Could interferon-γ (IFN-γ) be to blame for the encephalitis that brought an end to clinical trials of Elan’s amyloid-β (Aβ) vaccine? That’s one interpretation of results, published March 20 in the PNAS Online Early Edition, by Howard Weiner, Brigham and Women’s Hospital, Boston, and colleagues there, at Ben-Gurion University, Israel, and at the University of Southern Denmark in Odense. The scientists report that meningeal encephalitis, which was not observed in preclinical testing of the AN1792 vaccine, can be induced in mice if the animals also express limited IFN-γ in the central nervous system (CNS). The data suggest that small amounts of the cytokine in the brain may have exacerbated or triggered the inflammatory response in about 6 percent of the AD patients who took part in the Phase 2 trial (see ARF related news story).

As seen again this week, other species are not always predictable test models of the human immune system. Six healthy volunteers in Britain ended up in intensive care soon after receiving a monoclonal antibody slated for treatment of leukemia (see related news from the BBC). That antibody had apparently been pretested in at least two other species. Similarly, since the Elan trial was halted, speculation has been rife as to why a small number of patients had such a strong reaction to the antigen (see ARF related news story) even though prior animal tests had raised no flags about inflammation. The general consensus seems to be that those patients mounted an aggressive T cell response. Now Weiner and colleagues report that T cell responses can also be elicited from mice expressing certain major histocompatibility (MHC) haplotypes. These mice will then develop meningoencephalitis if they also express IFN-γ in the brain.

To study T cell responses to Aβ, first author Alon Monsonego and colleagues immunized two strains of mice with human Aβ42, then examined the popliteal lymph nodes behind the knee. They found that the number of T cells appearing in the lymph nodes of SJL mice (H2s haplotype of class II MHC) was more than 15-fold higher than in the same nodes from C57BL6 animals (H2b haplotypes of class II MHC). Furthermore, when the authors challenged these T cells with 10 different peptides that span the entire length of Aβ42, they found that SJL T cells proliferated in the presence of Aβ10-24 and Aβ7-21. In contrast, T cells from the C57BL6 nodes were unaffected by these peptides. C57BL6 cells did respond to Aβ16-30, albeit only at 10-fold higher peptide concentration. The findings suggest that the MHC background determines not only what epitope gets recognized, but also the strength and type of the response.

But even if the immune system does mount a T cell response, how would the cells access the brain to cause meningoencephalitis as seen in the AN1792 trial? Monsonego and colleagues wondered if microglia might provide the impetus because these brain cells can stimulate T cell infiltration. To check this idea, they immunized APP transgenic mice (see Mucke et al., 2000) with Aβ42, then examined the brains for signs of inflammation. They found a smoking gun—microglia co-localizing with both amyloid plaques and CD28, a T cell stimulator—but alas, no encephalitis. Something was still missing from the picture.

The researchers then wondered if IFN-γ might play a role because it is known to activate microglia. To test this, they crossed SJL mice expressing modest amounts of IFN-γ in the CNS with APP transgenics. Then they immunized the animals with Aβ10-24, the strongest antigen. In as little as 12 days, they found that the animals had mounted a robust response and had marked meningoencephalitis. “Overall, we demonstrate that Aβ10-24 immunization can induce temporary meningoencephalitis primarily targeted to sites of Aβ burden provided that IFN-γ is expressed in the brain,” conclude the authors.

The data are curiously similar to what has been reported for human patients. In the Swiss cohort of the Elan trial, three of 30 patients immunized developed aseptic encephalitis. Two of them had Aβ antibodies, suggesting that they had mounted a B cell response, but Aβ antibodies could not be detected in the sera of the third patient, suggesting primarily a T cell response (see Hock et al., 2003 and ARF related news story). It would be interesting to compare the MHC haplotypes of these T and B cell responders and also those who did/did not develop encephalitis.

Monsonego and colleagues note that it is unknown if IFN-γ was expressed in the brains of the patients who developed encephalitis during the trial, or even if the cytokine is produced in the adult brain under normal circumstances. It has recently been reported that expression of IFN-γ in the mouse brain increases with age, suggesting a gradual transition to a more proinflammatory environment as animals get older (see Frank et al., 2005), and some researchers have noted that expression of a number of proinflammatory cytokines appear to increase with human aging, as well. The cytokine may also be expressed in the brain in response to viral or bacterial infection, suggest Monsonego and colleagues, thus predisposing the brain to heightened microglial activation and T cell infiltration.

Whatever the cause of the encephalitis in the trial patients, it is curious that the severity of dementia in the Swiss cohort patient with no B cell response continued to worsen after recovery from the brain inflammation, while the two patients who generated Aβ antibodies were cognitively stable one year later, suggesting that the antibodies are beneficial even after the encephalitis had passed. In the case of the mice, encephalitis seemed to hasten clearance of Aβ from the brain. When Monsonego and colleagues challenged SJL/APP Tg mice with Aβ10-24, they found that many highly activated microglia and macrophages colocalized with Aβ deposits in the brain. Those hot spots of immune cell activity contained significantly decreased amounts of Aβ.

A challenge now is to achieve similar Aβ reductions in humans without touching off encephalitis. Pre-trial testing of patients (O’Toole et al., 2005), passive immunizations (see ARF related news story), and antigens designed to avoid T cell responses (see ARF related news story) are some of the strategies currently being investigated.—Tom Fagan


  1. This is a very interesting paper from Monsonego and colleagues. It provides us with another potential mechanism for the meningoencephalitis observed in the Aβ vaccination clinical trial. The two factors that appear critical to the development of meningoencephalitis in these experiments are the MHC class II haplotype and brain levels of IFN-γ. Also, the removal of amyloid in APP+IFN-γ transgenic mice appears to be antibody-independent as no titers were developed.

    The authors show that T cell proliferation following a single Aβ1-42 immunization in SJL mice with the I-As MHC-II haplotype is over 10-fold greater than in C57BL6 mice with the I-Ab haplotype. Also, different Aβ fragments stimulate T cell proliferation when different MHC-II haplotypes are present. This suggests that the HLA haplotype of the patient could act as a predictor for the development of meningoencephalitis in response to Aβ vaccination.

    Aβ vaccination of APP+IFN-γ transgenic mice on an SJL background results in marked meningoencephalitis only 12 days following immunization. Importantly, the IFN-γ transgene is under the control of the myelin-basic protein promoter, ensuring only CNS expression. The time course of this T cell infiltration is extremely interesting, as it shows that by day 20, the CD4 and Cd11b cells have moved from the meningeal tissue to the parenchymal plaques in the hippocampus. By day 60, no infiltrates were observed, suggesting the meningoencephalitis is transient. Hippocampal tissue close to the area of infiltration in the APP+IFN-γ transgenic mice shows significant activation of microglia and reduced levels of compact amyloid. These mice did not develop antibody titers, suggesting that the amyloid removal is antibody-independent and instead associated with the T cell response. This suggests a novel mechanism of amyloid removal in response to active vaccination. The IFN-γ is critical to the development of meningoencephalitis, as the single APP transgenic mice did not show T cell infiltrates. It would be interesting to see if administration of another immunization would result in new infiltration and whether this would be more severe than observed following the first dose.

    These data highlight the necessity of characterizing the role of inflammation in Alzheimer disease. This is now more important than ever since we are developing new, disease-modifying therapies, the efficacy of which could depend on the inflammatory state in the brain.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Human Aβ Vaccine Snagged by CNS Inflammation
  2. The Alzheimer's Vaccination Story, Continued
  3. Alzheimer’s Vaccine: In Some Patients, at Least, It Might Just Work
  4. Pilot Study Shows Promise of Passive Immunotherapy
  5. Sorrento: Immunotherapy Update Hot Off Lectern of AD/PD Conference

Paper Citations

  1. . High-level neuronal expression of abeta 1-42 in wild-type human amyloid protein precursor transgenic mice: synaptotoxicity without plaque formation. J Neurosci. 2000 Jun 1;20(11):4050-8. PubMed.
  2. . Antibodies against beta-amyloid slow cognitive decline in Alzheimer's disease. Neuron. 2003 May 22;38(4):547-54. PubMed.
  3. . mRNA up-regulation of MHC II and pivotal pro-inflammatory genes in normal brain aging. Neurobiol Aging. 2006 May;27(5):717-22. PubMed.
  4. . Risk factors associated with beta-amyloid(1-42) immunotherapy in preimmunization gene expression patterns of blood cells. Arch Neurol. 2005 Oct;62(10):1531-6. PubMed.

External Citations

  1. BBC

Further Reading

No Available Further Reading

Primary Papers

  1. . Abeta-induced meningoencephalitis is IFN-gamma-dependent and is associated with T cell-dependent clearance of Abeta in a mouse model of Alzheimer's disease. Proc Natl Acad Sci U S A. 2006 Mar 28;103(13):5048-53. PubMed.