As drug discovery in AD research is shifting from inhibiting the γ-secretase to inhibiting the β-secretase (BACE), interest in the entire life cycle of BACE has grown, concomitantly. In the current online Early Edition of PNAS, Vincent Mauro, George Rogers, and Gerald Edelman at the Scripps Research Institute in La Jolla, California, shine a light on structural features of the BACE1 mRNA that might help explain why levels of the enzyme increase in Alzheimer's disease.

Most evidence to date suggests that the increases in BACE seen in the brains of AD patients arise neither from genetic mutations or polymorphisms, nor from transcriptional changes. (However, see ARF concurrent news story on Li et al., who report increased BACE mRNA levels in AD brain.) Researchers have thus begun to hunt for mechanisms that could increase translation of the protein from its mRNA, or that extend the lifespan or activity of the protein.

Mauro and colleagues have helped to open this new chapter in AD research by focusing on the leader region of the BACE mRNA. For those who haven't cracked their molecular biology texts for a few years, the leader sequence follows the promoter and operator sequences at the 5' end (i.e., the beginning) of the mRNA. By definition, the leader ends at the start codon (AUG), which signals the translational machinery that the next open reading frame (ORF) contains the code for the actual protein. A potential problem is the presence of any upstream AUGs in the leader sequence; the BACE1 mRNA, for example, contains four of the imposter AUGs (though only three ORFS, because one of the AUGs is immediately followed by a stop codon).

Theoretically, these uAUGs have the capacity to trick the ribosome and its associated cogs and gears into starting translation ahead of the actual gene, and such a false start can block translation. (This need not always be a bad thing, as the current study will demonstrate.) However, the translational engine has its own tricks to avoid false starts. Rogers et al. set out to discover what might be at work here to allow increased production of BACE in Alzheimer's disease.

When the researchers transfected the BACE1 leader—in a construct with a reporter gene—into two different cell types (rat B104 neuroblastoma and rat PC12), they found that the gene was relatively efficiently translated, though there were clear differences in the amounts translated in the two cell types. They also demonstrated that the translation was 5' cap-dependent; this indicates that the translation machinery is not assembled at an internal ribosome entry site (IRES), one of the tricks that the translation engine can use to avoid uAUGs. Deleting or mutating some or all of the uAUGs also led to differential changes in translation between the two cell types. These changes were small (two- to fourfold), but they reinforced the notion that certain cellular conditions enable these uAUGs to inhibit BACE1 translation.

Another trick that ribosomes can use to avoid getting snagged on uAUGs is "leaky scanning," whereby the ribosome surveys the nucleotides immediately before and after the AUG, and bypasses sites that look troublesome. The authors found that this was unlikely in the case of BACE, as all four uAUGs were found individually to be efficient initiators of translation when attached to a gene other than BACE1.

How, then, does BACE ever get translated if these uAUGs can hijack the translational engine before it ever reaches the BACE1 AUG? The authors argue that the most likely way is by "shunting," wherein ribosomes simply jump over large stretches of the mRNA. Shunting often occurs in areas where the molecule makes hairpin turns or loops, and could be at work in AD, the scientists propose. "For example, the translation of a uORF might to some extent inhibit BACE1 translation in the normal brain, whereas during Alzheimer’s disease, translation might increase because of a shunting mechanism that enables ribosomes to bypass the upstream AUGs," they write.

One possible mechanism for this shift is that the relative accessibility of the uORFs and the BACE ORF may change in AD, or even during aging, as a function of changes in the three-dimensional structure of the BACE1 mRNA. This hypothesis could be tested in neurons cultured from brain tissue removed from AD patients during surgery for epilepsy, the authors suggest.—Hakon Heimer


  1. Elevated BACE at the mRNA, protein, and/or activity levels have been found in the Alzheimer's disease (AD) and aging brain. Clarifying the mechanism of BACE alterations in AD will contribute to identifying therapeutic targets. This rigorous study further elucidates the complex cell biology of BACE, and demonstrates that BACE is susceptible to regulation at the translational level. The authors show that BACE translation is cap-dependent; translation can be inhibited by nucleotides 61-74 and 180-190 in the leader sequence and by the second of four AUG sequences in the 5' UTR, depending on the cell line.

    Other factors can affect BACE levels, as well. Promoter sequences in the BACE gene include Sp1-responsive elements that regulate transcription. BACE undergoes alternative splicing, as well as multiple co- and post-translational modifications including N-glycosylation, sulfation, phosphorylation, and furin-mediated cleavage of the prodomain. Alternative BACE substrates, distribution of BACE between secretory and endosomal compartments, and cellular cholesterol metabolism may also modulate BACE activity. BACE cell biology is liable to be affected at multiple levels in AD, contributing to the accumulation of amyloid-β protein.

  2. The Rogers et al. paper (contributed by G.M. Edelman to PNAS) describes a possible mechanism of translational control in rat BACE-1 transcript. We have a paper in press in NAR (submitted September 25th 2003 and accepted March 2nd 2004) dealing with the same issue (De Pietri Tonelli et al., Nucl. Acids Res., 2004, in press). Here we would like to outline substantial differences between our paper and the Rogers et al. one that lead to diverging conclusions.

    Our control experiments clearly reveal that human BACE-1 transcript contains a cryptic promoter that is unraveled by standard DNA transfection. To dissect transcriptional from translational contribution, we adopted an expression system that confines transcription to the sole cytosol. Under these conditions we demonstrate a strong inhibition of translation initiation by BACE-1 transcript leader. This result was fully confirmed by in-vitro translation. Both polysomal analysis and controls on transcript stability are in line with this conclusion. It should also pointed out that in November 2000 we submitted the sequence of a splice variant of the BACE-1 transcript leader (Zacchetti et al., accession number AF324837) detected in human neuroblastoma and exocrine pancreas. Apparently, Rogers and colleagues did not consider the possibility that alternative splicing in the BACE-1 transcript leader might influence their results.

    On the whole, we propose that multiple mechanisms can control the translation in the human BACE-1 transcript, but we also present evidence that shunting does not provide a significant contribution.
    Obviously, we cannot rule out the possibility that the strong discrepancies between the two papers derive from a different control of BACE-1 translation in rat and human, a possibility that, in any extent, would raise a word of caution on the overall discussion of the results.
    Daniele Zacchetti (
    Davide De Pietri Tonelli (
    Marija Mihailovich (
    Alessandra Di Cesare (
    Franca Codazzi (
    Fabio Grohovaz (

    View all comments by Daniele Zacchetti
  3. Transcriptional regulation is probably the major mode of regulating eukaryotic gene expression; however, post-transcriptional mechanisms also contribute, in many instances, to determine the final amount of a gene product that is present in the cell at a given moment. In particular, the efficiency of translation of certain transcripts can be modulated, and in most cases the signals responsible for this modulation are located in the 5'-untranslated regions of the mRNAs (5'-UTRs or leader sequences, the region of the mRNA that is upstream of the translation initiation codon). The mRNA for BACE1, the enzyme involved in Alzheimer’s disease, seems to belong to this category, according to a paper published in last week’s PNAS, by first author George Rogers and colleagues at The Scripps Research Institute, La Jolla.

    In 1989, Marilyn Kozak proposed a mechanism for translation initiation known as scanning model. According to this model, the small ribosomal subunit and some translation initiation factors bind at the cap structure present at the 5'-end of all cellular transcripts, and scan linearly until they encounter the first AUG codon where translation initiates. This model fits most cellular mRNAs, since they contain unstructured leaders that are easily melted by the scanning ribosomes, and since translation initiates in these mRNAs at their first AUG codon. BACE1 mRNA, on the contrary, contains a 5'-UTR that is rich in GCs and has, therefore, the potential to fold into stable structures.

    Moreover, four AUG codons precede the BACE1 start codon, and thus, protein synthesis does not initiate at the most 5' AUG of the transcript. Like BACE1, other cellular and viral mRNAs contain weird leader sequences that violate Kozak’s rules. They are translated by mechanisms that are exceptions to the scanning model, including internal initiation, leaky scanning, reinitiation, and ribosome shunting. These “abnormal” expression strategies were first described for viral mRNAs and later usually also observed in rare cases of cellular gene expression.

    Rogers and colleagues propose that BACE1 mRNA is translated by ribosome shunting. This mechanism involves the recruitment of ribosomes in a cap-dependent manner, and their subsequent nonlinear migration, in which part of the leader is skipped and ribosomes are directly translocated to a site at or close to the start codon of the major open reading frame. This mechanism was originally described for a plant virus (of the Caulimoviridae family), and later shown to be used by some animal viruses, including adenovirus. To my knowledge, BACE1 would be the first cellular mRNA to use the shunting mechanism for translation initiation.

    The authors studied translation driven by the BACE1 leader both in an in-vitro translation system and in vivo upon transfection of different constructs in two rat-derived cell lines: B104 neuroblastoma cells and PC12, derived from a rat pheochromocytoma. The presence of the leader reduced expression of a reporter gene to 36 percent and 69 percent in B104 and PC12 cells, respectively, as compared to translation driven by the efficient β-globin leader. The fact that translation was inhibited by the presence of a strong stem-loop structure placed near the 5'-end both in vitro and in vivo, and by competition with a cap analogue, demonstrated that translation is cap-dependent. When the BACE1 leader was placed between two cistrons in a dicistronic mRNA, there was no expression of the second cistron, excluding the possibility that it functions as an internal ribosome entry site (IRES), allowing for internal initiation of translation.

    The authors went on to test a series of mutants that included either deletions of portions of the leader or point mutations of the AUG codons to AUU (in different combinations). Surprisingly, deletion of parts of the leader containing the AUGs or mutations of the AUG codons did not have a major effect on the efficiency of translation. Only the mutation of the second AUG resulted in a modest increase in PC12, but not B104 cells. These AUG codons, however, are efficiently recognized as initiation codons when they are transplanted to a different mRNA. In the context of the full-length BACE1 leader, they do not seem to be available for translation initiation, and this points to a nonlinear migration of ribosomes. Moreover, PC12 and B104 responded differently to the different mutants, indicating a cell-type-specific effect. For instance, an element that inhibited translation was mapped to a region between nucleotides 61-74 in B104, but to nucleotides 180-190 in PC12 cells. The authors propose that cell-specific factors might modulate the efficiency of shunting mediated by the BACE1 leader.

    Is there any evidence that translation of the BACE1 mRNA is regulated in vivo? If so, what are the implications for AD research? As the authors state, several studies have shown that BACE1 protein and activity are elevated in the brain of patients affected with AD, but this increase is not paralleled with an increase in mRNA levels. This would then indicate that BACE1 expression could indeed be modulated at the translational level by a mechanism that is altered in AD. Finding out exactly how this mechanism works and the factors implicated might reveal new therapeutic targets aimed at reducing BACE1 protein levels as an alternative to BACE1 enzyme inhibition.

  4. Response to comment by Daniele Zacchetti
    In a recent posting, Daniele Zacchetti and colleagues outlined substantial differences between the results reported in our paper and data from a paper that they have in press. Zacchetti suggests that the BACE1 transcript contains a cryptic promoter. This possibility was considered and addressed experimentally in our studies; however, our results could not be explained by cryptic promoter activity. In our transfection studies, the reporter constructs were transcribed via the strong CMV promoter and the translation efficiency of the reporter constructs containing the BACE1 5' leader was found to be up to approximately 70 percent of that of a reporter construct containing the 5' leader of the efficiently translated β-globin mRNA. In contrast, Zacchetti indicates that the BACE1 5' leader is inhibitory in their system, in which transcription occurs in the cytoplasm. If the high level of translation observed in our study was due to the production of shorter transcripts initiating at a cryptic promoter, they would have to be present at approximately the same level as the CMV-driven transcript, and thus should be easily detectable. However, no shorter transcripts were detected by Northern blot analysis.

    In addition, the introduction of a stable hairpin structure at the 5' end of the mRNA blocked translation by more than 98 percent, suggesting that the majority of the transcripts contained the full-length 5' leader. Comparable results were obtained in cell-free lysates using in-vitro transcribed RNA, corroborating the transfection results.

    Although Zacchetti suggests that we did not consider the possibility of a splicing event, which might remove the upstream AUGs from the 5' leader, this possibility was addressed in our studies by performing PCR reactions using primers corresponding to the extreme 5' end of the BACE1 mRNA and to the 5' end of the coding region of the reporter cistron. The results of these studies revealed only one band corresponding to the full-length 5' leader, which indicated that there were no shorter mRNAs with alternatively spliced 5' leaders.

    Zacchetti warns that the differences between the two studies might derive from differences between the rat and human sequences. We consider this possibility unlikely because of the strong similarity between these sequences, including the presence of four upstream AUGs and three upstream open reading frames in both sequences.

    Without access to the Zacchetti manuscript, we cannot comment directly on their data; however, there may be several reasons for the differences between the two studies. One possibility is that the translation mechanism mediated by the BACE1 5' leader varies substantially in different cell types, and we reported evidence consistent with this notion. A second possibility may be related to the use of a cytoplasmic expression system by Zacchetti. In our manuscript, we suggested that the translation mediated by the BACE1 5' leader might be altered by RNA secondary structures that affect the rate of translation by altering the accessibility of the initiation codon. Furthermore, it is possible that the RNA secondary structures of the BACE1 reporter mRNA, and consequently, the rate of translation, may be different when this mRNA is transcribed in the cytoplasm rather than in the nucleus. Variations in the RNA secondary structure might arise because of differences between RNA binding proteins that are present and accessible to the mRNA in the nucleus and those that are present and accessible to the mRNA in the cytoplasm.

  5. Over the last two years, several studies have suggested that levels of BACE1 protein are elevated in sporadic Alzheimer’s disease brains (1-4). It is not completely clear if the elevation of BACE1 protein is accompanied by an elevation in BACE1 message (as suggested in 4) or not (1-3). Obviously, based on postmortem studies, one cannot decide if the observed elevation in BACE1 protein is one of numerous biochemical abnormalities in advanced AD, or if it actually contributes to the pathogenesis. Because of the limited number of available studies, we do not know how robust this increase will turn out once a large number of cases are analyzed. With all these caveats in mind, it is nonetheless tempting to speculate that translational regulation of BACE1 could play a role in AD. Along these lines, several laboratories have initiated studies to investigate BACE1 mRNA translation.

    Rogers et al. (5) have now performed a detailed study of the complex 5' leader of BACE1 mRNA and how it affects translation efficiency. Analysis of the leader sequence revealed the presence of three conserved upstream ORFs (uORFs) preceding the BACE1 initiation codon. Disruption of the uORFs had no effect on translation in B104 cells. Christian Haass and coworkers have obtained similar data in 293 cells (C. Haass, personal communication), but Rogers et al. (5) saw an impact on translation in PC12 cells. A detailed analysis suggests that translation occurs by an unusual mechanism that is not consistent with processive scanning of the 5' leader or with internal initiation of translation, but rather with ribosomes starting at the 5' end and then skipping over segments of the leader as they move to the initiation codon (5).

    What does this have to do with AD? Translation properties of BACE1 mRNA could be altered in the disease, leading to increased BACE1 protein production. The different effects observed in different cell lines suggest the involvement of cell type-specific factors regulating BACE1 mRNA translation, which could account for a restriction of BACE1 protein increase to the brain. Clearly, regulation of BACE1 protein expression and its potential relationship to AD pathogenesis has emerged as an exciting topic for future studies.


    . Increased expression of the amyloid precursor beta-secretase in Alzheimer's disease. Ann Neurol. 2002 Jun;51(6):783-6. PubMed.

    . Beta-secretase protein and activity are increased in the neocortex in Alzheimer disease. Arch Neurol. 2002 Sep;59(9):1381-9. PubMed.

    . Elevated beta-secretase expression and enzymatic activity detected in sporadic Alzheimer disease. Nat Med. 2003 Jan;9(1):3-4. PubMed.

    . Amyloid beta peptide load is correlated with increased beta-secretase activity in sporadic Alzheimer's disease patients. Proc Natl Acad Sci U S A. 2004 Mar 9;101(10):3632-7. PubMed.

    . Differential utilization of upstream AUGs in the beta-secretase mRNA suggests that a shunting mechanism regulates translation. Proc Natl Acad Sci U S A. 2004 Mar 2;101(9):2794-9. PubMed.

  6. As mentioned above in the comment posted by Martin Citron (March 18, 2004), we have a paper in EMBO Reports dealing with translational control of BACE1 expression (Lammich et al., 2004). Like Mauro et al. and De Pietri Tonelli et al., we also found that the 5'UTR (but not the 3'UTR) of BACE1 inhibits the translation of a downstream open reading frame in different cell lines. In contrast to the other two studies, we did not rely on luciferase as a reporter gene, but instead, we directly used the human BACE1 cDNA and analyzed BACE1 expression by Western blot and Northern blot analysis. An extensive mutagenesis analysis predicts that the GC-rich region of the 5'UTR forms a constitutive translation barrier, which may prevent the ribosome from efficiently translating the BACE1 mRNA.


    . Expression of the Alzheimer protease BACE1 is suppressed via its 5'-untranslated region. EMBO Rep. 2004 Jun;5(6):620-5. PubMed.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Follow the BACE to Higher Aβ Levels

Paper Citations

  1. . Amyloid beta peptide load is correlated with increased beta-secretase activity in sporadic Alzheimer's disease patients. Proc Natl Acad Sci U S A. 2004 Mar 9;101(10):3632-7. PubMed.

Further Reading


  1. . Increased expression of the amyloid precursor beta-secretase in Alzheimer's disease. Ann Neurol. 2002 Jun;51(6):783-6. PubMed.
  2. . Beta-secretase activity increases with aging in human, monkey, and mouse brain. Am J Pathol. 2004 Feb;164(2):719-25. PubMed.
  3. . Elevated beta-secretase expression and enzymatic activity detected in sporadic Alzheimer disease. Nat Med. 2003 Jan;9(1):3-4. PubMed.

Primary Papers

  1. . Differential utilization of upstream AUGs in the beta-secretase mRNA suggests that a shunting mechanism regulates translation. Proc Natl Acad Sci U S A. 2004 Mar 2;101(9):2794-9. PubMed.