Okura Y, Miyakoshi A, Kohyama K, Park IK, Staufenbiel M, Matsumoto Y.
Nonviral Abeta DNA vaccine therapy against Alzheimer's disease: long-term effects and safety.
Proc Natl Acad Sci U S A. 2006 Jun 20;103(25):9619-24.
Please login to recommend the paper.
To make a comment you must login or register.
The recent PNAS paper by Okura and colleagues provides interesting new data regarding a nonviral Aβ DNA vaccine. It is encouraging that testing with two of the three naked DNA plasmid vaccines (Aβ-Fc and IgL-Aβ) against Aβ1-42 significantly lowered cerebral Aβ in APP23 transgenic mice when given prior to Aβ deposition as well as later, when plaque deposition was well underway. As has been shown previously in 3xTg-AD mice by LaFerla's group, the authors in this paper showed clearance or prevention of intracellular Aβ in neurons in addition to extracellular plaques. Antibody titers were lower with the Aβ DNA vaccines than using full-length Aβ1-42/CFA (Complete Freund’s Adjuvant) but were enough to reduce CNS Aβ burden. T cells and macrophages were absent from the immunized mouse brains, indicating a lack of neuroinflammation; however, microglial and astrocytic labeling were apparently not investigated and would shed further light on the state of inflammation in the brains of immunized mice. Interestingly, T cells isolated from draining lymph nodes of APP23 mice immunized with either the Aβ DNA vaccines (without CFA but with bupivacaine, and adjuvant commonly used for DNA vaccines) or Aβ1-42 peptide with CFA did not proliferate in response to restimulation with Aβ1-42 peptide in vitro, whereas T cells from B6 wild-type mice responded strongly to immunization with Aβ1-42/CFA but not the Aβ DNA vaccines. The authors believe that tolerance to the high levels of human Aβ transgenic peptide may be responsible for the lack of Aβ recognition by T cells in APP23 mice. In particular, it is curious that immunization with Aβ1-42 and CFA did not cause T cell proliferation in the APP23 mice. This result contrasts with what we have found in 3x AD-like Tg mouse models (PSAPP, J20 APP, 3xTg-AD) using Aβ1-42 peptide and various routes of administration and adjuvants, including CFA (by intraperitoneal injection). In the Okura paper, the T cell studies were performed after 3 weeks of immunization, whereas our studies examined splenocytes after the full-term multi-month vaccine. It would be informative to examine T cells after the long-term Aβ DNA immunization, especially in older mice, although according to the paper, the results were the same after 5 months of immunization (data not shown). In addition, differences in Aβ peptide conformation, adjuvant, route of administration, and the genetic background of the mice examined may also influence the T cell reaction. The APP23 mice used here were on a C57BL/6 background, which may partially explain their somewhat suppressed immune response. However, 3xTg-AD mice are also on the same genetic background and show a strong proliferative response to immunization with full-length Aβ and CFA or a heat labile enterotoxin adjuvant called LT(R192G).
In the future, it will be of interest to know if these nonviral Aβ DNA vaccines have any effect on cerebrovascular Aβ (including the induction of microhemorrhage), cerebral gliosis, and behavior. Immunoglobulin isotyping may help determine the extent of T helper cell contribution to the overall immune response. Unfortunately, because Aβ immunized mice do not show the same adverse events as humans in the AN1792 trial, it will be difficult to predict exactly how humans will respond to such a vaccine. However, these earlier studies are encouraging and may represent a safer, cost-efficient strategy for Aβ vaccination. Testing these Aβ DNA vaccines in APP Tg mouse models in which immunization with full-length Aβ and CFA results in a strong T cell proliferative response will help tease out the safety of the vaccine.
Okura et al. demonstrated effectiveness of no viral Aβ DNA vaccine in reducing Aβ load in APP23 Tg mice. The vaccine was administered without an adjuvant and appears to elicit no cellular immunity. Therefore, it may be considered safer that AN1792, containing Aβ and QS-21 adjuvant, which caused exaggerated cellular response resulting in autoimmune encephalitis in 5 percent of study subjects. No signs of toxicity were observed in APP23 and wild-type mice which received the DNA vaccine from the age of 3 to 18 months. One has to remember, however, that in the original study of Aβ vaccination administered with Freund adjuvant, no toxicity was also observed in Tg mice and similarly no toxicity was observed in phase 1 of the human clinical trial. Although the safety concepts behind development of this nonviral DNA vaccine sound reasonable, they have not been fully evaluated. One way to evaluate long-term safety prior to clinical trials would be a study in aged primates. The risk of an autoimmune reaction is known to significantly increase with aging. Risks here would include the generation of anti-DNA antibodies and the genome integration risk associated with DNA vaccination, as well as the remaining risk of these vaccines, inducing cellular immunity in a small proportion of patients. So far, DNA vaccines have fared well in some model systems of disease; however, in human clinical trials the results have been disappointing. On the other hand, two DNA vaccines have recently been licensed in animals (horse and fish), highlighting the potential of this approach.
The PNAS paper by Okura et al. describes results regarding a DNA-based Alzheimer disease vaccine. Significant advantages of DNA immunization in comparison to other means include ease of vaccine development; high stability of preparation; capability of modifying genes encoding desired antigen(s); ability to target cellular localization of an antigen by means of adding or removing signal sequences or transmembrane domains; and the ability to selectively elicit the desired type of immune response (humoral or cellular).
Three years ago we reported (Ghochikyan et al., 2003) that immunization of B6SJLF1 mice with a DNA vaccine encoding the human Aβ42 gene generated robust anti-Aβ42 antibodies of predominantly IgG1 and IgG2b isotypes, and that these antibodies were capable of binding to β-amyloid plaques in brain tissue from AD cases. Thus, we demonstrated for the first time that DNA immunization is capable of inducing therapeutically potent anti-Aβ42 antibodies in wild-type mice. However, later we reported (Cribbs, 2005) that this prototype vaccine induced only low titers of anti-Aβ antibody in APP/Tg 2576 mice. These results were not totally unexpected, because previously we, and others, observed that immune responses to Aβ in APP/Tg animals were significantly impaired even after protein immunizations with fAβ42 (Monsonego et al., 2001; Petrushina et al., 2003). Other groups also attempted to generate anti-Aβ antibodies using DNA vaccines encoding the Aβ42 peptide (Schultz et al., 2004; Qu et al., 2004; Kutzler et al., 2005). Schultz et al. demonstrated that DNA encoding human Aβ42 (pAβ42) was not capable of inducing anti-Aβ antibodies even in wild-type mice. Only pAβ42 mixed with aggregated Aβ42 peptide slightly increased antibody production. The second group (Kutzler et al., 2005) demonstrated that pAβ42 was capable of inducing T cell proliferation in mice of different immune haplotypes as well as HLA Class II transgenic mice; however, these authors did not report generation of anti-Aβ antibodies. The last group (Qu et al., 2004) induced in one out of three wild-type mice significant titers of anti-Aβ antibodies only after immunization with plasmid encoding the Aβ42 dimer gene. These results were expected, because DNA vaccination is generally known to induce relatively low levels of antibody production and/or cellular immune responses in mice, and it is not always successful in large animals and humans.
Okura et al. are basically reporting in PNAS the same immunological data with DNA vaccine based on Aβ42, although they did not cite all these papers published and reported in different conferences within the last 4 years. More specifically, they are showing that multiple intramuscular immunizations (sometimes it is more than 20 injections) with DNA vaccines encoding the Aβ42 gene with signal peptide induced low titers of anti-Aβ antibodies (titers are 1,000-6,000). With these low titers of antibodies (exact concentration of antibodies, unfortunately, is not provided), it is not surprising that the authors did not detect immune lymph node (LN) cell proliferation specific to Aβ42 in APP/Tg mice. However, it should be mentioned that Aβ42 is a T-dependent antigen, because immunization of two types of nude mice with fibrillar Aβ42 peptide formulated in a strong conventional adjuvant did not induce any humoral responses (our unpublished data). Therefore, without CD4+T helper cell responses, APP23 mice couldn’t produce antibodies, so we still have to assume that DNA vaccine is inducing some anti-Aβ cellular responses that the authors likely did not detect because of the methodology they used (for example, incorporation of H3 thymidine was analyzed “after the first injection”; Stimulation Index is not reported at all; see also comments of Dr. Lemere about cellular responses).
Data from the AN1792 clinical trial indicate that vaccinated patients with anti-Aβ titer of 1:1,000 had very low Aβ-plaques in postmortem brain (see Nicoll et al., 2003 and ARF related news story), and that all AD patients that showed significant improvement in cognitive functions and activities of daily living, had titers of anti-Aβ antibodies 1:2,200 at any time after injections (Hock et al., 2003; Gilman et al., 2005; Fox et al., 2005; Bayer et al., 2005). Recently, using our epitope peptide vaccine approach (Agadjanyan et al., 2005), we demonstrated that concentrations of anti-Aβ antibody from ~1.5 to ~250 mg/ml are sufficient for clearing/reducing of AD-like neuropathology in APP/Tg2576 mice (paper submitted and data presented in the last AD/PD conference in Italy). So it was not totally unexpected that Okura et al. showed that vaccination with DNA encoding secreted Aβ42 reduced Aβ burden in both prophylactic and therapeutic protocols and “there was significant correlation between the serum anti-Aβ antibody titer and the reduction of amyloid depositions at 7 months and 9 months of age.” What was unexpected was that therapeutic vaccination induced a more profound response than prophylactic vaccination. This was unexpected because previous reports by several investigators showed that clearance of pre-existing Aβ plaques is much more difficult than the prevention of Aβ accumulation. Another unexpected result was data with the DNA vaccine targeting the non-secreted form of Aβ42. The authors mentioned that this K-Aβ vaccine was also effective. However, because the authors did not analyze immune responses after vaccination with K-Aβ, it is not clear whether antibodies or some other mechanism(s) are involved in the significant reduction of Aβ burden (Fig 2E). While these questions could be clarified in the future, we would like to focus our discussion on the safety of the DNA vaccine based on Aβ42 peptide.
As we mentioned above, and as shown by Okura et al., anti-Aβ antibody concentrations may be critical for clearance/reduction of AD-like neuropathology in the brains of AD patients. Thus, immunotherapy can only be effective if it will induce therapeutically relevant titers of anti-Aβ antibodies capable of removing toxic species of this peptide (oligomers, fibrils, monomers) from the brain of AD patients. The DNA vaccine generated by Okura et al. induced very low concentration of such antibodies in APP23 mice. Hence, it should be expected that the concentration of these antibodies will be even lower in AD patients, since it is well known that DNA vaccination is less immunogenic in humans. Thus, to enhance such humoral responses, authors will ultimately need to enhance anti-Aβ CD4+T helper cell responses in AD patients. I think that this is not only a difficult task, but also not a very safe one. Such an approach is not very safe because of the potential for T cell-mediated toxicity, as detected in AD patients immunized with AN1792 vaccine (Nicoll et al., 2003; Ferrer et al., 2004; Masliah, 2005). Unfortunately, the absence of neuroinflammation in the brains of DNA vaccinated APP23 mice is not a good argument that such a vaccine really is “highly secure and easily controllable,” because firstly, as the authors show vaccinated APP23 mice do not generate robust anti-Aβ42 T cell responses, and secondly, it is well known that the mouse model is a poor predictor of Aβ vaccination-induced adverse effects in humans. Thus, one needs to use an alternative approach to make a safe DNA vaccine. For example, one theoretical solution that circumvents the need for autoreactive T cells is to use a DNA epitope vaccine composed of the self-B cell antigenic determinant of Aβ42 and the non-self T cell antigenic determinant as we and others demonstrated with using a peptide-based approach (Agadjanyan et al., 2005; Maier et al., 2006). This strategy appears to eliminate the need for the stimulation of autoreactive T cells, while simultaneously allowing for potent Th2 support of B cell-mediated therapeutic anti-Aβ42 antibodies. Such an approach may avoid the generation of systemic inflammatory responses, which can be propagated to the CNS. This and other DNA vaccine approaches should be considered in the future to ensure the best possible strategy for the development of an AD vaccine.