astrocytes provides a substrate for nuclear ANG II associated with chromatin or associated with nuclear AT1 receptors.
Of course, to invoke a logical argument for the intracellular (or even nuclear) production of ANG II from intracrine AGT, we first need to consider the source of renin. First, renin and prorenin can be internalized through a number of mechanisms, including a recently identified high-affinity receptor (12, 27, 28, 36). Second, we and others have reported the existence and function of an alternative form of renin in several tissues, including the brain (5, 23, 39). This alternative form of renin lacks the signal peptide and therefore remains inside the cell. Transfection of a mutated form of AGT lacking a signal peptide into hepatoma cells expressing the unsecreted form of renin was reported to upregulate PDGF and enhance cellular proliferation via a mechanism sensitive to losartan or renin antisense (5a). Third, using a dual reporter transgenic mouse approach, our group (20) recently reported that renin and AGT can be coexpressed in some cells or in adjacent cells in the brain. What we do not know is whether the cells coexpressing renin and AGT express the intracellular or secreted form of renin or whether renin gains access to the nucleus. Recent unpublished studies from our laboratory (Lavoie and Sigmund, unpublished observations) indicate that the unsecreted form of renin can cleave glial AGT to generate ANG II, which participates in the regulation of arterial pressure.
CCF-STTG1 cells themselves do not express renin. Although this may be a reflection of the transformed state of the cell line, more likely it indicates that renin is only expressed in a small proportion of AGT-expressing glial cells. Indeed, the patchy regional expression of glial renin in the brain is supported by studies that used transgenic mice expressing a tightly regulated renin transgene as well as renin promoter-reporter genes (21, 26).
AGT as serpin.
AGT is a member of the serine protease inhibitor family (serpin) because it is structurally similar to inhibitory serpins. Most serpins act by mimicking the structure of the substrate for their targeted protease. After binding, the protease cleaves the serpin, causing a change in its structure that shunts the protease to a site on the serpin where it can be inactivated or destroyed. Conventionally, AGT is considered a noninhibitory serpin because it does not specifically inhibit any known serine proteases or undergo the classic stressed-relaxed conformational change typical of inhibitory serpins. However, it is worth noting that des[ANG I]AGT, the cleavage product of AGT by renin, has been reported to inhibit the action of renin, an aspartyl protease, suggesting that there may be some related inhibitory activity (1, 32).
Noninhibitory serpins such as ovalbumin exhibit complex cellular distribution patterns that are nuclear, cytoplasmic, or both. Many intracellular serpins lack a classical NLS, as normally found in proteins such as SV40 T antigen (reviewed in Ref. 43). It was hypothesized that a conformational NLS comprising a number of noncontiguous residues functions in human proteinase inhibitor 9 (an ovalbumin-like serpin) (2). Based on the behavior of our eGFP-AGT fusion proteins, we propose that hAGT may be behaving similarly in that a sequence of amino acids is recognized by its three-dimensional conformation to target it to the nucleus. Indeed, alignment of the COOH terminus of AGT, containing the putative NLS, with other serpins that exhibit nuclear or nucleocytoplasmic localization reveals a number of invariant and highly conserved residues (Fig. 9) (2, 16). Mutagenesis experiments will be required to determine whether any of these form a functional NLS.
Most serpins (such as angiogenin) have angiogenic properties, although a few (such as maspin) have antiangiogenic properties. Celerier et al. (3) reported that full-length AGT and des[ANG I]AGT had antiangiogenic properties in the chorioallantoic membrane angiogenic assay, inhibited proliferation and migration of cultured endothelial cells, and attenuated capillary formation in the matrigel. Supporting the hypothesis that the antiangiogenic effects are independent of the classical RAS pathway was the absence of ANG I and ANG II in the assays and the retention of antiangiogenic activity of a reactive center-loop-cleaved AGT derivative. Consequently, our typical view of AGT only as the source for generating ANG II may not be complete; it may also have other important functions. The localization of AGT (but not renin) in the nucleus of glial cells may be consistent with a function for a native protein that has yet to be experimentally determined.
Interestingly, some AGT-containing glial cells in the brain surround capillaries. Kakinuma et al. (15) examined the importance of glial AGT in response to cold injury using AGT-deficient mice and demonstrated abnormal vascular permeability in the brain that was associated with defects in what would normally be AGT-containing glial cells. Interestingly, this abnormality was not observed in renin-deficient mice, which should be equally defective in the generation of ANG II (44). These results suggest that glial AGT may be an important factor regulating the stability of the blood-brain barrier.
This work was supported by National Heart, Lung, and Blood Institute Grants HL-58048, HL-61446, and HL-55006. We gratefully acknowledge the generous research support of the Roy J. Carver Trust.
Confocal microscopy was performed at the University of Iowa Central Microscopy Facility.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2005 the American Physiological Society