Acta Neuropathol (1998) 96 : 287–296
© Springer-Verlag 1998
R E G U L A R PA P E R
R. Veerhuis · I. Janssen · J. J. M. Hoozemans · C. J. A. De Groot · C. E. Hack · P. Eikelenboom
Complement C1-inhibitor expression in Alzheimer’s disease
Received: 18 December 1997 / Revised, accepted: 11 March 1998
Abstract In situ and in vitro studies suggest that activation of locally produced complement factors may act as a mediator between amyloid deposits and neurodegenerative changes seen in Alzheimer’s disease (AD). C1-esterase inhibitor (C1-Inh), which regulates activation of C1 of the complement classical pathway, can be detected immunohistochemically in its inactivated form in activated astrocytes and dystrophic neurites in AD plaque areas. In this study, designed to investigate the cellular source of C1-Inh, C1-Inh was found to be secreted in a functionally active form by astrocytes cultured from postmortem human brain specimens as well as by neuroblastoma cell lines. Recombinant human interferon-γ (IFN-γ), which stimulates C1-Inh synthesis in various cell types, several-fold stimulated C1-Inh protein secretion by cultured human astrocytes derived from different regions of the central nervous system and by one (SK-N-SH) of two neuroblastoma cell lines (SK-N-SH and IMR-32) included in this study. In contrast to IFN-γ, other cytokines [interleukin (IL)-1β, IL-6 and tumor necrosis factor (TNF)-α] that can be found in brain areas affected by AD, did not stimulate C1-Inh se-
R. Veerhuis · P. Eikelenboom Department of Psychiatry, Research Institute Neurosciences Vrije Universiteit, Vrije Universiteit, Graduate School Neurosciences Amsterdam, Amsterdam, The Netherlands R. Veerhuis · I. Janssen · J. J. M. Hoozemans · C. J. A. De Groot Department of Pathology, Research Institute Neurosciences Vrije Universiteit, Vrije Universiteit, Graduate School Neurosciences Amsterdam, Amsterdam, The Netherlands C. E. Hack Department of Pathophysiology of Plasma Proteins, Central Laboratory for Blood Transfusion, Amsterdam, The Netherlands R. Veerhuis (Y) Department of Pathology, Academic Hospital Vrije Universiteit, P.O. Box 7057, 1007 MB Amsterdam, The Netherlands e-mail:
[email protected], Tel.: 31-20-444-4096, Fax: 31-20-444-2964
cretion by astrocytes or neuroblastomas in vitro. This inability to secrete C1-Inh is probably due to unresponsiveness at the transcriptional level, since C1-Inh secretion paralleled the expression of the 2.1-kb C1-Inh mRNA. In situ hybridization with a C1-Inh RNA antisense probe labeled neurons rather than astrocytes, suggesting a role for neurons as producers of complement regulatory proteins in vivo. Since IFN-γ is apparently lacking in the brain parenchyma, and amyloid plaque-associated cytokines (IL-1β, IL-6, TNF-α) do not stimulate C1-Inh expression in vitro, the nature of the stimulus responsible for neuronal C1-Inh expression in AD brains remains to be investigated. Key words Alzheimer’s disease · Astrocytes · Complement · C1-esterase inhibitor · RNA in situ hybridization
Introduction Complement activation products can be found associated with parenchymal, as well as with vascular amyloid deposits in Alzheimer’s disease (AD) brains [13, 15, 20, 27, 39]. The presence in neuritic plaques of microglia expressing complement iC3b receptors [35], together with the finding that amyloid β (Aβ) peptides, the major constituent of amyloid plaques, can bind C1q and activate the complement classical pathway in vitro [34], suggests a role of complement activation as a mediator between amyloid deposition and neurotoxicity [14]. Presumably all complement factors can be synthesized in the brain parenchyma [5]. Although complement synthesis in the central nervous system (CNS) is generally low, it is up-regulated in AD brains [25, 41] and in experimental brain lesions [31]. Therefore, especially under these conditions regulators of complement activation are needed to prevent unwarranted complement activation. C1-esterase inhibitor (C1-Inh), a heavily glycosylated serine protease inhibitor, inhibits complement at the very first stage. C1-Inh rapidly dissociates the activated C1 subcomponents C1r
288
and C1s from the activator-bound C1q, thus preventing the activation of C4 and C2 by C1s. Besides inhibition of complement activation, C1-Inh is also involved in the regulation of coagulation, fibrinolysis and of kinin-releasing systems [3]. Locally produced thrombin [1] and activation products of the coagulation system, such as activated coagulation factor XII that can be found in the brain [43], may cause consumption of C1-inh, eventually leading to disregulation of the complement system. Inactivated and complexed C1-Inh was reported to be present in neurites [42] and activated astrocytes [40] in the neocortex of AD cases. Whether the complexed and/or inactivated C1-Inh had been ingested, or was produced by these cells, was not clear. The very low levels of native C1-Inh detected by Western blotting of both AD and non-demented control brain homogenates are probably not plasma derived, because reverse transcriptase-polymerase chain reaction (RT-PCR) studies revealed that C1-Inh mRNA is expressed in these brain areas [40, 42]. Although these results are suggestive for local C1-Inh production, the cellular source of cerebral C1-Inh remained elusive, because of absent [40] or variable [42] immunostaining for native C1-Inh in AD and control brains. Astrocytoma cell lines [18], as well as neuroblastoma cell lines [19], can synthesize C1-Inh in vitro. In the present study we investigated whether, besides interferon-γ (IFN-γ) that is known to induce C1-Inh synthesis in a number of cell types [26], the AD plaque-associated cytokines interleukin-1β (IL-1β, IL-6 and tumor necrosis factor-α (TNF-α) can induce C1-Inh expression by astrocytoma and neuroblastoma cell line cultures. To correlate the findings to the situation in vivo, the expression of C1-Inh by primary human astrocyte cultures, which most closely resemble astrocytes in vivo, was determined both at the RNA and at the protein level. In addition, in situ hybridization with a C1-Inh-specific cRNA probe was performed to investigate whether neurons and astrocytes in AD brains also express C1-Inh mRNA. For this, tissue specimens from AD cases with activated glial cells associated with neuritic plaques, and from non-demented controls without amyloid plaques or with only diffuse type plaques were subjected to in situ hybridization.
Table 1 Summary of cases used in this study (PMD postmortem delay, LBV Lewy body variant, CTL + pl control case with many amyloid plaques) Case no. Sex/Age
Diagnosis
PMD (h) Areas
003 192 276 282 21
!/87 ?/59 !/76 !/87 !/51
AD AD AD/LBV AD AD/DOWN
4:25 6:00 4:55 5:00 8 : 00
300 136 311
!/82 ?/72 ?/63
CTL + pl CTL CTL
6:30 9:00 2:45
Temporal cortex Temporal cortex Temporal cortex Temporal cortex Temporal cortex, Hippocampus Temporal cortex Temporal cortex Temporal cortex
PMD: post mortem delay time; LBV: Lewy body variant; CTL + pl: control case with many amyloid plaques
Polyclonal rabbit anti-glial fibrillary acidic protein (GFAP) was from DAKO (Glostrup, Denmark). Monoclonal antibodies specific for CD11c (LeuM5) and CD14 (LeuM3) were from Becton Dickinson (Erembodegem, Aalst, Belgium), for vimentin (clone V9) from Boehringer Mannheim, and for glutamine synthethase from Affinity (Nottingham, UK). Recombinant human IFN-γ (rhIFN-γ) was a gift from Dr. P. H. Van der Meide (BPRC, Rijswijk), Recombinant human IL-1β (Genzyme), IL-6 (CLB, Amsterdam) and TNF-α (PeproTech, Canton, Mass.) were obtained as indicated. Tissue specimens Human brain specimens were obtained at autopsy with a short postmortem delay through the Netherlands Brain Bank (coordinator Dr. R. Ravid). For in situ hybridization, temporal cortex specimens of four AD cases, three controls, and hippocampal and temporal cortex sections of a case of Down’s syndrome with AD were snap frozen and stored in liquid nitrogen until use. Cases used in this study are listed in Table 1. Clinical diagnosis of AD was neuropathologically confirmed on formalin-fixed, paraffin embedded tissue from different sites. Many senile plaques, neurofibrillary tangles and neuropil threads were shown to be present in the hippocampus and in the frontal, temporal and parietal neocortex using Bodian and Congo red staining. Non-demented controls were without clinical signs of AD. Temporal cortex specimens of two of three control cases had many diffuse type plaques that were tau negative.
Cells
Materials and methods Trypsin, gentamycin, streptomycin, penicillin, 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) and poly-L-lysine (PLL), were purchased from Sigma (St. Louis, Mo.), bovine pancreatic DNase I from Boehringer Mannheim (Mannheim, Germany). Dulbecco’s modified Eagle’s medium (DMEM), nutrient mixture HAM’s F-10 (HAM-F10) were obtained from GIBCO Life Technologies (Breda, The Netherlands), fetal calf serum (FCS) from ICN Biomedicals (Amsterdam, The Netherlands), and sterile pyrogen-free water from Baxter B.V. (Utrecht, The Netherlands). Microtiter plates (96 wells) for enzyme-linked immunosorbent assays (ELISA) and 24-well plates for cell culturing were obtained from Nunc (Roskilde, Denmark), 80-cm2 culture flasks from Greiner (Alphen a/d Rijn, The Netherlands). RNAzol B from Cinna/ Biotecx was purchased from Campro Scientific (Veenendaal, The Netherlands).
Primary astrocyte cultures from either a non-affected area of spinal cord of a multiple sclerosis case (male/age 61 years), or from the prefrontal cortex (female/age 85 years) or cerebral white matter (female/age 51 years) of non-demented controls, were established and characterized as described [10]. Briefly, brain specimens from which visible blood vessels are removed, were collected in DMEM/ HAM-F10 (1:1) containing gentamycin (50 µg/ml) and minced (approximately ± 2 mm2 cubes). Subsequent digestion (for 20 min at 37 °C) in Hanks’ balanced salt solution (HBSS) containing trypsin (0.025 mg/ml) and bovine pancreatic DNase I (0.1 mg/ml) resulted in cell suspensions that were washed with DMEM/HAMF10 (1:1), supplemented with 2 mM L-glutamine, 10% FCS, penicillin (100 IU/ml) and streptomycin (50 µg/ml) (DMEM/HAM-F10; 10% FCS), after which the cells were allowed to adhere to PLLcoated 80-cm2 culture flasks and were grown (at 37 °C; 5% CO2) as monolayer cultures in DMEM/HAM-F10; 10% FCS. More than 95% of the human astrocyte cultures from several passages expressed the astrocyte-specific markers GFAP and gluta-
289 mine synthetase, and also vimentin. No staining for LeuM5 (CD11c) and LeuM3 (CD14) was seen, indicating that contamination with macrophage/microglia was negligible. Human astrocytoma clone D384 [4] and astroglioma cell line U251 [32] were kindly provided by Drs. Balmforth (Department of Cardiovascular Studies, University of Leeds, Leeds, UK) and Darling (Institute of Neurology London, UK), respectively, through Dr. C. H. Langeveld (Department of Experimental Neurology, Vrije Universiteit, Amsterdam, The Netherlands). Neuroblastoma cell lines SK-N-SH (ATCC; HTB-11) and IMR-32 (ATCC; CCL127), as well as the astroglioma cell line U373 (ATCC; HTB-17) were obtained from American Type Culture Collection (ATCC, Rockville, Md.). Hepatoma cell lines HepG2 and Hep3B (ATCC) and primary cultures of human umbilical cord endothelial cells (HUVEC) as well as immortalized HUVEC (kindly provided by Dr. C. M. Schweitzer, Hematology Department, Vrije Universiteit and Dr. Van Nostrand, Stoney Brook, N.Y., respectively) were used as a positive control for C1-Inh production. Stimulation experiments At passage 3–5, three primary astrocyte cultures derived from different CNS areas (see above) were trypsinized and transferred from 80-cm2 flasks to 24-well plates at 2 × 104 cells/well in DMEM/ HAM-F10; 10% FCS. Likewise, astrocytoma or neuroblastoma cell lines were transferred to 24-well plates. After settling for 18 h, the medium was replaced by the same medium containing 0.1% FCS to reduce proliferation. Different doses of rhIFN-γ, IL-1β, IL-6 or TNFα were added (T = 0) to yield final concentrations of 4–1000 U/ml. At 24, 48 and 72 h after the start of the stimulation, supernatants were collected and, after centrifugation to remove cells, aliquoted and stored at –20 °C until assayed for the presence of C1-Inh. Cells remaining in the wells were either immunostained for the expression of complement factors and phenotypic characterization, lysed to isolate RNA for dot-blot analysis, or subjected to the MTT assay [2] to assess possible effects of cytokine stimulation on cell viability and proliferation. Formazan generated by the mitochondrial-dependent reduction of MTT by viable cells was solubilized in dimethylsulfoxide and measured in a microculture plate reader at 540 nm (Titertek; Flow labs). C1-Inh ELISA To determine the amount of functionally active C1-Inh in culture supernatants, 96-well microtiter plates (Nunc maxisorb) were coated (2 µg/ml in carbonate buffer pH 9.6) with monoclonal antibody RII, which recognizes native as well as complexed or inactivated C1-Inh [29]. RII-bound C1-Inh from culture supernatants was subsequently allowed to capture biotinylated activated C1s [11]. All dilutions were made in phosphate-buffered saline (PBS) containing 2% milk. Bound antibodies were detected with streptavidin polyHRP (CLB; Amsterdam) and visualized with TMB as a substrate. Colour development was stopped with 2 N H2SO4 and read at 450 nm with a Titertek Multiscan. Results were related to a calibration curve prepared for each ELISA plate by serial diluting fresh pooled plasma containing 275 µg native C1-Inh/ml, and expressed as ng C1-Inh/ml. To determine which percentage of secreted C1-Inh was functionally active, another ELISA was performed in which RII-bound C1-Inh was detected with a biotin-conjugated polyclonal anti-C1-Inh (total C1-Inh). As controls in these assays, either pooled fresh plasma (primarily containing native C1-Inh) or aged (37 °C for 7 days) serum (containing inactivated C1-Inh only) were used. Identical levels of C1-Inh were detected in culture supernatants with both ELISA systems, indicating that all C1-inh secreted by astrocytes, neuroblastomas or hepatoma cell lines was functionally active (data not shown). Therefore, only data from the assay measuring functionally active C1-Inh are shown.
Northern blots and spot blots For Northern blotting, astrocytes were cultured (1.3 × 106 cells) at subconfluency in 80-cm2 cell culture flasks, containing 10 ml DMEM/HAM-F10 with 0.1% FCS for 72 h. Supernatants from stimulated (1000 U/ml rhIFN-γ) and control cultures were harvested to assess C1-Inh secretion and RNA was isolated by guanidine isothiocyanate and phenol extraction (RNAzol B; Cinna/ Biotecx). Total RNA (15 µg/lane) was separated on 1% agarose gels containing 6% formaldehyde (40 mA; 18 h). RNA size markers (0.24–9.5 kb; Gibco BRL) were simultaneously run and stained with methylene blue. Size-separated RNA was capillary blotted (10 × SSC) onto nylon membrane (Qiabrane; Qiaex) and hybridized with in 0.5 M phosphate buffer and 7% SDS, pH 7.2, overnight at 65° C. After washing twice in 3 × SSC/0.5% SDS and twice in 0.1 × SSC/0.5% SDS, hybridization signal was visualized by phosphoimaging (Molecular Dynamics). To quantitate hybridization of RNA derived from stimulated astrocytes and neuroblastoma cells with the 32P-labeled C1-Inh probe, ethanol precipitated total RNA from each well of the 24-well plates was dissolved separately in 10 µl 1 mM EDTA and spotted onto nylon (Qiabrane; Qiaex). Spot blots were hybridized as described for the Northern blots. Hybridization signal as visualized by phosphoimaging was expressed as arbitrary units with use of ImageQuant analysis (Molecular Dynamics). Probes A 582-bp human C1-Inh cDNA sequence, corresponding to nucleotides 704–1286 of human C1-Inh cDNA, according to Bock et al. [7] was obtained after BglII and EcoRI digestion of plasmid pTZ 53, clone 53 [17], and ligated into pBluescript KS+ (Stratagene). For in vitro transcription, C1-Inh cDNA containing pBluescript KS+ was EcoRI or XbaI digested, size separated on 1% agarose gels and purified with Qiaex beads (Qiagen, Qiaex), according to the manufacturers instructions. Digoxigenin (DIG)-labeled RNA transcripts, both sense and antisense orientations, were generated [28] using DIG-11-UTP (Boehringer Mannheim) and either T3 or T7 RNA polymerases (Promega, Madison, Wis.). Templates were removed by RQ DNase (Promega) treatment (1 U/µg DNA; 30 min, 37 °C). Resulting cRNA was ethanol precipitated and dissolved in 200 µl deionized formamide containing yeast tRNA (50 µg/ml). Quantity of DIG-labeled probes was estimated by immunostaining for DIG of serial diluted aliquots of the probes spotted onto Qiabrane. Human elongation factor (hEF-1α) cDNA [8] was subcloned into pBluescript KS+ and used to check the integrity and yield of isolated mRNA. For use on Northern and dot blots, C1-Inh and hEF-1α cDNA probes were random primer labeled with [α-32P]dCTP (New England Nuclear). RNA in situ hybridization Cryostat sections (5 µm) were mounted on 3-amino-propyl-triethoxysilane (APES; Sigma, Mo.)-coated slides, air-dried and fixed in 4% paraformaldehyde. Hybridization mixtures (100 µl/cm2) consisted of 50% deionized formamide, 2 × SSC (0.3 M NaCl, 0.03 M sodium citrate), 10% (w/v) dextran sulfate and yeast tRNA (50 µg/ ml). After denaturation at 64 °C for 7 min, sections were hybridized for 3 h at 55 °C, after which coverslips were removed and slides were rinsed in 2 × SSC (30 min) and in 0.1 × SSC (2 × 30 min) at 55 °C and finally in PBS. RNA-RNA hybrids were detected immunohistochemically by successive incubations with blocking reagent (Boehringer Mannheim), peroxidase-conjugated sheep antiDIG (Boehringer Mannheim), biotinylated tyramine-HCl (Sigma) in 0.001% H2O2, and streptavidin-biotin-horseradish peroxidase complex (sABC; DAKO) [33]. Visualization with diaminobenzidine/NiCl2 and silver enhancement was performed as described [21]. Finally, sections were counterstained with hematoxylin and in some cases with Congo red to visualize the amyloid, dehydrated and mounted in Depex (BDH, Poole, UK). Negative controls included sense probes for each specimen and hybridization mix
290 without probe. The mRNA quality was checked by RNA in situ hybridization using β2-microglobulin or hEF-1α as control target mRNA.
Results Secretion of C1-Inh by astrocytes and neuroblastomas Under basal conditions, low C1-Inh levels ranging from undetectable to maximally 1 ng/ml secreted per 24 h were detected in the culture medium of primary astrocyte cultures (2 × 104 cell/well), as well as in that of U251, D384 and U373 astrocytoma cell lines and of two neuroblastoma cell lines (IMR-32 and SK-N-SH). Addition of rhIFN-γ to the human astrocyte cultures in a range of concentrations (15.6, 62.5, 250 and 1000 U/ml) clearly stimulated C1-Inh secretion in all three primary astrocyte cultures in a dose-dependent fashion (Fig. 1). C1-Inh accumulated with time, when assessed in supernatants harvested at either 24, 48 and 72 h after the start of the stimulation. The stimulatory effect of IFN-γ on C1-Inh production was not due to effects of IFN-γ on cell replication, since no differences in MTT values, reflecting mitochondrial activity, were observed between IFN-γ stimulated and control cultures (data not shown). Results comparable to those with primary astrocyte cultures were obtained with the astroglioma U251 (26.7 ± 3.2 ng/ml at 72 h) and U373 (17.5 ± 1.1 ng/ml at 72 h), whereas the other astrocytoma (D384) hardly responded to IFN-γ (1000 U/ml) stimulation (3.7 ± 0.3 ng/ml at 72 h). C1-Inh secreted by SK-N-SH increased from 1.8 ± 1.1 ng/ml by unstimulated cells to 56.7 ± 13.2 ng/ml in culture supernatants of cells stimulated for 72 h with 1000 units IFN-γ/ml (Fig. 2). Recombinant IL-1β, IL-6 or TNF-α, added at concentrations of 4, 16, 62, 250 and 1000 U/ml, did not stimulate the expression of C1-Inh by primary astrocytes, astrocytomas or neuroblastomas. Although approximately tenfold less potent than IFN-γ, IL-6 clearly stimulated C1-Inh synthesis by the hepatoma cell line Hep3B, whereas TNFα was without effect (Fig. 2). Similar results were obtained with hepatoma cell line HepG2.
Expression of C1-Inh mRNA
Fig. 1 rhIFN-γ increases in a dose-dependent fashion the secretion of functionally active C1-Inh by primary human astrocyte cultures, as assessed in the RII ELISA in which bound C1-Inh is allowed to capture biotinylated activated C1s. Cells were cultured in Dulbecco’s modified Eagle’s medium containing nutrient mixture HAM’s F-10 and 0.1% fetal calf serum alone ( ), or containing either 16 ( ), 62.5 ( ), 250 ( ) or 1000 ( ) U rhIFN-γ. C1-Inh in supernatants accumulates in time, as shown for supernatants harvested either 24, 48 or 72 h after the start of stimulation (rhIFN-γ recombinant human interferon-γ, C1-Inh C1-esterase inhibitor, ELISA enzyme-linked immunosorbent assay)
To investigate whether the inability of astrocyte cell cultures to secrete C1-Inh in response to other cytokines than IFN-γ is due to unresponsiveness at the transcriptional level or due to a defective protein synthesis or secretion, Northern blotting for C1-Inh was performed. Size-separated total RNA (15 µg) isolated from IFN-γ (1000 U/ml)-stimulated primary astrocyte cultures, but also from non-stimulated subconfluent cell cultures in 80 cm2 flasks that had accumulated more than 7 ng C1-Inh/ml after 72 h culture, expressed detectable 2.1-kb C1-Inh mRNA (Fig. 3) when probed with 32P-labeled C1-Inh cDNA. To quantify the C1-Inh mRNA expression by astrocytes in culture, dot-blots of total RNA derived from 2 × 104 astrocytes cultured in 24-well plates were hybridized with the 32P-labeled C1-Inh probe. A dose-dependent increase in C1-Inh mRNA, expressed as arbitrary units, was seen
291
Fig. 2 Astrocyte cultures derived from adult human CNS and neuroblastoma SK-N-SH secrete C1-Inh in response to IFN-γ, but not to IL-1β, IL-6 or TNF-α. No response of neuroblastoma IMR-32 to either cytokine was observed. C1-Inh secretion in response to IL-6 by the hepatoma cell line Hep3B is comparable with that of the astrocytes to IFN-γ (IL-1β was not tested on Hep3B). Cells (2 × 104/0.5 ml per well) were either unstimulated ( ) or cultured in the presence of 16 ( ), 62 ( ), 250 ( ),or 1000 ( ) U of either rhIFN-γ, IL-1β, IL-6, or TNF-α. C1-Inh levels are expressed as ng/ml (IL interleukin, TNF tumor necrosis factor)
occasional slight vascular wall staining. Counterstaining with Congo red revealed that, even in gray matter areas with neuritic plaques, neuron-like cells showed a hybridization signal, whereas cells with glial morphology in the region did not express C1-Inh mRNA (Fig. 7).
when primary astrocytes or astrocytoma U251 cells were cultured in the presence of IFN-γ. Stimulation with 1000 U IFN-γ/ml resulted in an approximately tenfold increase in C1-Inh expression, as compared with unstimulated cells (Fig. 4). C1-Inh mRNA expression by the astrocytoma cell line D384 also increased dose dependently, but remained low, which is in line with its barely detectable C1-Inh protein secretion. No stimulatory effect of IL-1β or IL-6, added in a concentration range from 16 to 1000 U/ml, on C1-Inh mRNA expression by either primary astrocytes, or astrocytoma U251 cultures was observed (Fig. 4). Similar results to those with the astrocytes were obtained with neuroblastoma cell lines SK-N-SH and IMR-32. In line with the ELISA results, SK-N-SH clearly expressed C1-Inh mRNA in response to IFN-γ, whereas IMR-32 did not (Fig. 5). Cellular localization of C1-Inh mRNA expression Hybridization signal with an antisense DIG-labeled C1-Inh RNA probe was observed in cells with neuron morphology in all layers of temporal cortex specimens of all AD cases and controls, as well as in the hippocampal and temporal cortex sections of a Down’s syndrome case (Fig. 6). Frequently, hybridization was observed in the walls of blood vessels. No labeling of the white matter was observed. Sense probe hybridizations were negative, except for some
Fig. 3 Northern blot showing a 2.1-kb hybridization signal, when total RNA (15 µg/lane) from rhIFN-γ-stimulated (IFN) astrocytes (A88 and A157) or human umbilical vein endothelial cells (EC) is hybridized with the 32P-labeled C1-Inh probe. Unstimulated (–) A88 that constitutively synthesize 7 ng C1-Inh/ml (as measured by ELISA), also show a hybridization signal. Hybridization of the same blot with a human elongation factor (hEF-1α) cDNA probe shows that comparable amounts of total RNA were loaded in each lane
292
Fig. 5 IFN-γ clearly stimulates C1-Inh mRNA expression by neuroblastoma cell line SK-N-SH, whereas it is without effect on IMR-32. Although a minor effect of IL-6 on SK-N-SH is seen, the AD plaque-associated cytokines IL-1β, IL-6 and TNF-α do not induce a dose-dependent response effect. Quantitation (arbitrary units) of hybridization signal (32P-labeled C1-Inh cDNA) on dotblots with total RNA isolated from neuroblastoma cell lines SK-NSH and IMR-32 (2 × 104 cells/well) cultured (72 h) in the absence ( ), or presence of 16 ( ), 62.5 ( ), 250 ( ) or 1000 ( ) U/ml of either rhIFN-γ or IL-6, and 62.5 ( ), 250 ( ) or 1000 ( ) U/ml of either IL-1β or TNF-α. Results from one representative experiment are expressed as mean of three replicate cultures
Discussion
Fig. 4 Quantitation (arbitrary units) of hybridization signal (32Plabeled C1-Inh cDNA) on dot-blots with total RNA isolated from either primary astrocyte cultures (A64 and A157) or astrocytomas (D384 and U251) (2 × 104 cells/well) cultured (72 h) in the absence ( ), or presence of 16 ( ), 62.5 ( ), 250 ( ) or 1000 ( ) U/ml of either rhIFN-γ, IL-1β or IL-6. Expression is highest in primary astrocyte cultures stimulated with IFN-γ, no effect of IL-1β or IL-6 on C1-Inh expression is seen. Results from one representative experiment are expressed as mean of two or three (U251) replicate cultures (ND IL-6 was not tested on U251)
IFN-γ, but not other cytokines such as IL-1β, IL-6 and TNF-α, the levels of which are elevated in AD brains [12], stimulated C1-Inh expression by astrocytoma cell lines, primary cultures of adult human astrocytes as well as by neuroblastoma SK-N-SH (Figs. 1, 2). The functional activity of the IL-1β used was confirmed by the finding that the astrocyte cultures, which did not respond to IL-1β with C1-Inh synthesis, clearly produced IL-6 in response to IL-1β [6], which also indicates the presence of functional IL-1 receptors on these cells. Another indication for the functional activity of the cytokines used was that, although neuroblastoma cell line IMR-32 did not produce C1-Inh, IMR-32 could be stimulated to secrete high levels of C3 in response to TNF-α and IL-1β and moderate levels of C4 in response to IL-6 and IFN-γ (data not shown). These findings also illustrate that, insofar as neuroblastoma can be regarded as neuron-like cells, neurons may be potential producers of more than one complement factor. The functional activity of the IL-6 used was confirmed by the fact that the hepatoma cell line Hep3B was found
293
Fig. 6A–F Localization of C1-Inh RNA as assessed with RNA in situ hybridization with digoxigenin-labeled C1-Inh antisense and sense probes. A Neurons in the temporal cortex specimen of a nondemented control (case 311) show extensive hybridization signal. B Neuronal expression of C1-Inh in the temporal cortex of an AD case with Lewy bodies (case 276). C Large hippocampal neurons in the CA4, as well as small neurons in the granular cell layer of a Down’s syndrome case (no. 21) show strong hybridization for C1Inh. D Control section adjacent to C hybridized with sense RNA. E. Higher magnification of CA4 area in C. F Higher magnification of granular layer in C. A, B, E, F × 100; C, D × 25
to secrete C1-Inh (up to 78 ng/ml by 2 × 104 cells in 72 h; Fig. 2) in response to IL-6 (1000 U/ml), in line with previous reports [45]. Not only astrocytes and neuroblastomas (Fig. 2), but also other cell types consistently fail to produce C1-Inh in response to IL-6, whereas they can be stimulated by IFN-γ [23, 38, 45]. The presence of an IL-6 response-like element upstream from the start site for C1Inh expression in hepatocytes, which may be lacking or may be lying further upstream in monocytes, could explain the differences in expression of C1-Inh by hepatocytes and monocytes in response to IL-6 [9], and possibly also those between the hepatoma cell lines and astrocytes and neuroblastomas observed in this study.
294
Fig. 7 Glial cells adjacent to a birefringent Congo red-stained plaque (arrow) in this hippocampal section of a Down’s syndrome case (no. 21) with severe AD are negative for C1-Inh hybridization signal, whereas neurons and also vascular lining express C1-Inh message. Near the birefringent ghost tangle (arrowhead) no hybridization signal is seen. × 100
Northern blotting experiments revealed that IFN-γ-stimulated primary astrocyte cultures express C1-Inh mRNA with an apparent size of 2.1 kb (Fig. 3), coinciding with the known size of C1-Inh mRNA expressed by other cell types [37]. Quantitation of hybridization signals on dotblots indicated, that the levels of astrocytic C1-Inh mRNA paralleled C1-Inh secretion by the cells (Fig. 4), indicating that astrocytes are refractory to the stimulatory effect of IL-1β and IL-6 at the transcriptional level, whereas the rhIFN-induced increase in C1-Inh synthesis is probably due to increased transcription. Likewise, neuroblastomas IMR-32 and SK-N-SH did not respond to IL-1β, TNF-α or IL-6. A possible minor effect, although not in a dosedependent fashion, of IL-6 on SK-N-SH C1-Inh mRNA expression was occasionally seen with some concentrations of IL-6 (Fig. 5). rhIFN-γ, on the other hand, clearly stimulated C1-Inh mRNA expression by SK-N-SH, but had no stimulatory effect on IMR-32, coinciding with the C1-Inh protein secretion data. Because activated lymphocytes, the main producers of IFN-γ, are undetectable in control and AD neocortex [16] it is unclear how C1-Inh production can be stimulated in the brain. Rat neurons produce IFN-γ -like factors [24], which have been reported to have a similar bioactivity to lymphocyte-derived IFN-γ [30]. However, it remains to be investigated whether such IFN-γ-like factors can also be produced by human neurons and can stimulate complement synthesis. In previous studies, RT-PCR was used to investigate whether C1-Inh is expressed in AD and control brain [40, 42]. From these studies the relative contributions of vascular endothelial cells that are known to produce C1-Inh [37] and cells in the brain parenchyma could not be determined. Our in situ hybridization findings indicate that, besides endothelial cells, neurons express C1-Inh mRNA (Fig. 6). Even in areas in which congophilic plaques with
activated astrocytes and microglia were present, neurons were the only cell type that clearly expressed C1-Inh mRNA (Fig. 7). This is in line with an earlier report [42], in which polyclonal anti-C1-Inh antibodies were shown to detect C1-Inh in neurons, whereas monoclonal antibodies specific for inactivated and/or complexed C1-Inh immunostained either activated astrocytes [40] or dystrophic neurites [42]. Taken together, these findings suggest that C1Inh after interaction with its target proteases may accumulate within astrocytes or neurons, and that neurons may be the major local site of synthesis of C1-Inh. The involvement of neurons as producers of inflammatory mediators in response to injury or inflammation was first suggested by the Finch group, based on RNA in situ hybridization [22, 25] and cell culture studies [22, 25, 36], in which neurons were shown to express C1q, C4 and clusterin mRNA. Our RNA in situ hybridization data suggest that neurons are also potential producers of regulators of complement activation in vivo. In conclusion, this study shows that, although functionally intact C1-Inh can be produced by human astrocytes isolated from postmortem brain specimens in vitro, the appropriate stimulus appears to be lacking in vivo. IFN-γ, which is apparently absent from the brain parenchyma, was found to be a potent stimulator of C1-Inh expression in vitro, both at the mRNA and at the protein level, whereas the plaque-associated cytokines IL-1β, IL-6 and TNF-α did not stimulate C1-Inh expression. RNA in situ hybridization shows that neurons rather than astrocytes express C1-Inh mRNA in situ, suggestive for a role for neurons as potential producers of regulators of complement activation in vivo. Because Aβ can activate complement via the classical pathway, and C1-Inh is the only known inhibitor of C1 activation, targeting the expression of C1-Inh in the brain may offer a potential strategy to prevent or retard the onset of AD pathology. Acknowledgements The authors thank the Netherlands Brain Bank (coordinator Dr. R. Ravid) for supplying the human CNS tissue, Dr. Eldering (CLB, Amsterdam) for providing the C1-Inh cDNA clone pTZ53, Drs. C. Langeveld and B. Drukarch (U251 and D384 cell lines), Dr. Schweitzer (primary HUVEC) and Dr. Van Nostrand (immortalized HUVEC) for providing the respective cell cultures, J. T. Van Veldhuisen and G. J. Oskam for preparing the photographs. This study was supported in part by a grants from the Netherlands Organization for Scientific Research (NWO; grant 903–51–108) and from the Internationale Stichting Alzheimer Onderzoek (ISAO; grant 95504).
References 1. Akiyama H, Ikeda K, Kondo H, McGeer PL (1992) Thrombin accumulation in brains of patients with Alzheimer’s disease. Neurosci Lett 146: 152–154 2. Alley MC, Scudiero DA, Monks A, Hursey ML, Czerwinski MJ, Fine DL, Abbott BJ, Mayo JG, Shoemaker RH, Boyd MR (1988) Feasibility of drug screening with panels of human tumor cell lines using a microculture tetrazolium assay. Cancer Res 48: 589–601 3. Aulak KS, Donaldson VH, Coutinho M, Davis AE III (1993) C1-Inhibitor: structure/function and the biologic role. Behring Inst Mitt 93: 204–213
295 4. Balmforth AJ, Ball SG, Freshney RI, Graham DI, McNamee HB, Vaughan PF (1986) D-1 Dopaminergic and β-adrenergic stimulation of adenylate cyclase in a clone derived from the human astrocytoma cell line G-CCM. J Neurochem 47: 715– 719 5. Barnum SR (1995) Complement biosynthesis in the central nervous system. Crit Rev Oral Biol Med 6: 132–146 6. Blom MAA, Twillert MGH van, Vries SC de, Engels F, Finch CE, Veerhuis R, Eikelenboom P (1997) NSAIDS inhibit the IL-1β induced IL-6 release from human post mortem astrocytes: the involvement of prostaglandin E2. Brain Res 777 : 210–218 7. Bock SC, Skriver K, Nielsen E, Thogersen HC, Wiman B, Donaldson VH, Eddy RL, Marrinan J, Radziejewska E, Huber R, Shows TB, Magnusson S (1986) Human C1 inhibitor: primary structure, cDNA cloning and chromosomal localization. Biochemistry 25: 4292–4301 8. Brands JHGM, Maassen JA, Hemert FJ van, Amons R, Möller W (1986) The primary structure of the α subunit of human elongation-factor 1. Structural aspects of guanidine-nucleotidebinding sites. Eur J Biochem 155: 167–171 9. Carter PE, Duponchel C, Tosi M, Fothergill JE (1991) Complete nucleotide sequence of the gene for human C1 inhibitor with an unusually high density of Alu elements. Eur J Biochem 197: 301–308 10. De Groot CJA, Langeveld CH, Jongenelen CAM, Montagne EJ, Van der Valk P, Dijkstra CD (1997) Establishment of human adult astrocyte cultures derived from postmortem multiple sclerosis and control brain and spinal cord regions: immunophenotypical and functional characterization. J Neurosci Res 49: 342–354 11. De Smet BJGL, De Boer J-P, Agterberg J, Rigter G, Bleeker WK, Hack CE (1994) Clearance of human native, proteinasecomplexed, and proteolytically inactivated C1-Inhibitor in rats. Blood 81: 56–61 12. Dickson DW (1997) The pathogenesis of senile plaques. J Neuropathol Exp Neurol 56: 321–339 13. Eikelenboom P, Stam FC (1982) Immunoglobulins and complement factors in senile plaques. An histochemical study. Acta Neuropathol (Berl) 57: 239–242 14. Eikelenboom P, Veerhuis R (1996) The role of complement and acivated microglia in the pathogenesis of Alzheimer’s disease. Neurobiol Aging 17: 673–680 15. Eikelenboom P, Hack CE, Rozemuller JM, Stam FC (1989) Complement activation in amyloid plaques in Alzheimer’s dementia. Virchows Arch [B] 56: 259–262 16. Eikelenboom P, Rozemuller JM, Kraal G, Stam FC, McBride PA, Fraser H (1991) Cerebral amyloid plaques in Alzheimer’s disease but not in scrapie-affected mice are closely associated with a local inflammatory process. Virchows Arch [B] 60: 329– 336 17. Eldering E, Nuijens JH, Hack CE (1988) Expression of functional human C1 Inhibitor in COS cells. J Biol Chem 263: 11776–11779 18. Gasque P, Ischenko A, Legoedec J, Mauger C, Schouft M-T, Fontaine M (1993) Expression of the complement classical pathway by human glioma in culture. J Biol Chem 268: 25068– 25074 19. Gasque P, Thomas A, Fontaine M, Morgan BP (1996) Complement activation on human neuroblastoma cell lines in vitro: route of activation and expression of functional complement regulatory proteins. J Neuroimmunol 66: 29–40 20. Ishii T, Haga S (1984) Immuno-electron microscopic localization of complements in amyloid fibrils of senile plaques. Acta Neuropathol (Berl) 63: 296–300 21. Jiwa NM, Kanavaros P, Valk P van der, Walboomers JMM, Horstman A, Vos W, Mullink H, Meijer CJLM (1993) Expression of c-myc and bcl-2 oncogene products in Reed-Sternberg cells independent of presence of Epstein-Barr virus. J Clin Pathol 46: 211–217
22. Johnson SA, Lampert-Ettchells M, Pasinetti GM, Rozovsky I, Finch CE (1992) Complement mRNA in the mammalian brain: reponses to Alzheimer’s disease and experimental lesioning. Neurobiol Aging 13: 641–648 23. Katz Y, Strunk RC (1989) Synthesis and regulation of C1 inhibitor in human skin fibroblasts. J Immunol 142: 2041–2045 24. Kiefer R, Kreutzberg GW (1990) Gamma interferon-like immunoreactivity in the rat nervous system. Neuroscience 37: 725–734 25. Lampert-Etchells M, Pasinetti GM, Finch CE, Johnson SA (1993) Regional localization of cells containing complement C1q and C4 mRNAs in the frontal cortex during Alzheimer’s disease. Neurodegeneration 2: 111–121 26. Lappin DF, Guc DD, Hill A, McShane T, Whaley K (1992) Effect of interferon-γ on complement gene expression in different cell types. Biochem J 281: 437–442 27. McGeer PL, Akiyama H, Itagaki S, McGeer EG (1989) Immune system response in Alzheimer’s disease. Can J Neurol Sci 16: 516–527 28. Melton DA, Krieg PA, Rebagliati MR, Maniatis T, Zinn K, Green MR (1984) Efficient in vitro synthesis of biologically active RNA and RNA hybridization probes from plasmids containing a bacteriophage SP6 promoter. Nucleic Acids Res 12: 7035–7056 29. Nuijens JH, Huijbregts CCM, Mierlo GM van, Hack CE (1987) Inactivation of C1 inhibitor by proteases: demonstration by a monoclonal antibody of a neodeterminant on inactivated, noncomplexed C1 inhibitor. Immunology 61: 387–389 30. Olsson T, Kelic S, Edlund C, Bakhiet M, Höjeberg B, Meide PH van der, Ljungdahl Å, Kristensson K (1994) Neuronal interferon-γ immunoreactive molecule: bioactivities and purification. Eur J Immunol 24: 308–314 31. Pasinetti GM, Johnson SA, Rozovsky I, Lampert-Etchells M, Morgan DG, Gordon MN, Willoughby D, Finch CE (1992) Complement C1qB and C4 mRNAs reponses to lesioning in rat brain. Exp Neurol 118: 117–125 32. Pontèn J, Westermark B (1978) Properties of human malignant glioma cells in vitro. Med Biol 56: 184–193 33. Raap AK, Corput MP van de, Vervenne RA, Gijlswijk RP van, Tanke HJ, Wiegant J (1995) Ultra-sensitive FISH using peroxidase-mediated deposition of biotin- or fluorochrome tyramides. Hum Mol Genet 4: 539–534 34. Rogers J, Cooper R, Webster S, Schultz J McGeer PL, Styren SD, Civin WH, Brachova L, Bradt B, Ward P, Lieberburg I (1992) Complement activation by β-amyloid in Alzheimer’s disease. Proc Natl Acad Sci USA 89: 10016–10020 35. Rozemuller JM, Eikelenboom P, Pals ST, Stam, FC (1989) Microglial cells around amyloid plaques in Alzheimer’s disease express leucocyte adhesion molecules of the LFA-1 family. Neurosci Lett 101: 288–292 36. Rozovsky I, Morgan TE, Willoughby DA, Dugich-Djordjevich MM, Pasinetti GM, Johnson SA, Finch CE (1994) Selective expression of clusterin (SGP-2) and complement C1q and C4 during responses to neurotoxins in vivo and in vitro. Neuroscience 62: 741–758 37. Schmaier AH, Murray SC, Ghashyam DH, Farber A, Kuo A, McCrae K, Cines DB (1989) Synthesis and expression of C1 inhibitor by human umbilical vein endothelial cells. J Biol Chem 264: 18173–18179 38. Tosi M (1993) Molecular genetics of C1-inhibitor and hereditary angiooedema. In: Whaley K, Loos M (eds) Complement in health and disease, 2nd edn. Kluwer, London, pp 245–267 39. Veerhuis R, Van der Valk P, Janssen I, Zhan S-S, Van Nostrand WE, Eikelenboom P (1995) Complement activation in amyloid plaques in Alzheimer’s disease brains does not proceed further than C3. Virchow Arch 426: 603–610 40. Veerhuis R, Janssen I, Hack CE, Eikelenboom P (1996) Early complement components in Alzheimer’s disease brains. Acta Neuropathol 91: 53–60 41. Walker DG, McGeer PL (1992) Complement gene expression in human brain: comparison between normal and Alzheimer disease cases. Mol Brain Res 14: 109–116
296 42. Walker DG, Yasuhara O, Patston PA, McGeer EG, McGeer PL (1995) Complement C1 inhibitor is produced by human brain tissue and is cleaved in Alzheimer disease. Brain Res 675: 75– 82 43. Yasuhara O, Walker DG, McGeer PL (1994) Hageman factor and its binding sites are present in senile plaques of Alzheimer’s disease. Brain Res 654: 234–240
44. Zahedi K, Prada AE, Davis AE III (1994) Transcriptional regulation of the C1 inhibitor gene by γ-interferon. J Biol Chem 269: 9669–9674 45. Zuraw BL, Lotz M (1990) Regulation of the hepatic synthesis of C1 inhibitor by the hepatocyte stimulating factors interleukin 6 and interferon γ. J Biol Chem 365: 12664–12670
