Lambracht-Washington D, Qu BX, Fu M, Eagar TN, Stüve O, Rosenberg RN.
DNA beta-amyloid(1-42) trimer immunization for Alzheimer disease in a wild-type mouse model.
JAMA. 2009 Oct 28;302(16):1796-802.
PubMed.
I’d like to offer for the field’s collective consideration my view that this manuscript is flawed in several regards.
First, some background on DNA-based vaccination. A unique property of DNA-based vaccination is its ability to induce prolonged, endogenous antigen synthesis and processing within the subject’s own cells. DNA immunization has been shown to generate humoral and cellular immune responses against multiple viral, bacterial, and tumor antigens as well as against amyloid-β (Aβ) self-antigen (1) as was first shown by our group. This approach allows also the rational inactivation or removal of sequences encoding potentially toxic domains as well as the inclusion of molecular adjuvants, such as chemokines, cytokines, or co-stimulatory molecules that can direct T helper cell responses toward the desired pathway.
With regard to the requirements for an Alzheimer disease (AD) vaccine, DNA immunization exhibits several important advantages when compared to peptide-based AD vaccines or passive immunization with monoclonal antibodies. For example, DNA vaccines do not require the use of a conventional adjuvant, scale-up of manufacturing for clinical trials is readily achieved, and cGMP-grade DNA vaccines are significantly less expensive than peptide-based cGMP vaccines.
However, despite considerable promise demonstrated in animal models, DNA vaccines have generally exhibited low and inconsistent immune responses in clinical studies. There is consensus in the field that a key factor in this limitation is the low delivery efficiency of current DNA administration technologies. Two methods that can be used to increase the efficiency of DNA delivery into cells are the use of a gene gun (the only device for human use belongs to PowderMed/Pfizer) and electroporation (devices for human use belong to two companies, Ichor Medical Systems and Inovio Biomedical Corporation). Our mouse studies demonstrate that both gene gun- and electroporation-mediated delivery of DNA-based AD vaccines promote very strong antibody responses to Aβ in both wild-type and APP/Tg mouse models (1-5).
In this volume of JAMA, Lambracht-Washington et al. presented data demonstrating that multiple immunizations (the exact number is unclear from the manuscript but looks like up to 14 times) of B6SJLF1 wild-type mice with DNA encoding three copies of Aβ42 generated ~15 μg/ml antibodies. Basically, these authors repeated a pioneering study of Ghochikyan et al. (1), except they used a complicated and, in my opinion, debatable (see below) plasmid system instead of the single and effective plasmid used in our previous work published six years ago. Publication of this simple and arguably non-innovative paper in a prestigious medical journal with a 23.5 impact factor raises several questions in my mind, and I would try to address some here.
First, AD vaccine researchers want to understand if it is safe to generate B and T cell responses to a DNA vaccine encoding full-length Aβ42 peptide. It is well known that an AN1792 immunotherapy vaccine trial was halted prematurely because a subset of vaccinated, but not control, individuals developed meningoencephalitis. Postmortem examinations revealed that the neuroinflammation consisted primarily of CD4+ and/or CD8+T cells, implying that auto-reactive T cell responses to self-epitopes within the Aβ42 peptide were responsible for this serious adverse event. If that is the case, it is likely that any DNA vaccine encoding full-length Aβ42 may also generate auto-reactive T cells and induce inflammation in humans. In this study the authors claim that their DNA vaccine did not activate T cells specific to amyloid, but that raises the question of how the vaccinated mice managed to produce not high, but still substantial, humoral responses. Aβ42 is a T-dependent antigen (in our lab immunizations of nude mice of H2b and H2d haplotypes with fibrillar Aβ42 peptide formulated in CFA/IFA did not induce any humoral responses), and that is why without CD4+T helper cell responses, mice couldn’t produce antibodies. Thus, in order to move toward a clinical trial, the authors of this paper should clearly demonstrate the differences between the peptide antigen used in AN1792 and their DNA vaccine encoding three copies (probably one copy is not immunogenic) of full-length Aβ42. Another problem is that not every kind of animal model is suitable for testing the safety of an AD vaccine, since even AN1792 did not induce inflammation in various types of animals, including APP transgenic mouse models of AD. In addition, I think it is unlikely that the FDA or other regulatory agencies will allow another clinical trial based on full-length Aβ42 peptide immunization.
Another important issue is the plasmid that has been delivered by a gene-gun device approved only for animal use. The authors of this paper suggest that the double plasmid system (DPS) used in this study could enhance Aβ42 expression and trigger stronger immunity. However, as presented in the text, the authors generated only ~15 μg/ml antibodies (exact titers, individual variability and SD are not presented); this is much lower than the antibody titers (500 μg/ml) detected in mice vaccinated with Aβ42 formulated in Quil A adjuvant. Incidentally, using a single and effective pCMVE plasmid encoding an epitope AD vaccine we generated ~1,000 μg/ml and ~450 μg/ml of antibodies in wild-type and APP/Tg mice of H2b haplotypes, respectively (3). Thus, using such a complicated plasmid system is not necessary to generate a strong immune response. To assess the safety of this DPS, I respectfully submit that one should check the antibody and T cell responses not only to amyloid, but also to the GAL4 transcriptional factor. This foreign protein could be exposed to B cells from apoptotic cells transfected with DPS and almost certainly will be presented through MHC class I and II to the immune system of the host. In this experiment the host was the B6SJL strain of mice, but in case of a clinical trial it would be a human population with high MHC polymorphism.
Finally, the authors of this study are claiming that their DNA vaccine is inducing only the Th2-type of immune responses. It is likely that when using full-length Aβ42, one should aim to avoid pro-inflammatory (Th1) immune responses and induce a Th2 polarized anti-inflammatory response. The authors used an indirect measure of Th responses, analyzing Ig isotypes after AD vaccination as we suggested previously (6). Next, the authors rightly decided to measure cytokine production in splenocyte cultures from mice immunized with DNA or peptide. Unfortunately, the data of these experiments are presented only in the text of the manuscript in such a way that it is extremely difficult to understand the results of this study. There are no graphs, statistical analyses, and information about control groups of mice non-immunized, or immunized with irrelevant antigen. Without such clearly presented data, it is impossible to understand the immunogenic potency of DPS expressing full-length amyloid.
In conclusion, I would like to mention that a very recent assessment of the relationship between Aβ42 immune responses, degree of plaque removal, and long-term clinical outcomes demonstrated that immunization with Aβ42 resulted in clearance of amyloid plaques in the brains of AD patients, but did not prevent progressive neurodegeneration. These data suggest that removal of existing plaques alone is not sufficient to stop the progression of neurodegeneration and improve cognitive function. Another potential problem of AD immunotherapy in general could be the fact that a reduction of insoluble Aβ (plaques) may lead to increased levels of soluble forms of this peptide, primarily oligomers, the most toxic form for neurons, and impair cognitive function. Thus, I believe that to avoid these problems one should generate an AD vaccine that could be administered before significant AD brain pathology has accumulated. Passive transfer of anti-Aβ42 antibodies may be impractical for long-term preventative/early therapeutic application in AD because of the requirement for frequent intravenous dosing and the high cost of treatment. Thus, an ideal AD vaccine should be, first of all, extremely safe and, secondly, highly effective so it can be used prophylactically in high-risk subjects or therapeutically in early-stage AD. Unfortunately, this paper does not address these important questions.
See also:
Movsesyan, N. et al. Alzheimer's Disease DNA epitope vaccine induces equally strong humoral immune responses after delivery both via electroporation and gene gun. In prep.
References:
Ghochikyan A, Vasilevko V, Petrushina I, Movsesyan N, Babikyan D, Tian W, Sadzikava N, Ross TM, Head E, Cribbs DH, Agadjanyan MG.
Generation and characterization of the humoral immune response to DNA immunization with a chimeric beta-amyloid-interleukin-4 minigene.
Eur J Immunol. 2003 Dec;33(12):3232-41.
PubMed.
Davtyan H, Mkrtichyan M, Movsesyan N, Petrushina I, Mamikonyan G, Cribbs DH, Agadjanyan MG, Ghochikyan A.
DNA prime-protein boost increased the titer, avidity and persistence of anti-Abeta antibodies in wild-type mice.
Gene Ther. 2010 Feb;17(2):261-71.
PubMed.
Movsesyan N, Ghochikyan A, Mkrtichyan M, Petrushina I, Davtyan H, Olkhanud PB, Head E, Biragyn A, Cribbs DH, Agadjanyan MG.
Reducing AD-like pathology in 3xTg-AD mouse model by DNA epitope vaccine - a novel immunotherapeutic strategy.
PLoS One. 2008;3(5):e2124.
PubMed.
Movsesyan N, Mkrtichyan M, Petrushina I, Ross TM, Cribbs DH, Agadjanyan MG, Ghochikyan A.
DNA epitope vaccine containing complement component C3d enhances anti-amyloid-beta antibody production and polarizes the immune response towards a Th2 phenotype.
J Neuroimmunol. 2008 Dec 15;205(1-2):57-63.
PubMed.
Cribbs DH, Ghochikyan A, Vasilevko V, Tran M, Petrushina I, Sadzikava N, Babikyan D, Kesslak P, Kieber-Emmons T, Cotman CW, Agadjanyan MG.
Adjuvant-dependent modulation of Th1 and Th2 responses to immunization with beta-amyloid.
Int Immunol. 2003 Apr;15(4):505-14.
PubMed.
We appreciate the comment of Dr. Agadjanyan. We are aware of the work done by his group in the field of Aβ immunizations and recognize this in references 31 to 33.
Our manuscript underwent the regular peer review process and was deemed sufficiently innovative for publication. A novel approach in DNA vaccinations was applied using a double plasmid system with which we achieved not high, but effective antibody levels resulting from a non-inflammatory Th2 polarized immune response. One of the authors, Roger Rosenberg, as disclosed in the paper, is a co-inventor of a US patent application for “Aβ gene vaccines”. The fact that a patent was obtained is another indicator that our approach is innovative and relevant. Our group has made published contributions to this field of research since 2003 and we have presented the double plasmid approach at the ICAD meetings in 2008 and 2009.
The multiple copies of Aβ are linked together, making a trimeric protein rather than just expressing the single peptide three times. This is the most novel part of the paper in terms of a vaccine. Because B cells recognize antigens in solution and require receptor crosslinking for activation, multimeric epitopes are able to elicit B cell responses more potently than single epitopes. The double plasmid system, in which one plasmid encodes a yeast transcription factor, which drives the transcription of the Aβ42 sequence on the second plasmid, resulted in an about 10-fold stronger antibody production and improves the DNA Aβ42 trimer vaccine substantially. It is still much less compared to the peptide-immunized mice, but as we stated in the paper “a predominant Th2 response is a more important objective in developing an effective and safe therapy for AD than increased antibody levels alone”. It must be noted that T cell help is obviously provided in this model because isotype switching has occurred.
A major finding of this paper is that despite multiple immunizations T cell responses are undetectable. T cell responses are likely to be the cause of the encephalitis found in the clinical trial of peptide vaccination.
As stated by Dr. Agadjanyan, Aβ42 is a T cell-dependent antigen. Is it safe to generate a T and a B cell response by a DNA vaccine encoding full-length Aβ42? First, to have an effective immune response, both T and B lymphocytes are needed, as both strongly influence each other. In our manuscript, we do not state that our vaccine does not activate T cells. Instead, we say that T cell have been involved in the early immunizations process, as it is clearly indicated by the isotype switch of the respective Aβ42 antibodies produced. However, we could not detect in-vitro T cell activation or proliferation against Aβ42 as documented by appropriate detection assays. Since we did not find a prolonged T cell response, we conclude this type of vaccine is probably safe for use in humans. We are currently performing experiments to characterize early T cell responses and to determine T cell fate longitudinally. Second, the exact B and T cell epitopes in humans have not yet been clearly defined, as most of the work to date is done in mouse strains. Thus, we currently have insufficient data to avoid certain components of the Aβ42 peptide (for example: the T cell epitopes in the mouse) in the development of an effective vaccine for human AD patients.
Sufficient statistical analyses are provided in our manuscript. Every data point consists of mean and standard deviation. Additional information on the number of animals and the statistical assessment are provided in the article.
We agree with Dr. Agadjanyan that there is ongoing scientific discussion regarding the relevance of the plaque hypothesis. Removal of established plaques did not ameliorate or prevent progression of the disease in one clinical trial. Although the plaque count was reduced dramatically in these patients, there was evidence of progressive cognitive decline (Holmes et al., 2008). While it is very challenging to study these two seemingly contradictory observations in human patients, intuitively one would speculate that therapy was initiated too late when irreversible progressive and independent pathologic changes had already occurred. The greatest challenge for investigators who attempt to prevent clinical and pathological aspects of AD by DNA immunization or other interventional means will be the identification of biomarkers that will allow an early therapy. In summary, we think that our manuscript addresses a number of important key questions and we welcome further discussion.
References:
Holmes C, Boche D, Wilkinson D, Yadegarfar G, Hopkins V, Bayer A, Jones RW, Bullock R, Love S, Neal JW, Zotova E, Nicoll JA.
Long-term effects of Abeta42 immunisation in Alzheimer's disease: follow-up of a randomised, placebo-controlled phase I trial.
Lancet. 2008 Jul 19;372(9634):216-23.
PubMed.
The development of innovative vaccination strategies for Alzheimer disease is of utmost importance. In this article, a novel and interesting DNA-vaccine approach with two plasmids is described. Unfortunately, the antibody response raised in the non-transgenic mice is still fairly low, with OD-values ~0.5 when plasma samples are diluted only 1:500 (Fig. 3). One of the main difficulties with a DNA vaccine is to raise a significant immune response in humans or primates, but this problem is not addressed. The description of experimental procedures lack important details and their findings are presented in a somewhat incomplete manner. For instance, in Fig. 4, splenocytes from only a few mice were restimulated, and then only with the Aβ42 peptide. The authors conclude that a T cell response is undetectable. The obvious control experiment would have been to restimulate with the trimeric Aββ42 peptide. It is unclear if the splenocyte cultures contained only T cells, or were a mixture of T and B cells. If so, it would have been interesting to investigate if the isotype profile of the humoral response in blood samples (in Fig. 2) could have been replicated when cultured cells were restimulated. We regard this article as potentially interesting, but still very preliminary. Efficacy and safety experiments in transgenic mice and/or primates will be needed to evaluate this novel DNA vaccine strategy before it can be clinically tested.
I'd like to thank Dr. Lambracht-Washington et al. for their reply and respond to a few points.
Patents for specific plasmids are fairly routinely granted. Such a patent is not in and of itself proof of scientific innovation on the present challenges of Alzheimer disease vaccine design. The T cell epitope of amyloid in humans has been reported by Monsonego et al. (1); the B cell epitope in humans is very well described by the Elan group (2), and our group has characterized the mouse T cell epitope (3).
I agree with the authors' claim that the novel part of their study would be the trimeric protein. If so, the study would be well advised to focus its data and discussion around this new aspect. However, the paper contains no data characterizing the form of peptide expressed from their plasmid. There are no data to show that multiple copies of Aβ are expressed linked together into a trimer rather than three times as single peptides. The presumed trimeric antigen is not characterized.
References:
Monsonego A, Zota V, Karni A, Krieger JI, Bar-Or A, Bitan G, Budson AE, Sperling R, Selkoe DJ, Weiner HL.
Increased T cell reactivity to amyloid beta protein in older humans and patients with Alzheimer disease.
J Clin Invest. 2003 Aug;112(3):415-22.
PubMed.
Lee M, Bard F, Johnson-Wood K, Lee C, Hu K, Griffith SG, Black RS, Schenk D, Seubert P.
Abeta42 immunization in Alzheimer's disease generates Abeta N-terminal antibodies.
Ann Neurol. 2005 Sep;58(3):430-5.
PubMed.
Cribbs DH, Ghochikyan A, Vasilevko V, Tran M, Petrushina I, Sadzikava N, Babikyan D, Kesslak P, Kieber-Emmons T, Cotman CW, Agadjanyan MG.
Adjuvant-dependent modulation of Th1 and Th2 responses to immunization with beta-amyloid.
Int Immunol. 2003 Apr;15(4):505-14.
PubMed.
The study by Lambracht-Washington and coworkers explores the possibility of using a DNA-based Aβ1-42 trimer vaccine to promote Th2-type, Aβ1-42-specific immune response in wild-type mice. As detailed above in comments from Michael Agadjanyan, this work extends from pioneering studies that were previously conducted in the area of DNA-based vaccination against Alzheimer’s. Those past studies, a number of which were conducted in the Agadjanyan and Cribbs labs, definitively established that DNA-based Aβ vaccines were efficacious in reducing cerebral amyloidosis in transgenic Alzheimer’s mouse models (for a review, see Town, 2009). I agree with Michael that the novelty in the present study is primarily owing to use of a trimeric Aβ1-42 vaccine administered via a gene-gun delivery system, and secondarily, due to utilization of a dual plasmid system that relies on GAL4 transactivation of the 12.2 kD Aβ1-42 trimer antigen.
I would like to add a few comments regarding this manuscript. It seems that the amount of anti-Aβ1-42 antibodies is rather low, even after a prime-boost regimen of up to eight total DNA vaccinations. For instance, it appears that DNA-vaccinated mice produced ~fivefold less antibodies against full-length anti-Aβ1-42 versus peptide-vaccinated mice (and that is a conservative estimate assuming linearity of OD readings for antibody titres; see Fig. 3) and ~five- to sixfold reduced stimulation of T cell proliferation following Aβ1-42 splenocyte challenge compared to peptide immunization (see Fig. 4). Because the authors only chose to examine wild-type mice, it is unclear whether this relatively weak Aβ1-42 antibody response would be efficacious, and thereby translate to reduced cerebral amyloidosis in mice, let alone in humans. It is possible that the trimeric Aβ1-42 antigen is not optimal for generating antibody responses to synthetic Aβ1-42; however, the authors do not seem to report results from trimeric Aβ1-42 splenocyte recall stimulation experiments, which may have resulted in stronger anti-Aβ antibody responses.
I’d like to build on a comment that Michael made regarding assessment of Aβ immunotherapy safety. In the “Comment” section of this article, the authors state that they can assess safety of the DNA-based approach by comparing immune responses between DNA-based trimeric vaccination and peptide-based immunization conducted side-by-side. This is flawed logic. With the exception of one unconfirmed report that utilized pertussis toxin in combination with “standard” Aβ immunotherapy using Aβ1-42 plus CFA/IFA (Furlan et al., 2003), none of the Aβ immunotherapy approaches have produced aseptic meningoencephalitis—nor any other severe adverse event—in mice. Because of this, it is not possible to make any conclusions about safety by simply assessing Th1 or Th2 immune responses after DNA Aβ1-42 trimer vaccination in mice.
Finally, I would like to call the authors’ attention to our previous work, which was one of the first demonstrations that “standard” Aβ1-42 immunotherapy (as originally developed by Schenk et al., 1999) primarily produces a Th2 response in mice as determined by 1) primarily IgG1 anti-Aβ1-42 antibodies that recognized amino acids 1-12 of the peptide (now widely regarded as the B cell epitope), and 2) primarily Th2 cytokines both in vivo in mouse plasma and ex vivo in splenocyte assays (Town et al., 2001; Town et al., 2002). More recently, we have shown that changing the route of vaccination to transcutaneous delivery results in copious amounts of anti-Aβ antibodies, a primarily Th2-type immune response, and clearance of cerebral amyloid (Nikolic et al., 2007). Thus, it is possible to generate Th2 immune responses in mice using simpler, peptide-based approaches to Aβ immunotherapy. The key challenges are to translate these approaches into the clinic by devising strategies that are both safe and efficacious.
References:
Furlan R, Brambilla E, Sanvito F, Roccatagliata L, Olivieri S, Bergami A, Pluchino S, Uccelli A, Comi G, Martino G.
Vaccination with amyloid-beta peptide induces autoimmune encephalomyelitis in C57/BL6 mice.
Brain. 2003 Feb;126(Pt 2):285-91.
PubMed.
Nikolic WV, Bai Y, Obregon D, Hou H, Mori T, Zeng J, Ehrhart J, Shytle RD, Giunta B, Morgan D, Town T, Tan J.
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.
Schenk D, Barbour R, Dunn W, Gordon G, Grajeda H, Guido T, Hu K, Huang J, Johnson-Wood K, Khan K, Kholodenko D, Lee M, Liao Z, Lieberburg I, Motter R, Mutter L, Soriano F, Shopp G, Vasquez N, Vandevert C, Walker S, Wogulis M, Yednock T, Games D, Seubert P.
Immunization with amyloid-beta attenuates Alzheimer-disease-like pathology in the PDAPP mouse.
Nature. 1999 Jul 8;400(6740):173-7.
PubMed.
Town T.
Alternative Abeta immunotherapy approaches for Alzheimer's disease.
CNS Neurol Disord Drug Targets. 2009 Apr;8(2):114-27.
PubMed.
Town T, Tan J, Sansone N, Obregon D, Klein T, Mullan M.
Characterization of murine immunoglobulin G antibodies against human amyloid-beta1-42.
Neurosci Lett. 2001 Jul 13;307(2):101-4.
PubMed.
Town T, Vendrame M, Patel A, Poetter D, Delledonne A, Mori T, Smeed R, Crawford F, Klein T, Tan J, Mullan M.
Reduced Th1 and enhanced Th2 immunity after immunization with Alzheimer's beta-amyloid(1-42).
J Neuroimmunol. 2002 Nov;132(1-2):49-59.
PubMed.
Clearly, more work needs to be accomplished to define the immune response with Aβ42 immunization in both experimental animal models and in patients, but we are dedicated to proceeding in this direction.
Two related hypotheses drive our research: First, the generation of an effective anti-Aβ42 B cell response that will result in an antibody-mediated clearance of amyloid plaques, leading to a reduction in AD severity. Second, from a bio-safety standpoint, the generation of an anti-Aβ42 Th1 effector autoimmune response may lead to immune mediated pathology. These hypotheses are based on research by many others, including the clinical trial using peptide vaccination. We have sought to develop a therapy that will generate an effective antibody response and will lack measurable Aβ42-specific Th1 T cell responses. At this point, the definition of "effective" may not be directly related to Ig titer but is rather a biological outcome-based definition, as we have detected clearance of plaque with low titers previously and there is no method currently available to test CSF or CNS titers which are more relevant than serum (1,2). We specifically chose to test these hypotheses in non-transgenic mice in order to provide the most stringent possible test for T cell responses, as these mice have no thymic negative selection against human Aβ42. Therefore, if the potential existed for the development of a Th1 response, it should be detectable in this model (null hypothesis). Therefore, we compared the immune responses in wild-type mice against DNA and Aβ42 peptide immunization.
This direct comparison is innovative and constructive, as the results show a clearly different immune response with a predominant Th2 immune response with DNA Aβ42 immunization (Th2/Th1 ratio 10) and a mixed immune response with Aβ42 peptide (Th2/Th1 ratio 1). Furthermore, T cell proliferation (as determined by the stimulation indices) following DNA versus peptide immunizations showed again very different outcomes with non-reactive T cells after re-stimulation with Aβ42 peptide after DNA Aβ42 immunization and reactive T cells after Aβ42 peptide immunization.
These are important findings and strongly support proceeding ahead with a clinical trial using DNA Aβ42 immunization in patients. Additional studies using this construct in APP transgenic mice are ongoing as well as further characterization of the trimeric Aβ42 peptide as the new and innovative antigen in our system. We are aware of the challenge that the outcome in humans may or may not be the same as the findings in mice, but our study offers a valid explanation as to why the clinical trial with Aβ42 peptide immunization had caused encephalitis in 6 percent of the patients. We agree with Terrence Town, that “none of the Aβ immunotherapy approaches have produced aseptic meningoencephalitis nor any other severe adverse event in mice,” but our data are in strong support that DNA Aβ42 immunization does offer a potentially lower risk for adverse effects in patients. Because the meningoencephalitis was likely due an inflammatory Th1 immune response caused by the type of adjuvant used, we agree that it is possible to generate an non-inflammatory immune response using simpler, peptide-based approaches to Aβ immunotherapy, but we would like to cite a statement by Michael Agadjanyan in the first comment to our paper: “scale-up of manufacturing for clinical trials is readily achieved, and cGMP-grade DNA vaccines are significantly less expensive than peptide-based cGMP vaccines.”
These discussions with our colleagues are constructive and help to provide focus for future research.
References:
Qu B, Boyer PJ, Johnston SA, Hynan LS, Rosenberg RN.
Abeta42 gene vaccination reduces brain amyloid plaque burden in transgenic mice.
J Neurol Sci. 2006 May 15;244(1-2):151-8.
PubMed.
Qu BX, Xiang Q, Li L, Johnston SA, Hynan LS, Rosenberg RN.
Abeta42 gene vaccine prevents Abeta42 deposition in brain of double transgenic mice.
J Neurol Sci. 2007 Sep 15;260(1-2):204-13.
PubMed.
I’d like to add a few points to this discussion. I agree that the double plasmid system with GLA4 Activator, UAS/AB42 Trimer responder is quite innovative; however, whether this approach has translational potential for a human clinical trial is doubtful based on the data provided in the current manuscript.
I found it interesting that the wild-type mice used in this study were female B6SJLF1/J mice, which is the background of the Tg2576 mice that show immune hypo-responsiveness to immunization with Aβ42 (1). No data were provided in the current manuscript using APP transgenic mice to show that the antibody response induced by the double plasmid system was capable of attenuating amyloid deposition or improving behavioral measures. This raises the issue of immune tolerance that needs to be broken in order to induce an adequate antibody response to a “self” peptide or protein. In addition, because the target population for anti-Aβ immunotherapy is the elderly, overcoming immunosenescence is a formidable hurdle as well. The generally low titers and the low number of positive responders in the AN1792 clinical trial (I say that with the proviso that the trial was halted before the patients completed the protocol) illustrate the difficulties facing future active immunization trials in this elderly patient population. Therefore, inducing robust anti-Aβ antibody responses in APP transgenic mouse models, as well as large animal models, will likely be a prerequisite before moving forward into AD patients. In fact, we have previously published a study in elderly dogs that used fibrillar Aβ42 and alum as a Th2 adjuvant approved for human use; this induced therapeutic levels of anti-Aβ antibody titers, sufficient to clear amyloid deposits, in all of the immunized canines (2).
Other commentators on Alzforum have already pointed out the large number of immunizations required to induce rather low titers of anti-Aβ antibodies in this study, as well as the significant potential to induce an anti-GLA4 protein immune response. The latter would likely result in the loss of cells expressing the GLA4 protein, thereby eliminating the expression of the Aβ42 trimer immunogen. I would like to emphasize the fact that just because you fail to measure a T cell response from immunized mice does not mean that there was no response. Typically a T cell response is required to get Ig class switching from IgM to the IgG isotypes; Lambracht-Washington et al. were able to measure IgG1 and IgG2a isotypes, thus indicating that there was a T cell response to the Aβ42 trimer immunogen. Thus, the problem was likely due to the weak T cell response that was induced with the double plasmid system, thereby making it difficult to detect in the re-stimulated splenocyte cultures, which was also reflected in the rather weak antibody response to the double plasmid system.
A second major problem is the large number of immunizations needed to get the anti-Aβ antibody titers to 15 ug/ml. In fact, the data in Figure 3 do not appear to show any difference between the mice (n = 4) that received six immunizations and those (n = 4) that received 14 immunizations if you compare the antibody binding in panels A and B, because all of the plasma samples were diluted 1:500. Multiple immunization protocols—especially DNA immunization protocols requiring 6-14 injections that also require specialized equipment, such as electroporation, to promote adequate expression levels of the antigen to induce therapeutic levels of antibodies—are probably not a viable approach for translation to human clinical trials. Therefore, inclusion of some sort of adjuvant in the vaccine design to enhance the antibody titers and reduce the number of immunizations, either as a fusion construct or as a supplement (co-injection or transcutaneous delivery) to the immunogen will be required.
References:
Monsonego A, Maron R, Zota V, Selkoe DJ, Weiner HL.
Immune hyporesponsiveness to amyloid beta-peptide in amyloid precursor protein transgenic mice: implications for the pathogenesis and treatment of Alzheimer's disease.
Proc Natl Acad Sci U S A. 2001 Aug 28;98(18):10273-8.
PubMed.
Head E, Pop V, Vasilevko V, Hill M, Saing T, Sarsoza F, Nistor M, Christie LA, Milton S, Glabe C, Barrett E, Cribbs D.
A two-year study with fibrillar beta-amyloid (Abeta) immunization in aged canines: effects on cognitive function and brain Abeta.
J Neurosci. 2008 Apr 2;28(14):3555-66.
PubMed.
We thank David Cribbs for his comment, and we would like to correct a few things. First of all, we did not show the results from 14 immunizations in our mice. We immunized one group receiving a total of six DNA immunizations and a second group which received a total of eight DNA Aβ42 trimer immunizations. The resulting antibody titers are 10-fold higher than antibody titers from a previous published immunization construct from our group which was an Aβ42 monomer. For immunizations with the monomer, we have shown an effective reduction in plaque load in an APP transgenic mouse model. Therefore, we conclude that the new DNA Aβ42 trimer is very likely to cause an even better or at least equal reduction in overall plaque load and is thus effective and sufficient to cause a therapeutic effect. Further studies using DNA Aβ42 trimer immunization in transgenic mouse models are underway. Again, we would like to emphasize that an initial T cell response was observed by the isotype switching of the respective Aβ42 specific antibodies, but then the T cell response disappeared, not due to weak T cell response but due to a Th2 response which is non-inflammatory. The reason why we choose B6SJLF1 mice was indeed because this strain is the genetic background for a number of APP transgenic mouse strains. B6SJLF1 is a good responder in Aβ42 immunizations, as it carries the mixed background of C57BL6 which is poorly responding to Aβ42 immunization but has a genetic background promoting brain plaque development and SJL which has a great Aβ42 immune response.
I would like to respond to a reply made above by Doris Lambracht-Washington concerning the Th2 response issue raised by David Cribbs. Her reply was “Again, we would like to emphasize that an initial T cell response was observed by the isotype switching of the respective Aβ42 specific antibodies, but then the T cell response disappeared, not due to weak T cell response but due to a Th2 response which is non-inflammatory.”
The type of T cell response, i.e., Th1 vs. Th2, is not defined in any way by the duration or persistence of the response. These Th profiles are defined based on the types of cytokines produced; e.g., Th1 responses are primarily associated with interferon-gamma and interleukin-12, whereas Th2 responses occur with interleukin-4 and interleukin-10 production, amongst others such as interleukin-15. Thus, the Th2 response to trimeric Aβ antigen observed by Lambracht-Washington and coworkers would not, by its very nature, have a limited duration because it is anti-inflammatory.
Furthermore, it is a misconception that Th2 responses are always “non-inflammatory.” A number of immune disorders, including allergy and atopic disorders such as asthma are driven by overly aggressive Th2 responses.
Comments
Vice President, The Institute for Molecular Medicine
I’d like to offer for the field’s collective consideration my view that this manuscript is flawed in several regards.
First, some background on DNA-based vaccination. A unique property of DNA-based vaccination is its ability to induce prolonged, endogenous antigen synthesis and processing within the subject’s own cells. DNA immunization has been shown to generate humoral and cellular immune responses against multiple viral, bacterial, and tumor antigens as well as against amyloid-β (Aβ) self-antigen (1) as was first shown by our group. This approach allows also the rational inactivation or removal of sequences encoding potentially toxic domains as well as the inclusion of molecular adjuvants, such as chemokines, cytokines, or co-stimulatory molecules that can direct T helper cell responses toward the desired pathway.
With regard to the requirements for an Alzheimer disease (AD) vaccine, DNA immunization exhibits several important advantages when compared to peptide-based AD vaccines or passive immunization with monoclonal antibodies. For example, DNA vaccines do not require the use of a conventional adjuvant, scale-up of manufacturing for clinical trials is readily achieved, and cGMP-grade DNA vaccines are significantly less expensive than peptide-based cGMP vaccines.
However, despite considerable promise demonstrated in animal models, DNA vaccines have generally exhibited low and inconsistent immune responses in clinical studies. There is consensus in the field that a key factor in this limitation is the low delivery efficiency of current DNA administration technologies. Two methods that can be used to increase the efficiency of DNA delivery into cells are the use of a gene gun (the only device for human use belongs to PowderMed/Pfizer) and electroporation (devices for human use belong to two companies, Ichor Medical Systems and Inovio Biomedical Corporation). Our mouse studies demonstrate that both gene gun- and electroporation-mediated delivery of DNA-based AD vaccines promote very strong antibody responses to Aβ in both wild-type and APP/Tg mouse models (1-5).
In this volume of JAMA, Lambracht-Washington et al. presented data demonstrating that multiple immunizations (the exact number is unclear from the manuscript but looks like up to 14 times) of B6SJLF1 wild-type mice with DNA encoding three copies of Aβ42 generated ~15 μg/ml antibodies. Basically, these authors repeated a pioneering study of Ghochikyan et al. (1), except they used a complicated and, in my opinion, debatable (see below) plasmid system instead of the single and effective plasmid used in our previous work published six years ago. Publication of this simple and arguably non-innovative paper in a prestigious medical journal with a 23.5 impact factor raises several questions in my mind, and I would try to address some here.
First, AD vaccine researchers want to understand if it is safe to generate B and T cell responses to a DNA vaccine encoding full-length Aβ42 peptide. It is well known that an AN1792 immunotherapy vaccine trial was halted prematurely because a subset of vaccinated, but not control, individuals developed meningoencephalitis. Postmortem examinations revealed that the neuroinflammation consisted primarily of CD4+ and/or CD8+T cells, implying that auto-reactive T cell responses to self-epitopes within the Aβ42 peptide were responsible for this serious adverse event. If that is the case, it is likely that any DNA vaccine encoding full-length Aβ42 may also generate auto-reactive T cells and induce inflammation in humans. In this study the authors claim that their DNA vaccine did not activate T cells specific to amyloid, but that raises the question of how the vaccinated mice managed to produce not high, but still substantial, humoral responses. Aβ42 is a T-dependent antigen (in our lab immunizations of nude mice of H2b and H2d haplotypes with fibrillar Aβ42 peptide formulated in CFA/IFA did not induce any humoral responses), and that is why without CD4+T helper cell responses, mice couldn’t produce antibodies. Thus, in order to move toward a clinical trial, the authors of this paper should clearly demonstrate the differences between the peptide antigen used in AN1792 and their DNA vaccine encoding three copies (probably one copy is not immunogenic) of full-length Aβ42. Another problem is that not every kind of animal model is suitable for testing the safety of an AD vaccine, since even AN1792 did not induce inflammation in various types of animals, including APP transgenic mouse models of AD. In addition, I think it is unlikely that the FDA or other regulatory agencies will allow another clinical trial based on full-length Aβ42 peptide immunization.
Another important issue is the plasmid that has been delivered by a gene-gun device approved only for animal use. The authors of this paper suggest that the double plasmid system (DPS) used in this study could enhance Aβ42 expression and trigger stronger immunity. However, as presented in the text, the authors generated only ~15 μg/ml antibodies (exact titers, individual variability and SD are not presented); this is much lower than the antibody titers (500 μg/ml) detected in mice vaccinated with Aβ42 formulated in Quil A adjuvant. Incidentally, using a single and effective pCMVE plasmid encoding an epitope AD vaccine we generated ~1,000 μg/ml and ~450 μg/ml of antibodies in wild-type and APP/Tg mice of H2b haplotypes, respectively (3). Thus, using such a complicated plasmid system is not necessary to generate a strong immune response. To assess the safety of this DPS, I respectfully submit that one should check the antibody and T cell responses not only to amyloid, but also to the GAL4 transcriptional factor. This foreign protein could be exposed to B cells from apoptotic cells transfected with DPS and almost certainly will be presented through MHC class I and II to the immune system of the host. In this experiment the host was the B6SJL strain of mice, but in case of a clinical trial it would be a human population with high MHC polymorphism.
Finally, the authors of this study are claiming that their DNA vaccine is inducing only the Th2-type of immune responses. It is likely that when using full-length Aβ42, one should aim to avoid pro-inflammatory (Th1) immune responses and induce a Th2 polarized anti-inflammatory response. The authors used an indirect measure of Th responses, analyzing Ig isotypes after AD vaccination as we suggested previously (6). Next, the authors rightly decided to measure cytokine production in splenocyte cultures from mice immunized with DNA or peptide. Unfortunately, the data of these experiments are presented only in the text of the manuscript in such a way that it is extremely difficult to understand the results of this study. There are no graphs, statistical analyses, and information about control groups of mice non-immunized, or immunized with irrelevant antigen. Without such clearly presented data, it is impossible to understand the immunogenic potency of DPS expressing full-length amyloid.
In conclusion, I would like to mention that a very recent assessment of the relationship between Aβ42 immune responses, degree of plaque removal, and long-term clinical outcomes demonstrated that immunization with Aβ42 resulted in clearance of amyloid plaques in the brains of AD patients, but did not prevent progressive neurodegeneration. These data suggest that removal of existing plaques alone is not sufficient to stop the progression of neurodegeneration and improve cognitive function. Another potential problem of AD immunotherapy in general could be the fact that a reduction of insoluble Aβ (plaques) may lead to increased levels of soluble forms of this peptide, primarily oligomers, the most toxic form for neurons, and impair cognitive function. Thus, I believe that to avoid these problems one should generate an AD vaccine that could be administered before significant AD brain pathology has accumulated. Passive transfer of anti-Aβ42 antibodies may be impractical for long-term preventative/early therapeutic application in AD because of the requirement for frequent intravenous dosing and the high cost of treatment. Thus, an ideal AD vaccine should be, first of all, extremely safe and, secondly, highly effective so it can be used prophylactically in high-risk subjects or therapeutically in early-stage AD. Unfortunately, this paper does not address these important questions.
See also:
Movsesyan, N. et al. Alzheimer's Disease DNA epitope vaccine induces equally strong humoral immune responses after delivery both via electroporation and gene gun. In prep.
References:
Ghochikyan A, Vasilevko V, Petrushina I, Movsesyan N, Babikyan D, Tian W, Sadzikava N, Ross TM, Head E, Cribbs DH, Agadjanyan MG. Generation and characterization of the humoral immune response to DNA immunization with a chimeric beta-amyloid-interleukin-4 minigene. Eur J Immunol. 2003 Dec;33(12):3232-41. PubMed.
Davtyan H, Mkrtichyan M, Movsesyan N, Petrushina I, Mamikonyan G, Cribbs DH, Agadjanyan MG, Ghochikyan A. DNA prime-protein boost increased the titer, avidity and persistence of anti-Abeta antibodies in wild-type mice. Gene Ther. 2010 Feb;17(2):261-71. PubMed.
Movsesyan N, Ghochikyan A, Mkrtichyan M, Petrushina I, Davtyan H, Olkhanud PB, Head E, Biragyn A, Cribbs DH, Agadjanyan MG. Reducing AD-like pathology in 3xTg-AD mouse model by DNA epitope vaccine - a novel immunotherapeutic strategy. PLoS One. 2008;3(5):e2124. PubMed.
Movsesyan N, Mkrtichyan M, Petrushina I, Ross TM, Cribbs DH, Agadjanyan MG, Ghochikyan A. DNA epitope vaccine containing complement component C3d enhances anti-amyloid-beta antibody production and polarizes the immune response towards a Th2 phenotype. J Neuroimmunol. 2008 Dec 15;205(1-2):57-63. PubMed.
Cribbs DH, Ghochikyan A, Vasilevko V, Tran M, Petrushina I, Sadzikava N, Babikyan D, Kesslak P, Kieber-Emmons T, Cotman CW, Agadjanyan MG. Adjuvant-dependent modulation of Th1 and Th2 responses to immunization with beta-amyloid. Int Immunol. 2003 Apr;15(4):505-14. PubMed.
University of Texas Southwestern Medical Center at Dallas
We appreciate the comment of Dr. Agadjanyan. We are aware of the work done by his group in the field of Aβ immunizations and recognize this in references 31 to 33.
Our manuscript underwent the regular peer review process and was deemed sufficiently innovative for publication. A novel approach in DNA vaccinations was applied using a double plasmid system with which we achieved not high, but effective antibody levels resulting from a non-inflammatory Th2 polarized immune response. One of the authors, Roger Rosenberg, as disclosed in the paper, is a co-inventor of a US patent application for “Aβ gene vaccines”. The fact that a patent was obtained is another indicator that our approach is innovative and relevant. Our group has made published contributions to this field of research since 2003 and we have presented the double plasmid approach at the ICAD meetings in 2008 and 2009.
The multiple copies of Aβ are linked together, making a trimeric protein rather than just expressing the single peptide three times. This is the most novel part of the paper in terms of a vaccine. Because B cells recognize antigens in solution and require receptor crosslinking for activation, multimeric epitopes are able to elicit B cell responses more potently than single epitopes. The double plasmid system, in which one plasmid encodes a yeast transcription factor, which drives the transcription of the Aβ42 sequence on the second plasmid, resulted in an about 10-fold stronger antibody production and improves the DNA Aβ42 trimer vaccine substantially. It is still much less compared to the peptide-immunized mice, but as we stated in the paper “a predominant Th2 response is a more important objective in developing an effective and safe therapy for AD than increased antibody levels alone”. It must be noted that T cell help is obviously provided in this model because isotype switching has occurred.
A major finding of this paper is that despite multiple immunizations T cell responses are undetectable. T cell responses are likely to be the cause of the encephalitis found in the clinical trial of peptide vaccination.
As stated by Dr. Agadjanyan, Aβ42 is a T cell-dependent antigen. Is it safe to generate a T and a B cell response by a DNA vaccine encoding full-length Aβ42? First, to have an effective immune response, both T and B lymphocytes are needed, as both strongly influence each other. In our manuscript, we do not state that our vaccine does not activate T cells. Instead, we say that T cell have been involved in the early immunizations process, as it is clearly indicated by the isotype switch of the respective Aβ42 antibodies produced. However, we could not detect in-vitro T cell activation or proliferation against Aβ42 as documented by appropriate detection assays. Since we did not find a prolonged T cell response, we conclude this type of vaccine is probably safe for use in humans. We are currently performing experiments to characterize early T cell responses and to determine T cell fate longitudinally. Second, the exact B and T cell epitopes in humans have not yet been clearly defined, as most of the work to date is done in mouse strains. Thus, we currently have insufficient data to avoid certain components of the Aβ42 peptide (for example: the T cell epitopes in the mouse) in the development of an effective vaccine for human AD patients.
Sufficient statistical analyses are provided in our manuscript. Every data point consists of mean and standard deviation. Additional information on the number of animals and the statistical assessment are provided in the article.
We agree with Dr. Agadjanyan that there is ongoing scientific discussion regarding the relevance of the plaque hypothesis. Removal of established plaques did not ameliorate or prevent progression of the disease in one clinical trial. Although the plaque count was reduced dramatically in these patients, there was evidence of progressive cognitive decline (Holmes et al., 2008). While it is very challenging to study these two seemingly contradictory observations in human patients, intuitively one would speculate that therapy was initiated too late when irreversible progressive and independent pathologic changes had already occurred. The greatest challenge for investigators who attempt to prevent clinical and pathological aspects of AD by DNA immunization or other interventional means will be the identification of biomarkers that will allow an early therapy. In summary, we think that our manuscript addresses a number of important key questions and we welcome further discussion.
References:
Holmes C, Boche D, Wilkinson D, Yadegarfar G, Hopkins V, Bayer A, Jones RW, Bullock R, Love S, Neal JW, Zotova E, Nicoll JA. Long-term effects of Abeta42 immunisation in Alzheimer's disease: follow-up of a randomised, placebo-controlled phase I trial. Lancet. 2008 Jul 19;372(9634):216-23. PubMed.
Uppsala University
The development of innovative vaccination strategies for Alzheimer disease is of utmost importance. In this article, a novel and interesting DNA-vaccine approach with two plasmids is described. Unfortunately, the antibody response raised in the non-transgenic mice is still fairly low, with OD-values ~0.5 when plasma samples are diluted only 1:500 (Fig. 3). One of the main difficulties with a DNA vaccine is to raise a significant immune response in humans or primates, but this problem is not addressed. The description of experimental procedures lack important details and their findings are presented in a somewhat incomplete manner. For instance, in Fig. 4, splenocytes from only a few mice were restimulated, and then only with the Aβ42 peptide. The authors conclude that a T cell response is undetectable. The obvious control experiment would have been to restimulate with the trimeric Aββ42 peptide. It is unclear if the splenocyte cultures contained only T cells, or were a mixture of T and B cells. If so, it would have been interesting to investigate if the isotype profile of the humoral response in blood samples (in Fig. 2) could have been replicated when cultured cells were restimulated. We regard this article as potentially interesting, but still very preliminary. Efficacy and safety experiments in transgenic mice and/or primates will be needed to evaluate this novel DNA vaccine strategy before it can be clinically tested.
Vice President, The Institute for Molecular Medicine
I'd like to thank Dr. Lambracht-Washington et al. for their reply and respond to a few points.
Patents for specific plasmids are fairly routinely granted. Such a patent is not in and of itself proof of scientific innovation on the present challenges of Alzheimer disease vaccine design. The T cell epitope of amyloid in humans has been reported by Monsonego et al. (1); the B cell epitope in humans is very well described by the Elan group (2), and our group has characterized the mouse T cell epitope (3).
I agree with the authors' claim that the novel part of their study would be the trimeric protein. If so, the study would be well advised to focus its data and discussion around this new aspect. However, the paper contains no data characterizing the form of peptide expressed from their plasmid. There are no data to show that multiple copies of Aβ are expressed linked together into a trimer rather than three times as single peptides. The presumed trimeric antigen is not characterized.
References:
Monsonego A, Zota V, Karni A, Krieger JI, Bar-Or A, Bitan G, Budson AE, Sperling R, Selkoe DJ, Weiner HL. Increased T cell reactivity to amyloid beta protein in older humans and patients with Alzheimer disease. J Clin Invest. 2003 Aug;112(3):415-22. PubMed.
Lee M, Bard F, Johnson-Wood K, Lee C, Hu K, Griffith SG, Black RS, Schenk D, Seubert P. Abeta42 immunization in Alzheimer's disease generates Abeta N-terminal antibodies. Ann Neurol. 2005 Sep;58(3):430-5. PubMed.
Cribbs DH, Ghochikyan A, Vasilevko V, Tran M, Petrushina I, Sadzikava N, Babikyan D, Kesslak P, Kieber-Emmons T, Cotman CW, Agadjanyan MG. Adjuvant-dependent modulation of Th1 and Th2 responses to immunization with beta-amyloid. Int Immunol. 2003 Apr;15(4):505-14. PubMed.
University of Southern California
The study by Lambracht-Washington and coworkers explores the possibility of using a DNA-based Aβ1-42 trimer vaccine to promote Th2-type, Aβ1-42-specific immune response in wild-type mice. As detailed above in comments from Michael Agadjanyan, this work extends from pioneering studies that were previously conducted in the area of DNA-based vaccination against Alzheimer’s. Those past studies, a number of which were conducted in the Agadjanyan and Cribbs labs, definitively established that DNA-based Aβ vaccines were efficacious in reducing cerebral amyloidosis in transgenic Alzheimer’s mouse models (for a review, see Town, 2009). I agree with Michael that the novelty in the present study is primarily owing to use of a trimeric Aβ1-42 vaccine administered via a gene-gun delivery system, and secondarily, due to utilization of a dual plasmid system that relies on GAL4 transactivation of the 12.2 kD Aβ1-42 trimer antigen.
I would like to add a few comments regarding this manuscript. It seems that the amount of anti-Aβ1-42 antibodies is rather low, even after a prime-boost regimen of up to eight total DNA vaccinations. For instance, it appears that DNA-vaccinated mice produced ~fivefold less antibodies against full-length anti-Aβ1-42 versus peptide-vaccinated mice (and that is a conservative estimate assuming linearity of OD readings for antibody titres; see Fig. 3) and ~five- to sixfold reduced stimulation of T cell proliferation following Aβ1-42 splenocyte challenge compared to peptide immunization (see Fig. 4). Because the authors only chose to examine wild-type mice, it is unclear whether this relatively weak Aβ1-42 antibody response would be efficacious, and thereby translate to reduced cerebral amyloidosis in mice, let alone in humans. It is possible that the trimeric Aβ1-42 antigen is not optimal for generating antibody responses to synthetic Aβ1-42; however, the authors do not seem to report results from trimeric Aβ1-42 splenocyte recall stimulation experiments, which may have resulted in stronger anti-Aβ antibody responses.
I’d like to build on a comment that Michael made regarding assessment of Aβ immunotherapy safety. In the “Comment” section of this article, the authors state that they can assess safety of the DNA-based approach by comparing immune responses between DNA-based trimeric vaccination and peptide-based immunization conducted side-by-side. This is flawed logic. With the exception of one unconfirmed report that utilized pertussis toxin in combination with “standard” Aβ immunotherapy using Aβ1-42 plus CFA/IFA (Furlan et al., 2003), none of the Aβ immunotherapy approaches have produced aseptic meningoencephalitis—nor any other severe adverse event—in mice. Because of this, it is not possible to make any conclusions about safety by simply assessing Th1 or Th2 immune responses after DNA Aβ1-42 trimer vaccination in mice.
Finally, I would like to call the authors’ attention to our previous work, which was one of the first demonstrations that “standard” Aβ1-42 immunotherapy (as originally developed by Schenk et al., 1999) primarily produces a Th2 response in mice as determined by 1) primarily IgG1 anti-Aβ1-42 antibodies that recognized amino acids 1-12 of the peptide (now widely regarded as the B cell epitope), and 2) primarily Th2 cytokines both in vivo in mouse plasma and ex vivo in splenocyte assays (Town et al., 2001; Town et al., 2002). More recently, we have shown that changing the route of vaccination to transcutaneous delivery results in copious amounts of anti-Aβ antibodies, a primarily Th2-type immune response, and clearance of cerebral amyloid (Nikolic et al., 2007). Thus, it is possible to generate Th2 immune responses in mice using simpler, peptide-based approaches to Aβ immunotherapy. The key challenges are to translate these approaches into the clinic by devising strategies that are both safe and efficacious.
References:
Furlan R, Brambilla E, Sanvito F, Roccatagliata L, Olivieri S, Bergami A, Pluchino S, Uccelli A, Comi G, Martino G. Vaccination with amyloid-beta peptide induces autoimmune encephalomyelitis in C57/BL6 mice. Brain. 2003 Feb;126(Pt 2):285-91. PubMed.
Nikolic WV, Bai Y, Obregon D, Hou H, Mori T, Zeng J, Ehrhart J, Shytle RD, Giunta B, Morgan D, Town T, Tan J. 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.
Schenk D, Barbour R, Dunn W, Gordon G, Grajeda H, Guido T, Hu K, Huang J, Johnson-Wood K, Khan K, Kholodenko D, Lee M, Liao Z, Lieberburg I, Motter R, Mutter L, Soriano F, Shopp G, Vasquez N, Vandevert C, Walker S, Wogulis M, Yednock T, Games D, Seubert P. Immunization with amyloid-beta attenuates Alzheimer-disease-like pathology in the PDAPP mouse. Nature. 1999 Jul 8;400(6740):173-7. PubMed.
Town T. Alternative Abeta immunotherapy approaches for Alzheimer's disease. CNS Neurol Disord Drug Targets. 2009 Apr;8(2):114-27. PubMed.
Town T, Tan J, Sansone N, Obregon D, Klein T, Mullan M. Characterization of murine immunoglobulin G antibodies against human amyloid-beta1-42. Neurosci Lett. 2001 Jul 13;307(2):101-4. PubMed.
Town T, Vendrame M, Patel A, Poetter D, Delledonne A, Mori T, Smeed R, Crawford F, Klein T, Tan J, Mullan M. Reduced Th1 and enhanced Th2 immunity after immunization with Alzheimer's beta-amyloid(1-42). J Neuroimmunol. 2002 Nov;132(1-2):49-59. PubMed.
University of Texas Southwestern Medical Center at Dallas
Clearly, more work needs to be accomplished to define the immune response with Aβ42 immunization in both experimental animal models and in patients, but we are dedicated to proceeding in this direction.
Two related hypotheses drive our research: First, the generation of an effective anti-Aβ42 B cell response that will result in an antibody-mediated clearance of amyloid plaques, leading to a reduction in AD severity. Second, from a bio-safety standpoint, the generation of an anti-Aβ42 Th1 effector autoimmune response may lead to immune mediated pathology. These hypotheses are based on research by many others, including the clinical trial using peptide vaccination. We have sought to develop a therapy that will generate an effective antibody response and will lack measurable Aβ42-specific Th1 T cell responses. At this point, the definition of "effective" may not be directly related to Ig titer but is rather a biological outcome-based definition, as we have detected clearance of plaque with low titers previously and there is no method currently available to test CSF or CNS titers which are more relevant than serum (1,2). We specifically chose to test these hypotheses in non-transgenic mice in order to provide the most stringent possible test for T cell responses, as these mice have no thymic negative selection against human Aβ42. Therefore, if the potential existed for the development of a Th1 response, it should be detectable in this model (null hypothesis). Therefore, we compared the immune responses in wild-type mice against DNA and Aβ42 peptide immunization.
This direct comparison is innovative and constructive, as the results show a clearly different immune response with a predominant Th2 immune response with DNA Aβ42 immunization (Th2/Th1 ratio 10) and a mixed immune response with Aβ42 peptide (Th2/Th1 ratio 1). Furthermore, T cell proliferation (as determined by the stimulation indices) following DNA versus peptide immunizations showed again very different outcomes with non-reactive T cells after re-stimulation with Aβ42 peptide after DNA Aβ42 immunization and reactive T cells after Aβ42 peptide immunization.
These are important findings and strongly support proceeding ahead with a clinical trial using DNA Aβ42 immunization in patients. Additional studies using this construct in APP transgenic mice are ongoing as well as further characterization of the trimeric Aβ42 peptide as the new and innovative antigen in our system. We are aware of the challenge that the outcome in humans may or may not be the same as the findings in mice, but our study offers a valid explanation as to why the clinical trial with Aβ42 peptide immunization had caused encephalitis in 6 percent of the patients. We agree with Terrence Town, that “none of the Aβ immunotherapy approaches have produced aseptic meningoencephalitis nor any other severe adverse event in mice,” but our data are in strong support that DNA Aβ42 immunization does offer a potentially lower risk for adverse effects in patients. Because the meningoencephalitis was likely due an inflammatory Th1 immune response caused by the type of adjuvant used, we agree that it is possible to generate an non-inflammatory immune response using simpler, peptide-based approaches to Aβ immunotherapy, but we would like to cite a statement by Michael Agadjanyan in the first comment to our paper: “scale-up of manufacturing for clinical trials is readily achieved, and cGMP-grade DNA vaccines are significantly less expensive than peptide-based cGMP vaccines.”
These discussions with our colleagues are constructive and help to provide focus for future research.
References:
Qu B, Boyer PJ, Johnston SA, Hynan LS, Rosenberg RN. Abeta42 gene vaccination reduces brain amyloid plaque burden in transgenic mice. J Neurol Sci. 2006 May 15;244(1-2):151-8. PubMed.
Qu BX, Xiang Q, Li L, Johnston SA, Hynan LS, Rosenberg RN. Abeta42 gene vaccine prevents Abeta42 deposition in brain of double transgenic mice. J Neurol Sci. 2007 Sep 15;260(1-2):204-13. PubMed.
UC, Irvine
I’d like to add a few points to this discussion. I agree that the double plasmid system with GLA4 Activator, UAS/AB42 Trimer responder is quite innovative; however, whether this approach has translational potential for a human clinical trial is doubtful based on the data provided in the current manuscript.
I found it interesting that the wild-type mice used in this study were female B6SJLF1/J mice, which is the background of the Tg2576 mice that show immune hypo-responsiveness to immunization with Aβ42 (1). No data were provided in the current manuscript using APP transgenic mice to show that the antibody response induced by the double plasmid system was capable of attenuating amyloid deposition or improving behavioral measures. This raises the issue of immune tolerance that needs to be broken in order to induce an adequate antibody response to a “self” peptide or protein. In addition, because the target population for anti-Aβ immunotherapy is the elderly, overcoming immunosenescence is a formidable hurdle as well. The generally low titers and the low number of positive responders in the AN1792 clinical trial (I say that with the proviso that the trial was halted before the patients completed the protocol) illustrate the difficulties facing future active immunization trials in this elderly patient population. Therefore, inducing robust anti-Aβ antibody responses in APP transgenic mouse models, as well as large animal models, will likely be a prerequisite before moving forward into AD patients. In fact, we have previously published a study in elderly dogs that used fibrillar Aβ42 and alum as a Th2 adjuvant approved for human use; this induced therapeutic levels of anti-Aβ antibody titers, sufficient to clear amyloid deposits, in all of the immunized canines (2).
Other commentators on Alzforum have already pointed out the large number of immunizations required to induce rather low titers of anti-Aβ antibodies in this study, as well as the significant potential to induce an anti-GLA4 protein immune response. The latter would likely result in the loss of cells expressing the GLA4 protein, thereby eliminating the expression of the Aβ42 trimer immunogen. I would like to emphasize the fact that just because you fail to measure a T cell response from immunized mice does not mean that there was no response. Typically a T cell response is required to get Ig class switching from IgM to the IgG isotypes; Lambracht-Washington et al. were able to measure IgG1 and IgG2a isotypes, thus indicating that there was a T cell response to the Aβ42 trimer immunogen. Thus, the problem was likely due to the weak T cell response that was induced with the double plasmid system, thereby making it difficult to detect in the re-stimulated splenocyte cultures, which was also reflected in the rather weak antibody response to the double plasmid system.
A second major problem is the large number of immunizations needed to get the anti-Aβ antibody titers to 15 ug/ml. In fact, the data in Figure 3 do not appear to show any difference between the mice (n = 4) that received six immunizations and those (n = 4) that received 14 immunizations if you compare the antibody binding in panels A and B, because all of the plasma samples were diluted 1:500. Multiple immunization protocols—especially DNA immunization protocols requiring 6-14 injections that also require specialized equipment, such as electroporation, to promote adequate expression levels of the antigen to induce therapeutic levels of antibodies—are probably not a viable approach for translation to human clinical trials. Therefore, inclusion of some sort of adjuvant in the vaccine design to enhance the antibody titers and reduce the number of immunizations, either as a fusion construct or as a supplement (co-injection or transcutaneous delivery) to the immunogen will be required.
References:
Monsonego A, Maron R, Zota V, Selkoe DJ, Weiner HL. Immune hyporesponsiveness to amyloid beta-peptide in amyloid precursor protein transgenic mice: implications for the pathogenesis and treatment of Alzheimer's disease. Proc Natl Acad Sci U S A. 2001 Aug 28;98(18):10273-8. PubMed.
Head E, Pop V, Vasilevko V, Hill M, Saing T, Sarsoza F, Nistor M, Christie LA, Milton S, Glabe C, Barrett E, Cribbs D. A two-year study with fibrillar beta-amyloid (Abeta) immunization in aged canines: effects on cognitive function and brain Abeta. J Neurosci. 2008 Apr 2;28(14):3555-66. PubMed.
University of Texas Southwestern Medical Center at Dallas
We thank David Cribbs for his comment, and we would like to correct a few things. First of all, we did not show the results from 14 immunizations in our mice. We immunized one group receiving a total of six DNA immunizations and a second group which received a total of eight DNA Aβ42 trimer immunizations. The resulting antibody titers are 10-fold higher than antibody titers from a previous published immunization construct from our group which was an Aβ42 monomer. For immunizations with the monomer, we have shown an effective reduction in plaque load in an APP transgenic mouse model. Therefore, we conclude that the new DNA Aβ42 trimer is very likely to cause an even better or at least equal reduction in overall plaque load and is thus effective and sufficient to cause a therapeutic effect. Further studies using DNA Aβ42 trimer immunization in transgenic mouse models are underway. Again, we would like to emphasize that an initial T cell response was observed by the isotype switching of the respective Aβ42 specific antibodies, but then the T cell response disappeared, not due to weak T cell response but due to a Th2 response which is non-inflammatory. The reason why we choose B6SJLF1 mice was indeed because this strain is the genetic background for a number of APP transgenic mouse strains. B6SJLF1 is a good responder in Aβ42 immunizations, as it carries the mixed background of C57BL6 which is poorly responding to Aβ42 immunization but has a genetic background promoting brain plaque development and SJL which has a great Aβ42 immune response.
University of Southern California
I would like to respond to a reply made above by Doris Lambracht-Washington concerning the Th2 response issue raised by David Cribbs. Her reply was “Again, we would like to emphasize that an initial T cell response was observed by the isotype switching of the respective Aβ42 specific antibodies, but then the T cell response disappeared, not due to weak T cell response but due to a Th2 response which is non-inflammatory.”
The type of T cell response, i.e., Th1 vs. Th2, is not defined in any way by the duration or persistence of the response. These Th profiles are defined based on the types of cytokines produced; e.g., Th1 responses are primarily associated with interferon-gamma and interleukin-12, whereas Th2 responses occur with interleukin-4 and interleukin-10 production, amongst others such as interleukin-15. Thus, the Th2 response to trimeric Aβ antigen observed by Lambracht-Washington and coworkers would not, by its very nature, have a limited duration because it is anti-inflammatory.
Furthermore, it is a misconception that Th2 responses are always “non-inflammatory.” A number of immune disorders, including allergy and atopic disorders such as asthma are driven by overly aggressive Th2 responses.
Make a Comment
To make a comment you must login or register.