. Transcutaneous beta-amyloid immunization reduces cerebral beta-amyloid deposits without T cell infiltration and microhemorrhage. Proc Natl Acad Sci U S A. 2007 Feb 13;104(7):2507-12. PubMed.

Recommends

Please login to recommend the paper.

Comments

Make a Comment

To make a comment you must login or register.

Comments on Primary Papers and News

  1. Drs. Nikolic, Tan, and their colleagues present an interesting paper demonstrating the efficacy of transcutaneous (TC) immunization with Aβ1-42 peptide and cholera toxin (CT) adjuvant. Wild-type and PSAPP (APPsw/PSEN1dE9) mice were immunized weekly for 4 weeks and then biweekly for an additional 12 weeks. Anti-Aβ titers were first detected at 4 weeks and rose thereafter. Anti-Aβ antibodies were of mostly IgG1 (Th2) isotypes, but lower levels of IgG2a (Th1) and IgG2b (Th2) antibodies were also generated. Splenocytes from Aβ/CT immunized mice secreted dramatically elevated levels of IL-4 along with less dramatic elevations in IFN-γ and IL-2 compared to CT-treated control mice. PSAPP mice were immunized from 4 to 8 months of age and showed significant reductions in cerebral plaque burden, soluble Aβ40 and Aβ42, and insoluble Aβ40 and Aβ42. Plasma Aβ was increased. T cell infiltration and microhemorrhage were not observed in any of the animals. The authors conclude that transcutaneous immunization with Aβ is effective in lowering cerebral Aβ (perhaps by a peripheral sink mechanism) and may be safer than the original Aβ vaccines tested in mice and humans because Langerhans cell precursors, a major antigen presenting cell in skin, drive a predominantly Th2 immune response.

    The data presented here bode well for a transcutaneous delivery of an Aβ vaccine for Alzheimer disease. They confirm, to some degree, data we reported last year in which transcutaneous immunization with a short Aβ immunogen (dendrimic Aβ1-15: 16 copies of Aβ1-15 on a lysine tree) and adjuvant LT(R192G) resulted in Aβ antibody generation in wild-type mice (Seabrook et al., 2006). Anti-Aβ levels and isotypes were similar between the two studies.

    Interestingly, in our study, TC immunization with full-length Aβ peptide did not lead to anti-Aβ production. This inconsistency with Nikolic's paper may be due to differences in adjuvant (CT vs. LT), dose (200 µg Aβ1-42 vs. 100 µg Aβ1-40/42), schedule (weekly for 4 weeks then biweekly for 12 weeks vs. biweekly for 16 weeks), and/or slight differences in the transcutaneous application of the vaccine.

    We did not report results for TC immunization with dAβ1-15 in AD mouse models; however, we believe that by avoiding an Aβ-specific T cell epitope in our immunogen, the vaccine may avoid an Aβ-specific T cell autoimmune response, such as that suggested to be responsible for the adverse events in the AN-1792 trial using full-length Aβ1-42 peptide. In our study, splenocytes from wild-type mice TC immunized with dAβ1-15 did not recognize full-length Aβ, as they did not proliferate upon restimulation with Aβ peptide. In the Nikolic et al. paper, splenocytes from wild-type and PSAPP mice TC immunized with Aβ/CT recognized full-length Aβ, resulting in increased levels of both Th2 and Th1 cytokines. Whether or not such a response in humans would predispose to an Aβ autoimmune response remains unclear.

    In the current paper, the authors found no T cell recruitment to brain; however, most Aβ vaccines studies in mice have shown the same result. It has been difficult to replicate in mice the adverse events observed in the human trial. Only a couple of papers have reported T cell recruitment to the brain upon Aβ immunization, and only when pertussis toxin was co-administered with the vaccine and again 2 days later (Furlan et al., 2003; Monsonego et al., 2006). In particular, Monsonego used an Aβ T cell epitope, Aβ10-24, for his immunogen to immunize APP double transgenic mice that overexpressed IFN-γ. Thus, in the present paper, it is not completely surprising that T cells were not recruited to brain. The need remains to test these vaccines (including ours) in an animal model that mimics the effects seen in the human subjects in the AN-1792 trial.

    Lastly, microhemorrhage has been associated with passive transfer of Aβ antibodies in older APP or PSAPP transgenic mice bearing cerebrovascular amyloid angiopathy (CAA) at the time of immunization; however, it is less clear that the same effect would occur if passive immunization were to be delivered to young animals prior to plaque deposition or CAA. The current paper, in which mice were first immunized prior to plaque deposition, suggests that the lack of microhemorrhage indicates safety; however, older mice with CAA should be examined. To date, we have not observed microhemorrhage with any of our active Aβ vaccines when given in a "prevention" trial in which the immunization began prior to plaque deposition.

    In summary, the authors are to be congratulated on the successful generation of anti-Aβ antibodies and lowering of cerebral Aβ in their study of transcutaneous immunization in PSAPP transgenic mice. Further studies are now needed to assess the safety issues raised. We agree with the authors that an effective and safe transcutaneous Aβ vaccine would be convenient, less costly, and less painful than an injectable vaccine that requires a doctor's office visit.

    View all comments by Cynthia Lemere
  2. The skin is a well-established effective route for vaccination. The authors evaluate the efficacy of transcutaneous immunization of PSAPP Tg mice in reducing cerebral amyloidosis using aggregated Aβ1-42 plus cholera toxin. Reduction in cerebral amyloidosis was not associated with deleterious side effects including brain T cell infiltration or cerebral microhemorrhage.

    Previous studies (Schenk et al., 1999) showed that this antigen works in mice, but it then proved to be dangerous to humans. Intraperitoneal immunization with Aβ1-42 generates antibodies, particularly against the N-terminal of the peptide, the immuno-dominant region of the Aβ peptide; therefore, it is not necessary to use whole peptide, which may induce an inflammatory T cell response. On the basis of endogenous reactivity to Aβ in patients with AD, the use of a full-length Aβ peptide might be expected to lead to T cell-mediated CNS inflammatory effects. Increased T cell reactivity to Aβ was not observed preclinically in APP transgenic mice, perhaps due to their high levels of peripheral Aβ and the consequent induction of T cell tolerance (Monsonego et al., 2001).

    Improved immunotherapeutic strategies may be used to obtain beneficial effects without untoward side effects. It is possible to induce Aβ antibodies with no T cell response using small peptides containing dominant B cell epitopes that bind to Aβ plaques but no T cell epitopes; it will be equally possible to find the most convenient way of delivery.

    View all comments by Beka Solomon
  3. I would like to comment on the following statement from this News Article:

    "It is important to note that the original vaccine did not produce T cell infiltration in mice, either, yet still caused problems in some people."

    I wonder how good the evidence is supporting the idea that the original vaccine (or the new skin vaccine) does not produce T cell infiltration in mice. How many mice would have to be examined (and how carefully?) to rule out an effect that only occurred in ~6 percent of subjects in the AN1792 trial?

    View all comments by DAVID HAWVER
  4. We appreciate the comments of Drs. Lemere and Solomon. I'd like to comment on a few of the issues they raised.

    Regarding cerebral microhemorrhage, it is correct that this adverse event has been observed in past "active" Aβ vaccines that have been administered to old transgenic AD mice bearing established amyloid plaques. So, I agree with Dr. Lemere that lack of detection of these microhemorrhages is not altogether unexpected in our transcutaneous Aβ vaccine, which was administered prophylactically starting in younger AD mice. I also agree that it has been difficult to understand why Aβ immunization does not produce the aseptic meningoencephalitis in AD mice that was observed in about 6 percent of patients who received Elan's AN-1792 vaccine (unless pertussis toxin is used as the adjuvant—often employed to promote brain T cell infiltration in mouse models of multiple sclerosis). Ultimately, this may come down to differences in immune systems of mice versus humans. It may be necessary to use mice with "humanized" immune systems (currently under development by Dr. Richard Flavell and colleagues) to better model this aspect of the Aβ vaccine and allow appropriate steps to be taken to avoid this harmful adverse event.

    Regarding the use of N-terminal Aβ "B cell epitopes" in lieu of Aβ middle-region "T cell epitopes," it would be great if this approach ultimately works in humans (as has been shown in AD mice by Dr. Lemere), should it make it to clinical trials. However, I have my doubts. It seems that production of Aβ antibodies is the key step to producing therapeutic benefit, whether it is the ~0.1 percent of circulating Aβ antibodies that we know from the work of Schenk and colleagues to cross the blood-brain barrier in mice, or the circulating antibodies from the work of Holtzman and colleagues that may act as a "peripheral sink" to draw Aβ out of the brain and into the blood. If we use a non-T cell reactive portion of the Aβ peptide as an immunogen, B cells will not receive the T cell help necessary to elicit a fulminant Aβ antibody response—and by "fulminant" I mean IgM to IgG class-switching and copious production of long-lasting Aβ IgG antibodies.

    An approach that in my opinion is more likely to succeed would be promoting anti-inflammatory, cognate Th2 cell Aβ-specific immunity. Th2 cells are more efficient than Th1 cells at providing help to B cells, and are defined on the basis of anti-inflammatory cytokines such as IL-4 and IL-10 (as opposed to pro-inflammatory Th1 cells, which may be the T cell subset that mediated the aseptic meningoencephalitis in the Elan/Wyeth trial). So, Th2 cells (unlike Th1 cells) should not provoke a pro-inflammatory, auto-aggressive immune response. We’ve seen from our past studies, as well as in studies from Dr. Lemere’s group and others, that there are different ways to promote Th2 immune responses in the context of Aβ; for example, depending on antigen route of delivery, adjuvant used, and dosing schedule. In our recent transdermal Aβ vaccine approach, we have targeted skin-resident Langerhans cells, a unique population of innate immune cells that we believe are important in promoting Th2 immune responses to Aβ. Not only is the skin a minimally invasive, convenient route of delivery for the Aβ vaccine, but it is also an organ that offers unique immunomodulatory potential.

    View all comments by Terrence Town
  5. In this interesting paper, Nikolic and colleagues examined the efficacy of transcutaneous immunization (TCI) with fAβ42 and cholera toxin (CT) in induction of immune responses to Aβ and reducing cerebral amyloidosis in PSAPP mice without development of significant amyloid deposits at the start of immunization (protective vaccination).

    It is well known that the first immunotherapy clinical trial (AN-1792 vaccine) in AD patients was halted when ~6 percent of the participants developed aseptic meningoencephalitis. Case reports from AN-1792 trials suggest that the aseptic meningoencephalitis detected in 22 percent of the vaccine-responsive subgroup (59 individuals with antibody titers ³1:2200) might have been caused by a T cell-mediated autoimmune response (Nicoll et al., 2003; Ferrer et al., 2004; Nicoll et al., 2006), although Aβ-specific CD4+ or CD8+ T cells have never been directly demonstrated in the brains. Importantly, AN-1792 vaccine utilized fibrillar Aβ42 (fAβ42) containing the B and T cell “self epitopes” of this peptide as an immunogen, and a Th1 adjuvant, QS21, which was also implicated as a cause of the reported adverse effects.

    Our previous studies with mice also suggested that both adjuvant and T cell epitopes of Aβ42 might be critical to safe AD vaccine design (Cribbs et al., 2003). Accordingly, we suggested to circumvent the side effects of the AN-1792 vaccine and reduce the potential for T cell-mediated autoimmune toxicity in AD patients, using a vaccine composed of B cell antigenic determinants of Aβ42 fused with a “non-self” T helper cell epitope (Agadjanyan et al., 2005; Petrushina et al., 2007 submitted).

    While we and others (Seabrook et al., 2006; Seabrook et al., 2006) were trying to avoid autoreactive T cell responses, Nikolic et al. have decided to keep both B cell and T cell self-epitopes using whole fAβ42 peptide, but to use the CT adjuvant instead of QS21 in order to generate an effective and potentially safe AD vaccine. CT, produced by various strains of Vibrio cholerae, is an exceptionally potent and safe immunoadjuvant not only for rodents, but also for humans, despite the fact that it induces strong anti-toxin immunity (Glenn et al., 1998; 1998; 1999). It was shown that TCI with tetanus toxoid and CT induced systemic Th2 (anti-inflammatory)-type (Hammond et al., 2001) responses, whereas CT mixed with OVA induced both Th1 (inflammatory) and Th2 responses (Anjuere et al., 2003).

    Nikolic et al. analyzed the humoral immune responses at day 0 and at weeks 4, 8, 12, and 16 after TCI, and demonstrated that the first four injections were not inducing much antibody responses to self Aβ in PSAPP mice. It remains to be investigated why the first four, or maybe even five immunizations were not potent (many reports with other antigens + CT indicate that 2-3 TCI are very potent, and this is important for use of a vaccine in humans). After the sixth injection with fAβ42 and CT, mice generated very high titers of anti-Aβ42 antibodies (average titers jumped to ~120 mkg/ml ± SD of ELISA) that continued to increase after eight (~180 mkg/ml) and 10 (>200 mkg/ml) immunizations. Although the antibody response of individual mice is not reported by the authors, it is clear that average titers of >200 mkg/ml are extremely high for APP/Tg mice.

    Along with these high titers of antibodies, one should expect strong anti-Aβ42 T helper cell responses in immune animals. Although the authors did not investigate proliferation of CD4-positive T helper cells in PSAPP or wild-type mice, they did analyze production of IL2 as well as IL4 (anti-inflammatory) and IFNg (inflammatory) cytokines in cultures of splenocytes obtained from C57Bl6 (Fig 1 in manuscript) and PSAPP (data were not shown in the paper) mice immunized with fAβ42 plus CT or CT alone. The authors reported folds of increase, but not concentrations of these cytokines in splenocyte cultures. The authors showed that immunizations with fAβ42 and CT, but not CT alone, induced elevation of the levels of all three cytokines after in vitro re-stimulation of cultures with fAβ42. Of note, the very strong non-specific polyclonal stimulators Con A and anti-CD3 antibodies somehow induced only a twofold increase in production of all three cytokines in experimental and control splenocyte cultures.

    Taken together, these results suggest that the CT adjuvant stimulates systemic anti-Aβ Th1 and Th2 types of immune responses, and such activated T cells may recognize Aβ42 in brains of vaccinated animals and induce encephalomyelitis. To analyze this possibility, the authors tried to detect CD3 + T cells in brains of immune and control mice. They did not see lymphocyte infiltration in brains of TCI mice, but this is not surprising, because all previous studies (except one unconfirmed report with Aβ42 and pertussis toxin, i.e., Furlan, et al., 2003) also did not identify T cells in the brains of different APP/Tg mice, or wild-type animals immunized with Aβ antigen. Hopefully other AD animal models (dogs? monkeys?) could be used to show the safety of any candidate Aβ vaccine; otherwise, we will need again to go to clinical trials to demonstrate safety and efficacy of active immunization strategy in AD patients.

    Having shown high titers of anti-Aβ antibodies in PSAPP mice that at the start of immunization were ~4 months old, the authors evaluated amyloid pathology after 16 weeks and demonstrated that these antibodies reduced cerebral amyloidosis in 8-month-old mice. Importantly, anti-Aβ antibodies reduced the levels of not only insoluble, but also soluble, likely more toxic forms of cerebral Aβ detected by capture ELISA. This finding is of interest because it suggests that immunotherapy for AD might not only remove amyloid deposits, but also take out most toxic, oligomeric Aβ species from the brains of AD patients. Previously, other researchers have reported that in AD patients vaccinated with AN-1792 “parenchymal amyloid was focally disaggregated” and “total soluble amyloid levels were sharply elevated in vaccinated patient gray and white matter compared with AD cases.”

    Interestingly, our therapeutic vaccine induced on average >50 mkg/ml anti-Aβ antibodies in APP/Tg2576 mice (~9 months old at start of immunization), and these antibodies significantly reduced insoluble, but not soluble Aβ, including 6-, 9-, and 12-mer oligomeric species of Aβ, detected in aged ~19-month-old animals (Petrushina et al., 2007, submitted). Collectively, these results and data reported by Nikolic et al. suggest that either higher concentration of anti-Aβ antibodies are needed for significant reduction of oligomeric amyloid in brains of immunized mice, or immunization should be initiated earlier, in mice without pre-existing AD-like pathology (preventive vaccination versus therapeutic vaccination). Interestingly, data from the AN-1792 report (Patton et al., 2006) also suggest that “anti-amyloid immunization may be most effective not as therapeutic or mitigating measures, but as a prophylactic measure when Aβ deposition is still minimal.”

    View all comments by Michael G. Agadjanyan
  6. The paper by Nikolic and colleagues reports a transcutaneous method to vaccinate mice with Aβ and induce anti-Aβ antibodies. The authors found that there was an increase in anti-Aβ antibodies and a decrease in cerebral Aβ. Overall, this is an interesting study and is in agreement with our forthcoming paper in Neurobiology of Aging. In that study we also immunized mice using the transcutaneous route but used instead a short Aβ fragment containing the B cell epitope.

    However, the cytokine data shown in the current manuscript also shows an increase in Th1 type cells as seen by the 4 fold increase in IFN-γ and 5 fold increase in IL-2 compared to PBS stimulation. In addition, it is often the case that IL-2 levels are higher at earlier time points of stimulation such as 24 and 48 hours, thus the peak levels of this cytokine may have been missed. I share the concern expressed in other commentaries of the very low levels of cytokines induced by ConA. Splenocyte proliferation data would also be useful to demonstrate the specificity and magnitude of the T cell response. Together, these data demonstrate a shift towards a Th2 type phenotype but this is not exclusive.

    The quantification of the anti-Aβ antibody in the brain of approximately 0.05 percent following immunization could be explained by blood contamination following perfusion. It would be good to report the total amount of IgG detected in both Aβ/CT and CT immunized WT and PSAPP mice to allow a comparison to be made.

    Based on these data and our previous work, I believe that transdermal immunization may be an effective method of immunization. However, I do not believe that it can avoid the generation of Th1 cells in humans. Humans, unlike mouse strains, express a wide variety of HLA haplotypes, and some of them are likely to generate a Th1 response to full-length Aβ. This suggests that the use of a B cell epitope-based vaccine in conjunction with a Th2 biasing adjuvant and administration route will all be required to avoid the deleterious immune response seen in the previous AN-1792 trial.

    View all comments by Tim Seabrook
  7. We would like to respond to some of the issues raised by Drs. Hawver, Agadjanyan, and Seabrook.

    Regarding Dr. Hawver's point of evaluating T cells in the brains of mice that received the transcutaneous Aβ vaccine in our study, it is of course difficult to conclude that there are no T cells in the brains of these mice. We examined multiple brain sections from these immunized mice, and in parallel sections from positive control brains where mice had been induced with experimental autoimmune encephalomyelitis (day 20 after induction), a mouse model of multiple sclerosis. We easily detected T cells by CD3, CD4, and CD8 immunostaining in the latter, but not in the former. Of course, absence of proof does not constitute proof of absence, but we feel confident that there is not appreciable/significant infiltration of T cells in the brains of mice immunized transcutaneously with Aβ.

    In response to Dr. Agadjanyan’s comment regarding Aβ antibody responses in C57BL/6 mice after four immunizations with Aβ plus CT, we agree that humoral responses of these mice were not strong until after the first four immunizations. Actually, we detected Aβ antibodies at the time of the fourth immunization, which markedly increased thereafter. As to why it took four+ immunizations to achieve Aβ antibody responses versus other reports where two to three vaccinations produced humoral responses as Dr. Agadjanyan mentions, this could be due to differences in Aβ versus other immunogens, variations in how the transcutaneous vaccine was applied, or perhaps mouse strain differences.

    Drs. Agadjanyan and Seabrook raise the issue of the responses that we obtained with the non-specific stimulators ConA and CD3 antibody. Specifically, they comment that the responses that we obtained in Fig. 1D on IFN-γ, IL-2, and IL-4 cytokines were not very strong. We represented these data as fold increases over splenocytes cultured from PBS-immunized control mice. We chose to represent the data this way so that one could easily see putative enhanced cytokine release from Aβ plus CT-immunized mice and mice immunized with CT alone. We observed approximately two- to threefold more cytokines from the latter groups over the former. One would not expect that non-specific stimulation of splenocytes from the latter groups would produce a marked increase in cytokine over splenocytes from PBS-immunized mice. After all, these are non-specific mitogens; why would we expect to get a marked increase in cytokines from splenocytes cultured from Aβ plus CT or CT-immunized mice, when we are not specifically stimulating Aβ and/or CT-specific T cells? On the other hand, Aβ recall challenge does produce clear increases in cytokines in splenocytes cultured from Aβ plus CT-immunized mice.

    Regarding Dr. Seabrook’s comment of measuring splenocyte proliferation, we did not conduct this assay. However, given the increase in IL-2 production after Aβ recall stimulation in splenocytes from Aβ/CT t.c. immunized mice, we may expect increased proliferation responses (as IL-2 is thought to be sufficient for T cell expansion). We appreciate Dr. Seabrook’s comment that we also observed Th1 cytokines (IFN-γ and IL-2), albeit at lower levels than Th2-indicative IL-4 production, in splenocytes from Aβ/CT t.c. immunized mice. Our interpretation of this was that given 1) IL-4 production at higher levels in vitro and 2) predominantly IgG1 Aβ antibodies in vivo, we were observing a mostly Th2-type response. The observation of appreciable IL-2 and IFN-γ from these splenocytes may represent an in-vitro phenomenon, as we did not observe evidence of a Th1-mediated humoral response in vivo.

    We appreciate Dr. Agadjanyan’s comment regarding reduction in both soluble and insoluble Aβ species after Aβ/CT t.c. immunization—we were also struck by this finding and are beginning to think about a possible mechanism for this result.

    Regarding Dr. Seabrook’s comment regarding blood contamination that could account for the Aβ antibodies in the brain, it is possible. However, we examined apolipoprotein B (present in blood but not normally in brain) levels in brain and could not detect this (see supplemental Fig. 7). Also, we performed Perl’s Prussian blue stain for ferric ion-hemosiderin (to detect blood contamination in brain, see Fig. 4), and we did not observe signal either in or around cerebral vessels. So we do not have evidence that poor perfusion contributed to our detection of Aβ antibodies in the brains of Aβ/CT t.c. immunized mice. Dale Schenk originally detected about 0.1 percent of Aβ antibodies in the brain (Schenk et al., 1999), and we show about half of that using our t.c. approach.

    Dr. Seabrook also comments that he does not believe that t.c. immunization with full-length Aβ1-42 can avoid Th1 responses in humans because humans have more HLA haplotypes than do mice. While we agree with the latter, why would more HLA haplotypes in humans result in a Th1 response? A number of adjuvants have been used in humans, and some of them (e.g., alum, CT) consistently produce a Th2 response, irrespective of the peptide used for immunization or the HLA haplotype of the individual. Our belief is that the “danger signal” provided by the adjuvant and the immune cells targeted (i.e., route of administration) may be more important than the peptide itself in promoting Th1 versus Th2 responses. Of course, we won’t ultimately know until data are available—if ever—in humans using these different approaches. That being the case, the more approaches that are explored, the better the chance that one or more of them will ultimately work.

    View all comments by Jun Tan

This paper appears in the following:

News

  1. Immunotherapy Update: Toward Patches and Creams