Upregulating CD59: a New Strategy for Protection of Neurons from Complement-Mediated Degeneration

Upregulating CD59: a new strategy for protection of neurons from complement-mediated degeneration

MV Kolev1,3, T Tediose1,3, An increasing number of studies have shown a critical role for the membrane 1 1 attack complex, synthesized on activation of the terminal pathway of B Sivasankar , CL Harris , the complement system, in causing demyelination and neuronal death in 2 1 J Thome , BP Morgan and neurodegeneration. The aim of this study was to develop a strategy to RM Donev1 increase the resistance of neurons to complement damage by modulating the expression of membrane complement regulatory protein CD59, the only 1Department of Medical Biochemistry and inhibitor of the terminal pathway of the complement cascade. We exploited Immunology, School of Medicine, Cardiff our recent finding that CD59 expression is regulated by the neural-restrictive University, Cardiff, UK and 2Academic Unit of Psychiatry, Institute of Life Science, The School silencer factor (REST) and designed a novel REST-derived peptide (REST5) of Medicine, Swansea University, Swansea, UK containing the nuclear localization domain of the wild-type protein. The effect of REST5 and the mechanism by which it modulates CD59 expression Correspondence: were modelled in neuroblastoma cells transfected with expression constructs, Dr Rossen Donev, Department of Medical Biochemistry and Immunology, School of and then confirmed in human neurons differentiated from neural progenitor Medicine, Cardiff University, Heath Park, cells. REST5 increased the expression of CD59 in neurons by fivefold and Cardiff CF14 4XN, UK. protected them from complement-mediated lysis spontaneously triggered E-mail: [email protected] by neurons. As a source of complement, we used either human serum or conditioned medium from primary human oligodendroglia. This study brings new insight into immunopharmacological research that may serve to inhibit neuronal death triggered by the terminal pathway of complement activation. The Pharmacogenomics Journal (2010) 10, 12–19; doi:10.1038/tpj.2009.52; published online 3 November 2009

Keywords: neurodegeneration; neuroprotection; complement lysis; CD59; complement regulators; neural-restrictive silencer factor (NRSF, REST)

Introduction

The complement system is a key component of innate immunity, having a central role in defence against pathogens.1 It is also a powerful drive to initiate inflammation and can, if unregulated, cause pathology leading to severe tissue damage. Complement has been implicated in various human neurodegenerative disorders, such as Alzheimer’s disease (AD), Huntington’s disease, Pick’s disease and multiple sclerosis (MS).2 Although viewed for years as an immunoprivileged organ, the central nervous system contains and synthesizes many components of the immune system. Cultured and primary neurons, astrocytes, microglia, 3These authors contributed equally to this work. oligodendroglia and endothelial cells were shown to produce all the complement components of both alternative and classical pathways, and also the proteins Received 1 July 2009; revised 24 September 3–8 2009; accepted 30 September 2009; involved in the terminal pathway forming membrane attack complex (MAC). published online 3 November 2009 Thus, the local production of complement components in the brain ensures the Targeting CD59 for neuroprotection from complement MV Kolev et al 13

complete functionality of the complement system without Materials and methods the need for infusion of plasma-derived complement proteins. Cells and treatment Inflammation and complement systems as a generator ReNcell VM human neural progenitor cells20 were obtained of proinflammatory molecules such as C3a and C5a have from Millipore UK (Watford, Hertfordshire, UK), and were previously been thought of as detrimental in the patho- expanded in T25-cm2 tissue culture flasks precoated physiology of neurodegenerative diseases. However, emer- with laminin (Sigma-Aldrich, Gillingham, UK) in Dulbecco’s ging genetic data, magnetic resonance imaging studies, modified Eagle’s medium:F12 medium containing B27 immunopathological evidence and in vivo studies in animal neural cell supplement mix (Invitrogen, Paisley, UK), models9,10 challenge this simplistic view. Evidence leads to L-glutamine (2 mM, Invitrogen), heparin (10 Units mlÀ1, the conclusion that inflammation is tightly regulated, and Sigma-Aldrich), gentamicin (50 mgmlÀ1, Invitrogen), Basic that its net effect may be beneficial in neurodegenerative fibroblast growth factor (bFGF) (10 ng mlÀ1, Invitrogen) and diseases such as MS and AD, thus explaining some of the epidermal growth factor (EGF) (20 ng mlÀ1, Sigma-Aldrich). results from recent trials of anti-inflammatory agents. Cells were differentiated (REN neurons) over a 2-week period Besides the proinflammatory role of complement compo- in the above medium, but without bFGF and EGF.20 Primary nent C5a, it was recently shown that this anaphylatoxin human oligodendroglia (TCS CellWorks, Buckingham, UK) may also have a protective role in AD brain through the were recovered and maintained in 2 ml of oligodendroglia activation of neuroprotective mitogen-activated protein medium in a T25-cm2 tissue flask according to the supplier’s kinases.11 In support of this, animals genetically deficient protocol. The culture medium was collected for complement in complement component C5 were found to be more lysis assay twice a week and replaced with fresh one until susceptible to hippocampal excitotoxic lesions.12 These cells were viable. Human neuroblastoma cell line IMR32 findings suggest a novel noninflammatory role for C5a in (European Collection of Animal Cell Cultures, Salisbury, UK) modulating neuronal responses to excitotoxins. However, was maintained in RPMI 1640 with 10% heat-inactivated several studies have unambiguously shown a critical role for fetal calf serum, supplemented with glutamine, penicillin MAC synthesized on activation of the terminal pathway and streptomycin (Invitrogen). All cells in culture were of the complement system in causing demyelination in maintained at 37 1C in a humidified atmosphere of 95% humans and experimental autoimmune encephalomyelitis air/5% CO2. and MS in animal models.13,14 REN neurons were treated with 10 mM His-TAT-REST5 or a CD59, a ubiquitously expressed membrane-bound control His-TAT-NLS peptide (Bio-Synthesis, Lewisville, TX, complement regulatory protein (mCReg), exists to protect USA) for 8, 16 or 24 h. His-TAT-REST5 peptide (Figure 1) self-cells from damage caused by the activation of the comprises a His-tag, HIV-1 transactivator protein transduc- complement cascade by blocking the formation of MAC.1 tion domain (TAT) and zinc-finger domain 5 (REST5) from There is very little information regarding the expression REST protein, separated from each other by three Gly of mCReg in neuronal cells. Studies to date suggest that residues to ensure flexibility of each of these domains.21 In cultured primary neurons express only low levels of CD59 the control His-TAT-NLS peptide (Figure 1), the REST5 and CD46.15,16 CD55 was not detectable by immuno- domain was replaced by the classical nuclear localization histochemistry staining.17 In contrast, oligodendroglia, the signal (NLS) for delivery of RNA polymerase II through an main function of which in higher vertebrates is to insulate importin-a1 receptor. axons exclusively in the central nervous system, abundantly Purification of neurons and separation of cytoplasmic and nuclear express CD59 and, although at a lower level, CD55 and proteins CD46.18 Furthermore, neuronal cell lines and primary To obtain high-purity neurons (497%), we used magnetic neurons were shown to activate the complement cascade beads coated with anti-human CD90 (Gentaur Europe, spontaneously, which, together with the very low levels Brussels, Belgium), following the supplier’s protocol. To of CD59, suggest their vulnerability to MAC lysis.19 separate the cytoplasm from nuclei, purified neurons bound Therefore, it is likely that CD59 might be an important to beads were lysed by incubation on ice for 5 min with resistance factor in protecting against MAC-mediated phosphate-buffered saline containing 0.3% NP-40. Beads demyelination and lysis. Thus, targeting the expression were removed and the lysate was centrifuged for 3 min of MAC regulator CD59 seems an attractive approach at 2000 g. The supernatant contained the cytoplasm of for protecting neurons from complement-mediated degene- neurons, whereas nuclei formed a pellet. Nuclei were lysed ration. with RIPA buffer (25 mM Tris–HCl pH 7.6, 150 mM NaCl, 1% The main objective of this study was to develop a strategy NP-40, 1% sodium deoxycholate, 0.1% SDS). Protein con- to increase the resistance of neurons to complement damage centration in each sample was determined by the Micro by modulating the expression of CD59. We achieved this by BCA Protein Assay Kit (Perbio Science UK, Cramlington, further exploiting our recent finding on the regulation of Northumberland, UK). the expression of CD59 by neural-restrictive silencer factor (REST), which allowed us to design a novel peptide that Dot-blot analyses overexpressed CD59 in neurons and protected them from Dot-blots were carried out using standard protocols. Poly- complement-mediated lysis. clonal anti-REST antibody, C-15, specific for the C-terminus

The Pharmacogenomics Journal Targeting CD59 for neuroprotection from complement MV Kolev et al 14

residues and the REST5 zinc-finger domain (Figure 1). A control plasmid in which the REST5 domain was replaced by the NLS (Figure 1) was also designed (sense: 50-CTAGAGCC ACCATGGCCCATCATCATCATCATCATGGCGGCGGCCCTA AGAAGAAGAGAAAGGTGTGAG-30; antisense: 50-GATCCTC ACACCTTTCTCTTCTTCTTAGGGCCGCCGCCATGATGATG ATGATGATGGGCCATGGTGGCT-30). The constructs were transfected into IMR32 cells using Effectene reagent (Qiagen, Crawley, West Sussex, UK) and cells were selected and maintained in medium containing 100 mgmlÀ1 hygromycin B (Invitrogen).

Flow cytometry The effect of the REST5 expression construct and REST5 peptide on the expression of CD59, CD55 and CD46 at the protein level was assessed by staining neuroblastoma cells or neurons (3 Â 105) with mouse monoclonal antibodies MEM43 (catalogue number ab9182), BRIC216 (catalogue number ab33111) and MEM258 (catalogue number ab789), all from Abcam plc, Cambridge, UK, against CD59, CD55 and CD46, respectively, and analysing by flow cytometry. Figure 1 Design of peptide for modulation of CD59 expression. A detailed description of this procedure was given pre- (a) Full-length REST contains the N-terminal repressor domain (black viously.23 All measurements were made at least in duplicate box), followed by nine zinc-finger domains (grey boxes). The ninth and each experiment was replicated twice. Results were domain is located at the C-terminus and serves as transcriptional repressor. Sequences of zinc-finger domain 5 of REST (REST5) used for combined and statistically analysed by Student’s t-test. A enhancing the expression of CD59 and classical nuclear localization P-value o 0.05 was considered to show statistically signi- signal (NLS) used as a control are given in panel (b). The TAT domain ficant differences. was used for delivery of synthesized peptides into neurons. The full sequences of peptides expressed by plasmids transfected into IMR32 Chromatin immunoprecipitation (ChIP) and those synthesized and given to REN neurons are also presented. The IMR32 cells transfected either with the REST5 or NLS expression construct were analysed as previously described,22 using Magna ChIP A kit (Millipore). Immunopre- of the protein (catalogue number sc-15120, Santa Cruz cipitation was carried out with rabbit polyclonal anti-Sp1 Biotechnology, Santa Cruz, CA, USA), was used to detect the (catalogue number 17-601, Millipore), rabbit polyclonal full-length REST on dot-blots from different cell types. For anti-AP2 (catalogue number ab52222, Abcam), sheep poly- detection of the REST5 peptide, a monoclonal mouse anti- clonal anti-CPBP (catalogue number AF3499, R&D Systems His antibody (clone HIS.H8, catalogue number 05-949, Europe, Abingdon, UK) or polyclonal anti-REST (C-15) Millipore) against the N-terminal His-tag was used. To (catalogue number sc-15120, Santa Cruz Biotechnology). control for the amount of sample loaded, blots were stained Nonimmune rabbit IgG supplied in the Magna ChIP A with a polyclonal anti-b-actin antibody (catalogue number kit was used as a control for the background of these ANA54590, Cambridge Bioscience, Cambridge, UK). Ana- experiments. The naked co-immunoprecipitated DNAs were lyses were carried out as previously described.22 then used as templates in quantitative PCR assays as described below. Statistical significance of differences Design of peptide for overexpression of CD59 between experimental groups was assessed using an un- To test the effect of REST-derived peptide on the expression paired, two-tailed Student’s t-test. of CD59, a pDR2DEF1a-based construct for expression in mammalian cells was designed. The sense (50-CTAGAGCCA Quantitative PCR CCATGATGCATCATCATCATCATCATGGTGGTGGTTATAAA Total RNAs from IMR32 cells and REN neurons treated as TGTGAACTTTGTCCTTACTCAAGTTCTCAGAAGACTCATC described above were purified using the GenElute Mamma- TAACTAGACATATGCGTACTCATTGAG-30) and antisense lian Total RNA Miniprep kit (Sigma-Aldrich). cDNAs were (50-GATCCTCAATGAGTACGCATATGTCTAGTTAGATGAGT synthesized using TaqMan Reverse Transcription reagents CTTCTGAGAACTTGAGTAAGGACAAAGTTCACATTTATAAC (Applied Biosystems, Warrington, UK), and amplification CACCACCATGATGATGATGATGATGCATCATGGTGGCT-30) was carried out on a Mini Opticon Real-Time PCR Unit oligonucleotides were synthesized (Biomers.net GmbH, Ulm, (Bio-Rad, Hemel Hempstead, Herts, UK) using iQ SYBR Germany) and annealed, leaving the XbaI and Bam HI over- Green Supermix (Bio-Rad) and primer pairs described in hangs at their 50 and 30 ends, respectively, for ligation into an Table 1. RNA expression levels were calculated using the appropriately digested pDR2DEF1a vector. This construct comparative Ct method (DDCt). At least two independent expresses a peptide comprising six His residues, three Gly experiments were carried out for each mRNA, and Student’s

The Pharmacogenomics Journal Targeting CD59 for neuroprotection from complement MV Kolev et al 15

Table 1 Primer pairs used for real-time quantitative PCR suppressor REST23 could be explored to design a peptide capable of enhancing neuronal resistance to com- Sequence Primer sequence Concen- plement, a plasmid was engineered to inhibit the delivery detected tration, nM of endogenous REST to the nucleus (Figures 1a and b). The plasmid encoded a peptide comprising the zinc-finger 0 0 CD59 5 -TAACCCAACTGCTGACTGCAA-3 (F) 50 domain 5 of REST (REST5) essential for nuclear delivery 0 0 5 -TTTGGTAATGAGACACGCATCAA-3 (R) 300 through a REST-specific receptor,24,25 and a His-tag to allow CD55 50-TTCCCCCAGATGTACCTAATGC-30 (F) 300 discrimination between endogenously expressed REST and 50-TTACAGTATCCTCGGGAAAACTTGT-30 (R) 50 CD46 50-GCGCTTTCCTGGGTTGCT-3’ (F) 300 REST5. A control construct in which the REST5 domain was 50-ACAGGCATCGGAGAAGGAGTAC-30 (R) 300 replaced by the classical NLS for delivery through im- b-Actin 50-ACGGCCAGGTCATCACTATTG-30 (F) 50 portin-a1 was also designed (Figures 1b). Therefore, the 50-ACGGCCAGGTCATCACTATTG-30 (R) 50 expressed control peptide should not compete with endogenous REST for nuclear delivery. As primary neurons F, indicates forward primer and R, indicates reverse primer. are difficult to transfect, plasmids were tested in IMR32 neuroblastoma cells that, similar to primary neurons, express full-length REST.23 Transfected cells were selected t-test was applied to calculate significance in changes of and maintained in the presence of hygromycin B in culture the expression pattern. In a similar manner, we quantified medium. Dot-blot analysis of REST and REST5 in nuclear the binding of different transcription factors to the cd59 extracts unambiguously showed that expression of REST5 promoter. Immunoprecipitated DNAs were used as tem- in IMR32 cells resulted in a 12-fold decrease in the amount plates (10 ng per reaction) in a quantitative assay with a of full-length REST delivered into the nuclei (Figures 2a). primer pair for detection of the responsive element within The control NLS peptide was delivered within the nuclei 23 the cd59 promoter. The Ct value calculated for the input with a similar efficiency to that of REST5; however, it did DNA used in ChIPs was used to normalize the cell number. not affect the nuclear delivery of endogenous REST. ChIP Two independent analyses of immunoprecipitated DNAs with antibodies against transcription factors previously were carried out for each antibody. showntobeinvolvedinregulationofcd59 expres- sion23 confirmed that this REST5-mediated depletion of Complement lysis assay endogenous REST in nuclei allowed the binding of Lysis was assessed by measuring lactate dehydrogenase transcriptional activators CPBP, Sp1 and AP2 within (LDH) release using a Colorimetric Cytotoxicity assay kit the responsive promoter sequence of the cd59 gene (Oxford Biomedical Research, Oxford, MI, USA) as pre- (Figures 2b). viously described.23 Different dilutions of normal human Treatment with REST5 yielded a fivefold increase in the serum (NHS) or conditioned medium from primary human expression of CD59 at both mRNA and protein levels in oligodendroglia were used as the source of complement. As IMR32 (Figures 3a and b). Expression of the other two negative controls for the complement-mediated lysis, we mCReg, CD55 and CD46, was not affected by the REST5 either inactivated complement by heat treatment (15 min peptide. Although expression of CD55 was detectable at the at 56 1C) of NHS and the conditioned medium or depleted mRNA level, it was negligible at the protein level, suggesting the NHS/conditioned medium of C8 using a monoclonal no role for this protein in the protection of neurons from affinity column. In experiments with CD59 blocking, excess complement-mediated lysis. It has been previously shown of Fab fragments (10 mgmlÀ1) generated from MEM43 mAb that fetal human neurons can spontaneously activate the against CD59 (ImmunoPure Fab preparation kit, Perbio classical pathway of complement cascade without any Science UK) was preincubated with cells for 30 min at 37 1C additional stimuli, resulting in susceptibility to complement and washed twice with phosphate-buffered saline; there- lysis.17 The alteration in the expression of CD59 led to the after, serum/conditioned medium was added at appropriate protection of IMR32 cells from spontaneously triggered dilutions. All experiments were carried out in triplicate for complement lysis (Figure 3c). Around threefold less cells each condition. The percentage of lysed cells was calculated transfected with the REST5 expression plasmid were lysed at using the following formula: the maximum serum concentration compared with cells % Lysis ¼ððmeasured LDH release À spontaneous releaseÞ transfected with the control construct or nontransfected =ðmaximum release À spontaneous releaseÞÞÂ100 cells. Fab fragments from CD59-blocking antibody were used in control experiments, which resulted in equal lysis in The experiment was replicated twice and data were analysed the three cultures, showing that the increased neuronal by Student’s t-test. resistance to complement damage is a result of CD59 upregulation. Results REST5 peptide increases the expression of CD59 in human neurons Modelling of a peptide for upmodulation of CD59 expression Having shown that plasmid expressing REST5 transfected To test whether the previously reported mechanism for into IMR32 protected cells from complement lysis, a peptide suppression of CD59 expression by the transcriptional was synthesized comprising a His-tag, HIV-1 transactivator

The Pharmacogenomics Journal Targeting CD59 for neuroprotection from complement MV Kolev et al 16

Figure 2 Peptide derived from the nuclear delivery domain of REST, Figure 3 Expression of the REST5 peptide protects neuroblastoma cells REST5, depletes nuclei from endogenous REST and allows the binding of from complement-mediated damage. (a) Quantitative PCR analysis of transcriptional activators to the cd59 promoter. (a) Dot-blot analysis of the expression of CD59, CD55 and CD46 mRNA in IMR32 cells endogenous REST in the nuclei isolated from IMR32 cells transfected transfected with REST5 or nuclear localization signal (NLS) expression either with the REST5 expression construct or nuclear localization signal constructs. NLS-transfected cells were used as control and their (NLS) control. Anti-His antibody was used to determine the expression expression was set as 100%. Results from three independent experi- of both REST5 and NLS peptides and anti-b-actin was used as a control ments (±s.d.; *Po0.001). (b) Cells treated as in panel a were stained for the amount of loaded proteins (10 mg per well). Representative with antibodies against mCReg and their expression was quantified by autoradiograph from two experiments is shown on the left-hand side. flow cytometry. Results are from three independent experiments (±s.d.; Mean data from densitometric analysis of duplicates±s.d. are given on *Po0.001). (c) IMR32 cells transfected either with REST5 (-m-) or NLS the right-hand side. Expression in cells transfected with VLS onstruct was (-K-) expression constructs, or nontreated (-’-), were tested for their set as 100%, except for the anti-His antibody, in which REST5- sensitivity to spontaneously triggered complement activation. Normal transfected cells were set as 100% (*Po0.001). (b) Chromatin human serum (NHS) was used as a source of complement. Lysis assay immunoprecipitation (ChIP) was carried out on IMR32 cells transfected with preincubation of the three cell lines (REST5, -n-; NLS, -J-; with either REST5 or NLS expression constructs with antibodies against nontreated, -&-) with CD59-blocking Fab fragment was carried out as a REST, CPBP, Sp1 and AP2. Immunoprecipitation with nonimmune control. Values are mean±s.d. for three independent experiments rabbit IgG was carried out as a control for the assay background. Pull- (*Po0.01). down and input DNAs were characterized for the presence of the responsive cd59-promoter element by quantitative PCR. Levels of binding for each transcription factor in cells transfected with the control NLS construct were set as 100%. Values are mean±s.d. for two presence of endogenous REST in nuclei (Figure 4b). independent ChIP experiments, each analysed in duplicate (*Po0.001). Densitometric analysis of dot-blot data for endogenous REST showed a decrease of around 50% within the first 8 h of REST5 treatment. However, this was associated with only protein transduction domain (TAT) and REST5 sequence a slight upregulation of CD59 mRNA expression and no (His-TAT-REST5), separated from each other by three Gly effect on CD59 protein at this time (Figures 5a and b). At residues to ensure flexibility/functionality of each of these 16 h, endogenous REST was decreased eightfold compared domains26 (Figure 1b). The TAT domain efficiently delivers with controls, and this was sustained at 24 h (Figure 4b). cargo molecules to the cytoplasm in a wide range of species, At these time points, the expression of CD59 at both mRNA including human.27 A peptide comprising His-TAT-NLS and protein level was increased fivefold (Figures 5a, b). domains was used as control. When neurons were exposed to His-TAT-REST5 (REST5) peptides, the peptide was detected REST5 protects neurons from complement-mediated damage in nuclear extracts within 2 h and accumulation increased Once we established the optimum treatment of human through 4 h, slowly decreasing thereafter (Figure 4a). The neurons with the REST5 peptide, which results in a fivefold control NLS peptide was delivered to nuclei with similar overexpression of CD59, we tested the effect of this efficiency to that of the REST5 peptide. Human neurons treatment on the expression of other mCReg proteins were treated with these peptides every 4 h over a period of (Figure 6a). The REST5 peptide was added every 4 h over a 24 h. Samples were taken every 8 h and analysed for the period of 16 h. Expression of CD46 was not affected during

The Pharmacogenomics Journal Targeting CD59 for neuroprotection from complement MV Kolev et al 17

Figure 4 The REST5 peptide successfully depletes endogenous REST from nuclei of neurons. (a) Dot-blot analysis of intracellular and nuclear delivery of REST5 (left-hand side) and nuclear localization signal (NLS) (right-hand side) peptides in REN neurons. The peptides were added to the medium at a concentration of 10 mM and neurons were purified with anti-CD90 beads at different treatment times (0, 2, 4 and 6 h). An equal amount of protein (10 mg) was loaded into each well. (b) The amount of endogenous REST in nuclear extracts (NE) from neurons repeatedly treated every 4 h either with REST5 or NLS peptide as a control was detected at 0, 8, 16 and 24 h from the first addition. Antibody that does not recognize REST5 was used to detect endogenous REST and results were quantified densitometrically (right-hand side). Values are mean±s.d. for two independent experiments (*Po0.001).

Figure 6 The REST5 protects human neurons from complement- mediated damage. (a) Flow cytometry analysis of the expression of CD59, CD55 and CD46 in neurons repeatedly incubated with either REST5 or nuclear localization signal (NLS) for 16 h. Values are mean±s.d. for two independent experiments (*Po0.001). (b) Comple- ment killing of neurons (normal human serum (NHS) used as a source of complement) treated with REST5 (-m-), a control NLS peptide (-K-) or nontreated (-’-). Lysis assay with preincubation of cells with CD59- blocking Fab fragments (REST5, -n-; NLS, -J-; nontreated, -&-) was Figure 5 The REST5 peptide increases the expression of CD59 in carried out as a control. (c) Complement-killing assays using condi- neurons at the mRNA and protein level. (a) Quantitative PCR analysis of tioned medium from primary human oligodendroglia as a source the time course of expression of CD59 in neurons repeatedly treated of complement. Neurons treated with REST5 (-m-) or control NLS every 4 h either with REST5 or nuclear localization signal (NLS) as a peptide (-K-) were incubated with different dilutions of the condi- control. The expression in neurons at the beginning of NLS treatment tioned medium. Lysis assay with preincubation of cells with CD59- was set as 100%. Values are mean±s.d. for two independent blocking Fab fragments (REST5, -n-; NLS, -J-) was carried out as a experiments (*Po0.001). (b) Cells treated as in panel a were stained control. Values are mean±s.d. for three independent experiments with antibody against CD59 and their expression was quantified by flow (*Po0.01). cytometry. Results are from two independent experiments (±s.d.; *Po0.001). Fab fragments derived from MEM43-blocking monoclonal the course of this treatment and CD55 remained negligible. antibody. This resulted in an equal lysis of neurons treated However, as a result of increased CD59 expression by the with REST5, control NLS peptides or nontreated. These REST5 management, we observed a substantial increase in data show that the increased neuronal resistance to the protection of neurons from complement-mediated complement damage conferred by REST5 is a result of killing spontaneously triggered by neurons (Figure 6b). CD59 upregulation. Neurons treated with REST5 had around 2.5-fold less lysed Our next step was to study whether complement compo- cells by complement sourced from serum, compared with nents synthesized locally from brain cells could lyse neurons cells treated with the control NLS peptide or nontreated and whether the REST5 peptide confers resistance to ones. In control experiments, CD59 was preblocked by eventual complement-mediated damage (Figure 6c). We

The Pharmacogenomics Journal Targeting CD59 for neuroprotection from complement MV Kolev et al 18

carried a complement lysis assay with REN human neurons To deliver the REST5 peptide to the cytoplasm of human similar to that presented in Figure 6b; however, as a source neurons, a TAT domain was added to the N-terminus. This of complement, we used conditioned medium from primary enabled efficient cell uptake and nuclear delivery of the human oligodendroglia, cells that were shown to synthesize peptide and a subsequent reduction in endogenous REST in all components of the complement system at both mRNA the nucleus, as the REST5 competed for nuclear translocation and protein levels.8 Indeed, the conditioned medium was through the REST-specific nuclear receptor (Figure 4). The capable of lysing neurons. This damage was inhibited by endogenous REST protein is reported to have a relatively short the REST5-mediated upmodulation of CD59 expression by half-life in embryonic stem cells (2 h) and neural stem cells B2.5-fold in comparison with the control treatment with (1 h).28 Improving the stability of the REST5 peptide might the NLS peptide. Control experiments with preblocking of further improve the efficiency of the removal of endogenous CD59 in both neuronal cultures showed that the upregu- REST from the cd59 promoter to increase gene expression. lated CD59 was responsible for the observed neuroprotec- The strategy proposed here for upregulation of CD59 tion in our in vitro assay system, proving that neuronal lysis expression would presumably affect the expression of other was a result of the activation of the terminal pathway of genes regulated by REST. These targets include a number of complement cascade. genes that require degradation of REST to initiate neuronal differentiation.29,30 Therefore, the mechanism of action of the peptide described here suggests that treatment with Discussion REST5 should not result in neuronal dedifferentiation. However, further genome-wide studies, such as microarray The critical role of the complement system and particularly of expression analysis, are necessary to determine any un- MAC in neuronal death in diverse neurodegenerative diseases wanted side effects that this peptide might have on neurons. is now much better understood.10,13,14 Although oligo- Our complement lysis data using conditioned medium dendroglia, fetal astrocytes and astrocyte cell lines are well from oligodendroglia (Figure 6) are in agreement with those protected from complement damage by an abundant expres- from previous study showing that oligodendroglia synthe- sion of mCReg,18 neurons are particularly susceptible to MAC size all components of the complement cascade,8 and attack because of the low expression of CD59 and CD46 on further support the hypothesis that the brain locally their surface.15 To increase the resistance of neurons to produces a functional complement system that could be complement-mediated damage, we designed a new strategy an important mediator in a number of events related to to upmodulate the expression of CD59 by targeting REST on neurodegenerative diseases. Neurons are particularly sensi- the basis of our recent demonstration that REST has a key role tive to effects driven by the complement because of their in the regulation of cd59 gene expression.23 Zinc-finger very weak protection from complement attack compared domain 5 of REST (REST5), essential for the nuclear delivery with the rest of the central nervous system cells.31 Although of the protein,25 either expressed in IMR32 neuroblastoma targeting the complement system in different neurodegen- cells or added as a synthetic peptide designed for uptake into erative diseases has been proposed as a very prospective neurons, successfully competed with endogenous REST for strategy,32 such an approach requires careful consideration the REST-specific nuclear translocation receptor24 (Figures 2– because a number of studies have suggested that certain 4). This resulted in the removal of endogenous REST from the components generated during complement activation (for cd59 promoter and recruitment of transcriptional activators example, C5a) are actually of benefit, at least in animal CPBP, Sp1 and AP2, previously shown to be involved in the models of MS and AD.11,12,33 The strategies described here modulation of CD59 expression.23 This interplay between cause an upregulation of CD59 on the neurons through activators and inhibitors of CD59 transcription yielded a mechanisms that are very clearly defined. By increasing the fivefold upregulation of the expression of CD59 and expression of CD59, formation of lytic MAC is inhibited, markedly decreased the susceptibility of both IMR32 and whereas generation of early activation products, including neurons to complement-mediated damage spontaneously C5a, which, for example, might contribute to plaque triggered by neurons (Figures 3 and 6). We observed that clearance in AD, is unimpeded. Targeting MAC in this way 8 h after beginning REST5 treatment, the amount of en- is thus an exceptionally attractive approach. However, dogenous REST in nuclei was reduced by around 50% (Figure further studies are needed to understand the exact role of 4). This was sufficient to observe a significant upmodulatory each component of the complement system in human effect on the expression of CD59 at the mRNA level (B40%). neurodegeneration to ensure the best choice of therapeutic However, longer time was needed to start accumulating more target. The strategy pioneered here should be further CD59 protein and reach the maximum effect of a fivefold extended in an in vivo animal model to better understand increase in CD59 expression (Figure 5). Flow cytometry any side effects that the REST5 peptide may have. analysis did not detect any significant expression of CD55 protein in both IMR32 cells and neurons differentiated from human neural progenitor cells. This observation is in concordance with an earlier study17 in which investigators Conflict of interest failed to detect the expression of CD55 by immunohisto- chemical staining of primary human neurons. The authors declare no conflict of interest.

The Pharmacogenomics Journal Targeting CD59 for neuroprotection from complement MV Kolev et al 19

Acknowledgments and neurons of the adult rat brain. J Comp Neurol 1994; 339: 387–400. This study was funded by the Medical Research Council New 17 Singhrao S, Neal J, Rushmere N, Morgan B, Gasque P. Spontaneous Investigator Grant G0700102 (RM Donev) and by the Wellcome classical pathway activation and deficiency of membrane regulators render human neurons susceptible to complement lysis. Am J Pathol Trust Programme Grant 068590 (BP Morgan). 2000; 157: 905–918. 18 Gasque P, Morgan B. Complement regulatory protein expression by a human oligodendrocyte cell line: cytokine regulation and comparison References with astrocytes. Immunology 1996; 89: 338–347. 19 Yang L, Li R, Meri S, Rogers J, Shen Y. Deficiency of complement defense 1 Walport MJ. Advances in immunology: complement (first of two parts). protein CD59 may contribute to neurodegeneration in Alzheimer’s N Engl J Med 2001; 344: 1058–1066. disease. J Neurosci 2000; 20: 7505–7509. 2 Gasque P, Neal J, Singhrao S, McGreal E, Dean Y, Van B et al. Roles 20 Donato R, Miljan E, Hines S, Aouabdi S, Pollock K, Patel S et al. of the complement system in human neurodegenerative disorders: Differential development of neuronal physiological responsiveness in pro-inflammatory and tissue remodeling activities. Mol Neurobiol 2002; two human neural stem cell lines. BMC Neurosci 2007; 8: 36. 25: 1–17. 21 Harris C, Williams A, Linton S, Morgan B. Coupling complement 3 Barnum S. Complement biosynthesis in the central nervous system. regulators to immunoglobulin domains generates effective anti- Crit Rev Oral Biol Med 1995; 6: 132–146. complement reagents with extended half-life in vivo. Clin Exp Immunol 4 Silva C, Rio M, Cruz C. Local immunoglobulin synthesis and blood-brain 2002; 129: 198–207. barrier assessment in subacute sclerosing panencephalitis. Eur Neurol 22 Donev RM, Cole DS, Sivasankar B, Hughes TR, Morgan BP. p53 1985; 24: 128–133. regulates cellular resistance to complement lysis through enhanced 5 Gasque P, Fontaine M, Morgan B. Complement expression in human expression of CD59. Cancer Res 2006; 66: 2451–2458. brain—biosynthesis of terminal pathway components and regulators in 23 Donev R, Gray L, Sivasankar B, Hughes T, van den Berg C, Morgan B. human glial cells and cell lines. J Immunol 1995; 154: 4726–4733. Modulation of CD59 expression by restrictive silencer factor-derived 6 Gasque P, Julen N, Ischenko A, Picot C, Mauger C, Chauzy C et al. peptides in cancer immunotherapy for neuroblastoma. Cancer Res Expression of complement components of the alternative pathway by 2008; 68: 5979–5987. glioma cell lines. J Immunol 1992; 149: 1381–1387. 24 Shimojo M, Hersh LB. REST/NRSF-interacting LIM domain protein, a 7 Morgan B, Gasque P. Extrahepatic complement biosynthesis: where, putative nuclear translocation receptor. Mol Cel Biol 2003; 23: 9025–9031. when and why? Clin Exp Immunol 1997; 107: 1–7. 25 Shimojo M, Lee JH, Hersh LB. Role of zinc finger domains of the trans- 8 Hosokawa M, Klegeris A, Maguire J, McGeer P. Expression of cription factor neuron-restrictive silencer factor/repressor element-1 complement messenger RNAs and proteins by human oligodendroglial silencing transcription factor in DNA binding and nuclear localization. cells. Glia 2003; 42: 417–423. J Biol Chem 2001; 276: 13121–13126. 9 Martino G, Adorini L, Rieckmann P, Hillert J, Kallmann B, Comi G et al. 26 Harris C, Hughes C, Williams A, Goodfellow I, Evans D, Caterson B et al. Inflammation in multiple sclerosis: the good, the bad, and the complex. Generation of anti-complement ‘prodrugs’: cleavable reagents for Lancet Neurol 2002; 1: 499–509. specific delivery of complement regulators to disease sites. J Biol Chem 10 Kolev M, Ruseva M, Harris C, Morgan B, Donev R. Implication of 2003; 278: 36068–36076. complement system and its regulators in Alzheimer’s disease. Curr 27 Lee K, Luu N, Kaneski C, Schiffmann R, Brady R, Murray G. Improved Neuropharmacol 2009; 7: 1–8. intracellular delivery of glucocerebrosidase mediated by the HIV-1 TAT 11 Osaka H, Mukherjee P, Aisen P, Pasinetti G. Complement-derived protein transduction domain. Biochem Biophys Res Commun 2005; 337: anaphylatoxin C5a protects against glutamate-mediated neurotoxicity. 701–707. J Cell Biochem 1999; 73: 303–311. 28 Westbrook T, Hu G, Ang X, Mulligan P, Pavlova N, Liang A et al. 12 Pasinetti G. Inflammatory mechanisms in neurodegeneration and SCFbeta-TRCP controls oncogenic transformation and neural differen- Alzheimer’s disease: the role of the complement system. Neurobiol tiation through REST degradation. Nature 2008; 452: 370–374. Aging 1996; 17: 707–716. 29 Johnson D, Mortazavi A, Myers R, Wold B. Genome-wide mapping of 13 Mead R, Singhrao S, Neal J, Lassmann H, Morgan B. The membrane in vivo protein-DNA interactions. Science 2007; 316: 1497–1502. attack complex of complement causes severe demyelination associated 30 Weissman A. How much REST is enough? Cancer Cell 2008; 13:381–383. with acute axonal injury. J Immunol 2002; 168: 458–465. 31 Singhrao S, Neal J, Rushmere N, Morgan B, Gasque P. Differential 14 Mead R, Neal J, Griffiths M, Linington C, Botto M, Lassmann H et al. expression of individual complement regulators in the brain and Deficiency of the complement regulator CD59a enhances disease choroid plexus. Lab Invest 1999; 79: 1247–1259. severity, demyelination and axonal injury in murine acute experimental 32 Lucas S, Rothwell N, Gibson R. The role of inflammation in CNS injury allergic encephalomyelitis. Lab Invest 2004; 84: 21–28. and disease. Br J Pharmacol 2006; 147(Suppl 1): S232–S240. 15 Vedeler C, Ulvestad E, Bjørge L, Conti G, Williams K, Mørk S et al. The 33 Maier M, Peng Y, Jiang L, Seabrook T, Carroll M, Lemere C. expression of CD59 in normal human nervous tissue. Immunology 1994; Complement C3 deficiency leads to accelerated amyloid beta plaque 82: 542–547. deposition and neurodegeneration and modulation of the microglia/ 16 Pasinetti G, Johnson S, Oda T, Rozovsky I, Finch C. Clusterin (SGP-2): macrophage phenotype in amyloid precursor protein transgenic mice. a multifunctional glycoprotein with regional expression in astrocytes J Neurosci 2008; 28: 6333–6341.

The Pharmacogenomics Journal