A Cyclic Undecamer Peptide Mimics a Turn in Folded Alzheimer Amyloid β and Elicits Antibodies against Oligomeric and Fibrillar Amyloid and Plaques

The 39- to 42-residue amyloid β (Aβ) peptide is deposited in extracellular fibrillar plaques in the brain of patients suffering from Alzheimer's Disease (AD). Vaccination with these peptides seems to be a promising approach to reduce the plaque load but results in a dominant antibody response directed against the N-terminus. Antibodies against the N-terminus will capture Aβ immediately after normal physiological processing of the amyloid precursor protein and therefore will also reduce the levels of non-misfolded Aβ, which might have a physiologically relevant function. Therefore, we have targeted an immune response on a conformational neo-epitope in misfolded amyloid that is formed in advance of Aβ-aggregation. A tetanus toxoid-conjugate of the 11-meric cyclic peptide Aβ(22–28)-YNGK′ elicited specific antibodies in Balb/c mice. These antibodies bound strongly to the homologous cyclic peptide-bovine serum albumin conjugate, but not to the homologous linear peptide-conjugate, as detected in vitro by enzyme-linked immunosorbent assay. The antibodies also bound—although more weakly—to Aβ(1–42) oligomers as well as fibrils in this assay. Finally, the antibodies recognized Aβ deposits in AD mouse and human brain tissue as established by immunohistological staining. We propose that the cyclic peptide conjugate might provide a lead towards a vaccine that could be administered before the onset of AD symptoms. Further investigation of this hypothesis requires immunization of transgenic AD model mice.


Introduction
Alzheimer's disease (AD) is a neurodegenerative disorder and the most common cause of dementia in elderly [1,2]. A characteristic of the disease is formation of plaques in the brain or in brain blood vessels. These plaques originate from a membrane-bound protein, amyloid precursor protein (APP). An a-helical fragment of 39-42 amino acid residues is cleaved by band c-secretases from APP thus forming a soluble amyloid b (Ab) peptide. Soluble Ab initially adopts an extended conformation but at high concentrations, soluble Ab will undergo conformational changes and form oligomers, protofibrils, and fibrils. In AD, fibrillar Ab is deposited in the brain as amyloid plaques, which is one of the main neuropathological hallmarks of the disease. However, accumulating studies suggest that the soluble oligomeric Ab instead of insoluble Ab in amyloid plaques is the culprit in AD [3][4][5][6][7][8] and therapeutic approaches aimed at preventing the formation of these oligomeric isoforms may be able to reduce the progression of the disease. In line with this concept, immunization of transgenic mice [9] with a suspension of ''preaggregated'' Ab  and the adjuvant quillaja saponin 21 appeared to be beneficial. Based on these results, a phase I clinical trial was started. Antibodies present in human sera recognized plaques and Ab deposits in brain blood vessels [10]. The antibodies did not recognize APP or soluble Ab. In the following phase II clinical trial, 20% of the vaccine recipients generated anti-Ab antibody titers. Unfortunately, this trial had to be terminated since 6% of the patients developed meningoencephalitis as a vaccine-related side effect. This side effect was caused by a cellular inflammatory reaction, attributed to a T helper cell type 1 response to epitopes located in the central and Cterminal part of Ab   [11,12].
Multiple ongoing studies aim at improving the Ab vaccination strategy [13][14][15][16][17][18]. The use of T helper cell type 2 directing adjuvants [19,20] or the use of formulations without any adjuvant [21] are under investigation. In addition, it has been proposed to use C-terminally truncated Ab peptides [22,23] or peptide mimics (affitopes) of the N-terminus [24]. Antibodies induced by Ab  are dominantly directed against the linear N-terminal epitope [25,26], although generation of conformation-specific antibodies against other regions within aggregated Ab has been reported [27]. A disadvantage of a vaccine against the N-terminus of Ab is that it will interfere with the normal physiological processing of APP. It may not be without risk to administer such a vaccine before onset of symptoms of AD.
By targeting an immune response exclusively on misfolded Ab the undesired response against the N-terminus of Ab may be avoided all together. A structural model of fibrillar Ab  predicts folding of monomeric Ab(1-42) into a cross-b unit. Two antiparallel extended b-strands, residues 11-25 and 28-42, are connected via a sharp bend around amino acid residues S26 and N27 [28]. Fig. 1 shows a simplification of the original model. A recent study of a particular oligomer of N-Met-Ab(1-42) suggests a bend around the sequence V 24 -G-S-N 27 [29].
The models of folded Ab show some resemblance with the bturn structure in surface loops of meningococcal outer membrane protein PorA. Previously, we have stabilized the b-turn conformation of small meningococcal peptides by adding an artificial sequence YNGK9, in which K9 is a modified lysine residue for selective conjugation to a carrier protein, followed by main chain (''head to tail'') amide cyclization [30,31]. Likewise, it appeared possible to prepare small cyclic peptides mimicking the turn in folded Ab. A panel of cyclic decameric and undecameric peptides spanning six or seven residues from the region 21-31 of Ab and YNGK9 was prepared and conjugated to tetanus toxoid (TTd). The conjugates were used for immunization of mice. A tetanus toxoid-conjugate of one of the peptides, cyclo[Ab(22-28)-YNGK9], elicited antibodies that specifically recognized misfolded Ab.

Immunization of mice
Balb/c mice were immunized twice with oligomeric or fibrillar Ab(1-42) without adjuvant or with the different amyloid peptide/ TTd-conjugates using aluminum phosphate as adjuvant. The pooled sera from each group immunized with peptide conjugate were analyzed by ELISA using plain TTd, homologous peptide/ BSA-conjugate and oligomeric or fibrillar Ab(1-42) as coating antigens. The sera showed a good antibody titer against TTd and the homologous cyclopeptide (titers of 10 4 -10 5 were obtained, data not shown).
Six out of eight pooled anti-cyclo-Ab peptide antisera were not cross-reactive with oligomeric or fibrillar Ab  in ELISA or a dot blot assay. Antibodies against cyclo[Ab(23-29)-YNGK9] showed a weak cross-reactivity. However, anti-cyclo[Ab(22-28)-YNGK9] antibodies were strongly cross-reactive with oligomeric or fibrillar Ab  in ELISA (Fig. 2) and the antibodies recognized Ab deposits in AD mouse and human brain tissue (Fig. 3). The anti-peptide antibodies did not recognize full length APP in Western blot assay (Fig. 4).
The cyclization of the peptide was essential to produce a functional antigen. Conjugates of linear control peptides, i.e. Ac-K9-Ab(21-31)-NH 2 or the fully homologous Ac-K9-Ab(22-28)-YNG-NH 2 , induced good anti-peptide antibody titers but the antibodies failed to recognize oligomeric or fibrillar Ab(1-42) (Fig. 2). The incorporation of the artificial sequence YNGK9 in the cyclopeptide was essential to mimic the turn in folded Ab . Incorporation of the sequence GAIK9, i.e. Ab(29-31)-K9, instead of YNGK9 did not provide a turn-mimicking peptide, since antibodies against cyclo[Ab(22-31)-K9] were not functional.  [28]. Residues in red are solvent accessible; residues in blue (and to a lesser extent in gray) are shielded from the solvent. The peptide cyclo[Ab (22)(23)(24)(25)(26)(27)(28)-YNGK9] is a mimic for misfolded Ab. YNGK9 is a turn-stabilizing sequence and K9 is a side-chain-modified lysyl residue for selective conjugation to a protein carrier. Other YNGK9-containing cyclic peptides prepared ( The anti-cyclo[Ab(22-28)-YNGK9] antibodies were very specific for the common Ab sequence. Amyloid b with the Dutch, Arctic, or Italian mutations (E22Q, E22G, or E22K, respectively) was not recognized. Unrelated but partially homologous proteins like the islet amyloid polypeptide (amylin)or the human Wiskott-Aldrich Syndrome protein (WASP) were also not recognized (data not shown).
Two immunizations with fibrillar Ab(1-42) gave only two sera with titers $2000 on the homologous coating, two very low responders and four non-responders (titer,100). Poor responsiveness to aggregated Ab(1-42) was found earlier in clinical and animal studies [35,36].
In conclusion, cyclo[Ab(22-28)-YNGK9] mimics a conformational epitope in folded Ab(1-42) that normally does not induce an antibody response. Cyclo[Ab (22)(23)(24)(25)(26)(27)(28)-YNGK9]-conjugate may be an interesting vaccine candidate against AD. Since the response is directed specifically to misfolded Ab, it is unlikely that vaccination will interfere with normal physiological processing of the amyloid precursor protein. Thus, vaccination might be possible at an early stage of AD symptoms or before. This seems to be indicated since plaque deposits have already reached a near maximum level at a mild stage of AD [37].
Further studies will aim at improvement of the immune response to cyclo[Ab(22-28)-YNGK9]-TTd by variation of the

Ethics statement
Tissue from AD patients was obtained from the Netherlands Brain Bank (NBB; Amsterdam, The Netherlands). The NBB performs brain autopsies with short post-mortem intervals, and the brain donors have given written informed consent for using the tissue and for accessing the extensive neuropathological and clinical information for scientific research, in compliance with ethical and legal guidelines [38]. The study was approved by the independent Review Board (''Medisch Ethische Toetsingscommissie, METc'') of the VU University Medical Center, Amsterdam, The Netherlands.
Morphological studies of mouse brain tissue were approved by the Ethical Committee on Animal Experiments of the Royal Netherlands Academy of Arts and Sciences (approval ID: NIN 06.51).
Immunization of mice was approved by the Ethical Committee on Animal Experiments of the National Institute of Public Health and the Environment (approval ID: 2007 00331 and 2008 00179), as required under Dutch Law on the use of laboratory animals. All efforts were made to minimize suffering of the animals.

Disaggregation of Ab(1-42)
Lyophilized Ab  was dissolved in trifluoroacetic acid at a concentration of 1.0 mM, left to stand at room temperature for 1 h and dried under a stream of nitrogen and, thereafter, in a vacuum (1 mm Hg) for 15 min. The peptide was then redissolved in hexafluoroisopropanol at a concentration of 1.0 mM and, after 1 h of incubation at room temperature, dried as described above [39]. The peptide was stored at 220uC for 18-20 h.

Preparation of oligomeric or fibrillar Ab(1-42)
Disaggregated of Ab(1-42) was dissolved in dimethyl sulfoxide at a concentration of 5.0 mM, diluted 50-fold with either phosphate buffered saline (PBS), pH 7.2, or 10 mM hydrochloric acid. The solution in PBS was incubated at 4uC for 24 h (to give oligomers), whereas the solution in 10 mM HCl was incubated at 37uC for 24 h (to give fibrils) [32].

Immunization of mice
Groups of eight female Balb/c mice of 6-8 weeks of age were immunized with 25 mg oligomeric or fibrillar Ab

ELISA
Microtiter plates (Greiner 655092) were coated with Ab  or peptide-BSA conjugates. Freshly prepared Ab(1-42) oligomers or fibrils were diluted to a final concentration of 2.5 mM (11.3 mg/ ml) in 0.04 M sodium carbonate/bicarbonate buffer, pH 9.7. Peptide-BSA conjugates in phosphate buffered saline, pH 7.2 (PBS), had a total protein concentration of 0.5 mg/ml. Aliquots (100 ml) of these solutions were transferred into wells of the plates.
The plates were incubated for 90 min at 37uC. The plates were further processed as described earlier [42]. Titers were calculated as the reciprocal serum dilution at 50% of the maximum optical density (OD 50 ).

Immunohistochemical staining
Human brain sections of the hippocampus of several donors with Alzheimer disease, Braak 5 or 6 [43], were used (Netherlands Brain Bank). In addition, brain sections of 9 month old APPswe/ PS1dE9 [42] mice with a significant plaque deposition were used. Cryosections (10 mm) were cut from unfixed, directly frozen tissue, thaw-mounted, dried for 1 hour and stored in a sealed box at 220 C. For immunostaining, sections were fixed in 4% PFA-PBS solution for 10 min, washed in 0.05 M phosphate buffer (PB) for 10 min with 2 exchanges and blocked with 10% normal donkey serum (NDS) +0.4% Triton X-100 in 0.05 M PB for 1 hour at RT. The blocking solution was discarded and diluted mouse sera (1:300; first antibody) in 3% NDS +0.4% Triton X-100 in 0.05 M