This web page was produced as an assignment for an undergraduate course at Davidson College.

Garrett Smith's Genomics Web Assignment Three:

Review of Badiola et al.'s, The Proton-Pump Inhibitor Lansoprazole Enhances Amyloid Beta Production

Article Summary

        The goal of this study was to document the effects of proton pump inhibitors (PPIs) – one class of drugs commonly used for the treatment of peptic ulcers and gastroesophageal reflux disease – in Aβ protein production in both in vitro and in vivo Alzheimer's disease (AD) models. Findings from both models differed somewhat but generally demonstrated that PPIs exacerbated Aβ production, which suggests that PPIs may contribute to an acceleration in the onset of AD in humans, as the extracellualr accumulation of certain forms of Aβ is a hallmark of AD. This finding was particularly notable given that PPIs are the third most common class of drug prescribed in the United States (Gardner 2010), suggesting a potentially substantial influence of these drugs in AD pathology.

        In a first set of experiments, which were conducted in vitro, researchers exposed Chinese hamster ovary (CHO) cells (a model mammalian cell commonly used to test interactions of recombinant proteins) expressing genes for wild type human amyloid precursor protein (APP; a protein of uncertain function that can be degraded into different forms of amyloid, including Aβ as well as a soluble form of APP) and presenilin 1 (PS1; which plays a major role in cleaving APP into its soluble form and its amyloid products) to either the PPI lansoprazole or the vehicle for the drug. Researchers main finding here was a positive dose-dependent increase in Aβ40 and Aβ42 expression following lansoprazole administration relative to vehicle. APP cleavage normally results mostly in Aβ40, but Aβ42 is thought to contribute more directly to AD pathology. Other PPIs including omeprazole, pantoprazole and esomeprazole also showed similar effects in increasing Aβ42 levels, though these effects were less pronounced and differed slightly by drug.
        Analyses of specific protein content of these cells showed relative increases in Aβ42,
Aβ40, and Aβ37; and decreases in Aβ38 following lansoprazole administration relative to vehicle administration, implying that lansoprazole was an inverse GSM (iGSM), a class of drugs which tend to increase Aβ42 and decreases Aβ38 by modifying activity of the protein γ–secretase. This was substantiated following when lansoprazole was found to neutralize the effects of R-flurbiprofen (a straight GSM which by itself decreases Aβ42 and increases Aβ38) when the two were co-administered, and still manage to elevate Aβ42 levels. Researchers also found that lansoprazole likely increased Aβ40 and Aβ37 content by elevating activity of the protein BACE1 as elevated levels of one of this protein's product (sAPPβ) were found following lansoprazole administration in contexts while levels of this protein itself remained the same relative to controls.

        In a second set of experiments, conducted to test the effects of
lansoprazole in vivo, researchers injected lansoprazole into wild type and AD triple-transgenic (3xTg-AD) mice every day for five days at doses of either 20 or 100 mg/kg/day.  These 3xTg-AD mice normally display age-related increases in tauopathy and amyloid plaques, much like in human AD. Soluble Aβ40 levels were found to increase with 100 mg/kg doses in wild type mice and following both doses in 3xTg-AD mice. Increases in Aβ42 were found to be non-significant in both wild type and AD models following lansoprazole administration relative to vehicle.

        Overall the study demonstrated that
lansoprazole (and other PPIs to some extent) can modify the production of Aβ species - one hallmark of AD - in cell cultures, wild type mice and mouse models of AD to levels that in some instances reflect those found in brains of humans with AD. In vitro evidence suggested that lansoprazole exerted its effects by modulating the γ–secretasecomplex and increasing activity of BACE1.  Of particular note were the findings that in mouse AD models, human equivalent doses of lansoprazole, even when administered over a relatively short time, could lead to measurable changes in some Aβ species. This study thus begin to target lansoprazole and other PPIs as potential exacerbators of AD pathology.


        Overall, this paper established a logical though maybe not comprehensive progression of ideas to suggest that proton pump inhibitors (PPIs), namely lansoprazole, may contribute to the accumulation of different Aβ species - some of which are implicated in Alzheimer's disease (AD) pathology. In vitro experiments first established a clear potential of the drug and other PPIs to affect this Aβ accumulation then went so far as to begin to establish a likely basis by which lansoprazole exerted its effects. Administration of lansoprazole to mice (wild type and AD model) revealed slightly different effects than were observed in culture that it seems could have been more thoroughly explored, but overall this study established a firm rationale for future studies into the effects of PPI (at least lansoprazole) in AD pathology.

        In terms of methodology, it did not necessarily seem that the researchers' chosen in vitro model of Chinese hamster Ovary (CHO) would provide the best representation of how Aβ might actually behave at the cellular level within neurons. Though I don't claim to be well aware of the range of model neural cells available, I imagine model neurons or neural precursors capable of cell division may exist (for glutamatergic or cholinergic neurons in particular, as loss of these cells are strongly implicated in AD pathology) that the researchers might have used. It might have been helpful for the researchers to have perhaps performed such studies and noted them in supplementary material. Using other model cells may have minimized some of the differences in Aβ expression observed between in vitro and in vivo models observed in the study. Perhaps the choice made for the paper was the most economical, but I can't be sure.

        The researchers also note that some of the differences they observed between cellular and mouse models might be explained by "Aβ quantization," given that they only measured extracellular Aβ in cells but they measured extra- and intracellular Aβ in whole brain homogenates. General results were reported to be similar between models (that lansoprazole altered Aβ production in cellular and animals models); however, I was curious as to why the researchers chose not to measure intracellular Aβ in cell cultures. No direct explanation is provided for this choice, though the choice may stem from the fact (visualized in figure 4) that non-soluble Aβ species are always made extracellular. It is noted that "the complete results of this study will be published elsewhere," so maybe more studies were underway to control for those discrepancies between experimental approaches.

        The fact that the hypothesized most toxic forms of Aβ (Aβ42) were not statistically significantly increased in wild type or AD mouse models following lansoprazole administration (relative to vehicle) fails to strengthen the case that PPIs contribute to AD pathology; however, it may just be that the drug was not administered over a long enough time span to see its effects in these animals. Longer-term studies would thus have been ideal as well to demonstrate increases of Aβ especially in at the human equivalent dose of lansoprazole. The researchers note that cognitive impairment and Aβ burden increase in 3xTg-AD after 2-3 months, but by 8 months these mice show significant levels of intracellular and soluble Aβ. It would have been insightful to have data of how lansoprazole may influence the rate at which these compounds accumulate over these longer time spans, as this would demonstrate greater pertinence to scenarios in which humans are exposed to the drug. Also though it is noted that Aβ40 production was significantly greater in non-transgenic mice, little discussion is provided for why this may be.

        Though it wasn't a behavioral neuroscience lab that conducted this research, the case that PPIs may exacerbate AD pathology would certainly have been more compelling had the team behind this study provided results from behavioral tests for changes in cognitive performance (e.g. memory) in mice following increased Aβ accumulation after lansoprazole administration. Even if lansoprazole had not led to the accumulation of Aβ species thought to contribute to AD pathology (Aβ42), cognitive decline in mice (wild type or AD model) given lansoprazole (relative to those given vehicle) would have demonstrated a notable detrimental effect of drug, especially if it were given for only for a short time. Results of these studies may released elsewhere in the future. Even if researchers had not chosen to perform behavioral tests, measurements on the effects lansoprazole or the other PPIs on tau hyperphosphorylation might have also been informative, as this phenomenon is also associated with AD pathology. Perhaps these results will be published in the future as well.

        A few times throughout the report, authors report that two means were different but not significantly different (e.g. "Levels of soluble Ab42 were also slightly increased, although they did not reach statistical significance" and "Similarly to the non- transgenic mice, we also observed a moderate increase in soluble Ab42 levels, although they were not statistically significant either"). This commentary seems largely unnecessary without noting p values (i.e. how close they may be to .05 or .01), which might have suggested that a statistically significant effects might have been observed with a larger sample size. Also, no experiments were conducted in vivo for effects of the other PPIs surveyed in vitro, which makes it difficult to generalize that PPIs may be involved in AD pathology. It might have been at least helpful to have conducted in vivo trials for omeprazole as it is the most commonly prescribed PPI (Jones et al., 2001).

        Regarding more trivial concerns I had with the paper, in figure 2 part E it was curious as to why authors chose not to provide some sort of loading control. Also, the caption of figure 3 seems to repeat a mistake in stating that mice were administered "100 kg/mg" of lansoprazole (a substantially different dose than was reported in the methods).

Figure Summaries

Figure 1

        Researchers first wanted general in vitro data regarding how PPI administration influenced manufacture of
Aβ proteins. Figure 1 portrays how administration of different PPIs at varying concentrations for 24 hours each affects expression of Aβ40 and Aβ42 relative to vehicle (control) as measured by ELISA immunoassays of PS70 cells. The same effects after DAPT (a γ–secretase inhibitor known to suppress the manufacture of Aβ and thus serve as a positive control condition) administration are also shown. Bar area above or below the dotted horizontal line indicates degree of difference from vehicle ("+" and "*" indicate significant differences at p<0.05 and p<0.01 respectively). Part A includes information only from lansoprazole and part B includes information only from the other drugs tested, reporting only results form higher drug dosages (relative to those used in part A).
        No particular drug appeared to contribute to a significant increase in Aβ40 at any of the concentrations employed relative to vehicle (no different from 100% of vehicle value, given by horizontal dotted line). Though data trended toward increases in Aβ40 at the three higher lansoprazole doses, Aβ42 showed clear dose-dependent increases with most drugs, with lansoprazone (part A, right graph) contributing to the greatest magnitudes of Aβ42 increase of all the drugs at its highest and second highest doses (about 300% and 200% respectively). These results established lansoprazole as the most likely candidate to contribute to amyloidosis as seen in AD, as Aβ42 is more strongly implicated in AD pathology.

Figure 2


        To follow up on results shown in Figure 1 that lansoprazole was the most likely candidate PPI to contribute to AD amyloid pathology, researchers measured the content of other Aβ species in the supernatant (extracellular) following 24-hour lansoprazole administration to PS70 cells. Part A shows the results of MALDI-MS analysis, where relative amounts of each protein species are shown. Relative to vehicle, lansoprazole administration led to an increase in Aβ42 and Aβ37; and a decrease in Aβ38 (as indicated by height of each correspondingly labeled spike). Proteins were differentiated by their mass to charge ratio (m/z) as listed on the x axis. 2B merely represents a confirmation of the results in 2A using Western blot, showing a relative increase in Aβ42 and a decrease in Aβ38 following lansoprazole administration relative to vehicle; long (exposure) is provided for enhanced visualization of differences. These results suggested that lansoprazole may act as an inverse GSM (as discussed earlier).
        Part C further evidences that lansoprazole acts as an inverse GSM, as combined treatment of cells with the straight GSM R-flurbiprofen (which has the opposite effect of an inverse GSM with respect to
) combined with lansoprazole (lime green column) led to an increase in Aβ42 relative to treatment with R-flurbiprofen or vehicle control alone. Lansoprazole essentially overpowered/outcompeted R-flurbiprofen in terms of Aβ42 proteins produced, suggesting that lansoprazole (at 50 mM) may be capable of mitigating or neutralizing the effects of at least some neuroprotective agents.
       To explore how lansoprazole exerted its unique effects researchers tested whether the drug
increased the quantity of APP or BACE1 available or whether it enhanced the activity of BACE1. Following administration of either 50 mM lansoprazole or vehicle, cells did not demonstrate different quantities of APP or BACE1 (as seen by equal intensity of protein density between conditions in the Western blots of part D), however they did differ in quantity of APPβ, a product of BACE1, implying that lansoprazole functioned by increasing BACE1's activity. sAPPα - a product of α-secretase (involved in a non-pathogenic processing of APP) - remained the same after either treatment (part E).

Figure 3

        Researchers certainly wanted to follow up on their intriguing in vitro results in vivo, and they began this line of inquiry by administering either lansoprazole (100 mg/kg/day) or vehicle alone to either wild type or 3xTg-AD model mice for five consecutive days. 3xTg-AD were also tested at another smaller dose of lansoprazole (20 mg/kg/day).
        Whole brain extracts of wild type mice analyzed with ELISA immunoassays contained significantly elevated levels of that Aβ40 – but not Aβ42 – following lansoprazole administration relative to vehicle (Figure 3A). This is somewhat the reverse of the trend seen in culture (significant elevation of Aβ42 but not Aβ40) following lansoprazole administration. The same assessments of 3xTg-AD model mice brains showed results similar to that of wild type mice. Aβ40 – but not Aβ42 – was found elevated in these brains following lansoprazole administration. Generally results seen following administration of a low dose of lansoprazole to 3xTg-AD mice mirrored those seen from wild type mice given a high dose of the drug (blue columns in B versus green columns in A). Increases in Aβ40 were lansoprazole dose-dependent in 3xTg-AD mice, with higher doses of lansoprazole boosting Aβ40 production by as much as 250%. Such a dose-dependent relationship cannot be determined from the less extensive information given for the wild type mice.

Figure 4