Certain inherited genetic mutations lead to Alzheimer’s disease (AD), but they are relatively rare. A recent study from my laboratory, however, shows that gene alterations that are not passed along by one’s parents may also play a key role in triggering the disease. This happens as a result of a process that occurs in the cell nucleus, known as gene recombination (GR), which can make changes to the DNA “blueprint” in human neurons.

Neuronal GR acts on a gene called APP (amyloid precursor protein), which plays a central role in Alzheimer’s, producing thousands of APP DNA variations. Such variations can occur in the normal brain but are altered further in AD. If our data are confirmed, this would indicate that recombination in these neurons may be involved in the disease process that leads to Alzheimer's. And our findings point as well to a class of existing medicines, approved for other disorders, which can interrupt GR and thus might be used to treat Alzheimer’s.

Historically, brain cells—and most cells of our body—have been thought to contain an identical DNA blueprint, or genome. We knew of one exception in cells of the immune system—B and T cells—that were the first and thus far only cell types known to undergo somatic GR, meaning that GR changes are not passed to offspring, unlike germ line changes affecting sex cells.

In the immune system, gene recombination creates specialized receptors recognizing self from non-self (technically, immunoglobulins and T-cell receptors formed by GR). The discovery of GR in the immune system by Susumu Tonegawa in the 1970s was preceded by theoretical work and observations on fish nervous systems suggesting that recombination might be relevant for the brain. However, unlike the immune system, there was no molecular candidate for GR in fish let alone humans, and the notion of gene recombination in the brain languished.

But at the start of the 21st century, researchers uncovered a harbinger for GR. We discovered that DNA sequences vary from cell to cell, meaning that our brains are a vast mosaic of distinct genomes, a phenomenon aptly referred to as “genomic mosaicism.” These changes are distinct from epigenetic changes that do not directly affect DNA sequences. Scientists have now identified multiple sequence changes that are quite varied and seemingly random, consisting—in order of decreasing size—of entire chromosomes (aneuploidies), smaller copy number variations, even smaller LINE1 retrotransposon repeat elements and single nucleotide variations that alter individual nucleotides.

Brain genomic mosaicism thus exists, but what is it good for and how does it work? General observations have lent support to the impact of mosaicism on gene expression and cell survival. However, specifically altered genes have not been identified to date. Of note, a number of candidate genes were examined over the years for GR—genes for olfactory receptors and certain cell adhesion proteins. Other approaches identified DNA strand breaks in neural genes during early mouse brain development that might be involved with gene recombination. However, once again, no proven genes emerged.

Without a bona fide candidate protein or gene, this research is like looking for the proverbial needle in a haystack. Moreover, immune cells—most notably, immune cell tumors—can grow identically as a cell divides through mitosis (or “clonal expansion”) to amplify the same genome and thus allow its analysis by conventional means, contrasting with neurons that do not continue to divide. Assessments at the level of single or several cells are therefore essential to understand GR. The problem that must be addressed can be illustrated through an analogy: a brown paint might be homogenously composed of brown pigment molecules, or instead formed by colorful pigment molecules that also appear brown when mixed.

Taking all this into account led us to conduct studies assessing mosaicism in AD brains. Our findings showed greater mosaicism—in particular, for increased APP copy numbers. Most notably, segments of DNA in the APP gene were found to not only be amplified in some neurons, but certain APP segments increased in number more than others, hinting at GR.

We therefore closely analyzed the APP gene using nine distinct technical approaches applied to single cells or small collections of neurons to account for genomic mosaicism from normal and AD brains. All these analyses yielded the same conclusions, discovering thousands of new APP variations characterized by an array of different sequence changes within the genomic DNA blueprint that resembled what are known as complementary DNAs. These CDNAs, for short, are copies of RNA molecules that provide the code for making proteins.

Involvement of a famous enzyme called reverse transcriptase discovered by David Baltimore and Howard Temin, appeared to create cDNAs that inserted themselves back into the genome (gencDNAs), a process that is different from immune system GR, which does not involve reverse transcriptase, and occurs in mitotic cells. In neurons, even a single gene can apparently give rise to many thousands of distinct forms through this process, vastly increasing genomic diversity.

Gene recombination occurs in response to stimuli that can be broadly thought of as a form of recording of a cellular event using gencDNAs. Subsequently, gencDNA “playback” may have the advantage of not requiring as much time and energy as the normal process of transcribing a gene into a protein. For instance, GR may be tied to sensory stimuli like sight, sound, taste, touch and scent, as well as internal neurochemical factors—even drugs—that could effectively record and store produced gencDNAs while later allowing their playback by the same or perhaps different stimuli.

In Alzheimer’s, our research implicates an instance of GR gone wild, producing APP gencDNAs with markedly increased numbers and forms, including nucleotide changes identical to those found in inherited AD mutations but occurring somatically and mosaically only in AD neurons. The existence of these myriad APP variations may explain past therapeutic trial failures, which were unable to target the multitude of varying molecular entities. The involvement of reverse transcriptase suggests the possibility of new therapeutics aimed at inhibiting the enzyme.

In fact, there is some evidence already that HIV patients who have been taking reverse-transcriptase inhibitors for many years may have a lower incidence of Alzheimer’s as they age. In principle, FDA-approved medications such as reverse-transcriptase inhibitors could be used today and may have special benefit for people in high-risk categories for whom no effective treatments currently exist. GR affecting different genes may underlie one or more of hundreds of other brain diseases and might also affect other cell types beyond the brain.