Rev Neurosci. 2017 Jan 1;28(1):21-30.

Brain metabolic DNA in memory processing and genome turnover.

Giuditta A, Grassi-Zucconi G, Sadile AG.

Department of Biology, Federico II University, Via Mezzocannone 8, I-80134 Napoli, Italy


Sophisticated methods are currently used to investigate the properties of brain DNA and clarify its role under physiological conditions and in neurological and psychiatric disorders. Attention is now called on a DNA fraction present in the adult rat brain that is characterized by an elevated turnover and is not involved in cell division or DNA repair. The fraction, known as brain metabolic DNA (BMD), is modulated by strain, stress, circadian oscillations, exposure to enriched or impoverished environment, and notably by several training protocols and post-trial sleep. BMD is frequently localized in glial cells but is also present in neurons, often in the perinucleolar region. Its distribution in repetitive and non-repetitive DNA fractions shows that BMD differs from native DNA and that in learning rats its profile differs from that of control rats. More detailed knowledge of the molecular, cellular, and time-dependent BMD features will be necessary to define its role in memory acquisition and processing and in the pathogenesis of neurologic disorders.
PMID: 27665419



To explain the meaning and prospective of the recently published review on brain metabolic DNA (BMD) it is of help to start with a summary of the main questions raised and experimental data obtained. In my lab, interest in brain DNA initially regarded the vertical lobe of octopus brain that was involved in learning and was densely populated by small neurons while the subesophageal lobe was a high motor centre that contained many large neurons. It turned out that during octopus lifetime nuclear DNA remained diploid in the former lobe but progressively increased in the latter lobe to the point of gaining 60% more DNA in adult specimens. Feulgen analyses of individual nuclei confirmed these differences and demonstrated the hyperdiploidy of subesophageal DNA (1,2). At that time, neuronal hyperdiploidy was disputed (for a review, see 3), but mammalian brain has been recently described as a genomic mosaic (4) containing considerable extra DNA specially in the prefrontal cortex.

The nature of hyperdiploid subesophageal DNA was investigated by comparing it with the diploid vertical DNA. Unfortunately, data were mislaid and were luckily recovered only recently. They suggest that hyperdiploid DNA is enriched in AT sequences that appear to be present in the cytoplasm (5). As to the origin of this DNA, the required administration of radiolabeled precursors to living octopus seemed a more difficult operation than that feasible in laboratory rats. Opting for the mammalian brain made us aware of Pelc’s pioneering data on metabolic DNA synthesized during cellular activity (6) and of Reinis’ striking demonstration that brain DNA synthesis markedly increased in mice learning a passive avoidance task (7, 8). We were led to examine the effect of electroconvulsive shock (ECS) on rat brain since ECS was known to massively stimulate the brain. We found instead that DNA synthesis was strongly inhibited for about 40 min before returning to control values (9). The data were nonetheless in line with Pelc’s interpretation since the ECS strong but very brief stimulation was known to produce an almost immediate, sustained brain silencing. Hence, the inhibition of DNA synthesis had only reflected the period of brain silencing, and brain stimulation had been too short to increment DNA synthesis.

Reinis’ results had also shown that the learning effect persisted for days despite the waning of newly synthesized DNA. These results prompted biochemical analyses in rats kept under normal conditions that confirmed the high turnover of brain metabolic DNA (BMD) (10), and in rats learning a two-way active avoidance task (11). In the latter experiment, control rats were kept in the same room in which training was taking place. As we later learned, this arrangement made them receive the distressing signals (odors and sounds) emitted by the training rats, that were obviously more intense when coming from non-learning rats receiving more foot-shocks. Data were expressed as brain to liver ratios of BMD synthesis to correct for the variable amount of injected radiolabeled thymidine, since liver DNA synthesis was not affected by training. Ratio were significantly higher in learning rats than in their control mates (3.5±0.8* vs 1.1±0.3), but were higher in non-learning rats and in their controls; in addition, the latter difference lacked significance (5.1±1.3 vs 3.8±1.2). These results confirmed the learning BMD increase demonstrated by autoradiographic analyses, but also indicated that non-learning rats subjected to a more distressing experience responded with an even higher BMD synthesis. A comparable effect also regarded the control rats that received a higher number of distressing signals from non-learning rats. In other words, BMD synthesis was also enhanced by stress, and this effect was stronger when stress was directly perceived but still of considerable intensity when it was indirectly felt.

The peculiar fate of the BMD synthesized by non-learning rats was revealed when the latter data were compared with those determined in non-learning rats that were allowed a 3 h post-trial sleep after completing a comparable training session. The comparison was suggested by the sleep processing of newly acquired memories that initially requires the involvement of slow wave sleep (SWS) and eventually paradoxical sleep (PS). The former sleep was known to reverse the synaptic modifications storing familiar memory traces, thus reinstating them to their native condition, while remaining adaptive memories were transferred from their synaptic stores to more readily retrievable brain places (12). Recently, the latter location was suggested to be the quantum layer (13).

This premise will hopefully clarify the interpretation of data demonstrating that non-learning rats lost 50% of their newly synthesized BMD in post-trial sleep while none was lost by learning rats (14). In addition, SWS episodes followed by PS were inversely correlated with BMD in the post-trial sleep of non-learning rats, while no correlation regarded learning rats (15). These contrasting effects were assumed to indicate that newly synthesized BMD was a molecular correlate of all acquired memories that were prevalently adaptive memories related to avoidances in learning rats, but prevalently innate memories related to escapes and freezings in non-learning rats. Accordingly, BMD correlates were kept if related to avoidance memories but lost if related to innate responses.

Data on the enhanced BMD synthesis in rats learning the two-way active avoidance task were in agreement with comparable results obtained by Ashapkin et al. (16). Likewise, the biochemical data regarding mice learning a passive avoidance task confirmed the autoradiographic data regarding the same learning condition (8). Indeed, the brain to liver ratios of BMD synthesis was more than two-fold higher in learning mice than in active and passive controls (respectively 1.7±0.4 vs 0.7±0.1* and 0.8±0.2*) (17). These increments were at variance with the significant decrements in BMD synthesis determined in rats learning appetitive or spatial habituation tasks. With regard to the former condition exemplified by the reverse handedness training (18), the lower BMD synthesis was attributed to the training task itself that did not require punitive foot-shocks. It can however be noted that in the latter experiment, BMD synthesis was determined in the last training session in which the rat behavior had not significantly improved with respect to the previous session, at variance with the marked increment recorded in the two preceding sessions. Hence, the decrement in BMD synthesis of learning rats could also be attributed to their acquired familiarization with the novel condition. This interpretation cannot detract from comparable learning paradigms also inducing decreased BMD synthesis, such as in spatial habituation learning in which BMD synthesis was lower in some brain regions but higher in other regions.

Data from the reverse handedness experiments are relevant on two additional counts: i) BMD was also present in the mitochondrial fractions, and ii) BMD from control and learning rats were widely distributed in DNA fractions differing in their degree of repetitiveness. The former data demonstrated the BMD presence in cytoplasmic particles rather than only in nuclei, as expected from the conventional belief. A mitochondrial BMD localization had been glimpsed in the early seventies, but data were attributed to leakage from the nuclear fraction. This possibility could not be applied to the reverse handedness experiment since mitochondrial BMD exhibited a specific activity much higher than nuclear BMD. As shown in those previous experiments mitochondrial BMD could not be identified as mitochondrial DNA since the expected correspondence between BMD and cytochrome oxidase was lacking (19).

The BMD distribution in DNA fractions differing in their degree of repetitiveness was determined by the Cot analytical procedure that was based on the initial fragmentation of DNA, the thermal separation of the two strands, and their following reannealing. Last process was concentration-dependent and time-dependent, as indicated by Co (concentration) and t (time). This meant that double stranded chains that reannealed at different times had to be separated from single stranded chains. Data demonstrated that the highly repetitive DNA fraction recovered at a low Cot value contained an excess of control and learning BMD with respect to native DNA. In addition, the excess was larger for control BMD. More relevantly, at very high Cot values (50,000) the amount of learning BMD that remained single stranded was much higher than in control BMD. The latter difference indicated that learning BMD might harbor a different sequence than control BMD. Comparable data were reported with regard to rats learning a two-way active avoidance task but their analyses only regarded low Cot values (16).

We should now mention that, according to Pelc (6), metabolic DNA was shown to replace the nuclear DNA that was degraded (or possibly transferred to other cells) during the increased cell activity. The most striking demonstration of this turnover came from experiments on the adrenal gland whose activity was kept high for a sustained period of time by exposing rats at 4 °C 15 hours a day for several days (20-22). Under these conditions, metabolic DNA was synthesized during the 9 hours rats were kept at room temperature and nuclear DNA was progressively lost. This regime produced an almost unbelievable 43% loss of nuclear DNA in Italico rats, a 24% loss in Long-Evans rats, a 13% loss in Sprague Dawley rats and an 8% loss in Wistar rats. The lost DNA was progressively recovered when the cold regime was stopped and rats were kept at room temperature all day. Strain differences were attributed to the different balance in rates of DNA loss and DNA synthesis. In simpler words, adrenal nuclear DNA behaved as the water level of a tank receiving water and loosing water. The key triggering the process was in the hands of cell activity which started to lose DNA beyond a certain threshold.

Why the activated DNA started to be deleted remains a problem. Possibly the question was not sufficiently emphasized. Hence, we may assume that beyond that threshold it could have been more fruitful to discard existing DNA unable to sustain cell efforts in view of its forthcoming replacement by newly synthesized, possibly better adapted DNA. Something analogous to the stressed adrenal gland may be conceived to regard the BMD of a learning rat, but things are not entirely the same. Learning to avoid a punishing footshock or to obtain food under a different condition requires behavioral responses whose novelty cannot be neglected. Nonetheless, by thinking it over a bit more, some novelty may also appear in the adrenal response to a lasting stress just in view of its unusual duration. True, the adrenal response might only involve an increment in the number of DNA segments turning over but a comparable quantitative support could also regard the brain cells of the learning rat if we keep in mind that the BMD we determine derives from an extremely heterogeneous population of brain cells rather than from the much more homogeneous adrenal cells. In brain, each cell is likely to respond differently, that is in its own way to incoming stimuli from the heterogeneous, changing environment. On the face of this marked cellular heterogeneity, we are merely analyzing average BMD data produced by an entire brain region.

Before describing results suggesting that comparable relations exist between BMD and brain DNA content, let me mention a further set of data that still concern the adrenal response but regard rats that were fed with the diet for another strain. In brief, if the progeny of an Italico rat suffering a 40% loss in adrenal DNA is fed the diet of a Wistar rat responding to the same stressful condition with only an 8% DNA loss, the first generation of Italico rats exhibits a less intense response, the second generation a still lower response, and the third generation a response comparable to that of a Wistar rat. A comparable reversal in the intensity of the adrenal response occurs in the progeny of Wistar rats fed the diet of Italico rats. In conclusion, the ingested food modulates the balance of DNA inflow and outflow in adrenal nuclei. To my knowledge, no information is available on the nature of diet components responsible for these effects.

Let me now mention experiments suggesting the presence of comparable relations between BMD synthesis and brain DNA content. They regard the circadian oscillations of these variables determined every 4 h in adult rats kept under normal conditions. The BMD maximum (acrophase) occurred in the dark period (11 pm), that is in the waking period of this species. When compared to the lowest BMD synthesis, the acrophase represented a 48% increase (23). Circadian oscillations of BMD synthesis were also determined in rats exposed to enriched or impoverished social and environmental conditions for a few days. In the enriched group the circadian BMD synthesis became sharper than in rats kept under normal conditions. The acrophase still occurred in the nocturnal period (11 pm) and exhibited a 28% increase with respect to the minimum. At variance with these results, in rats exposed to the impoverished condition a higher BMD synthesis occurred in the late diurnal period with respect to the enriched condition, the acrophase occurred earlier (at 7 pm) and the increment over the lowest value was only 8% (24).

These results were reproduced in a comparable circadian experiment in which BMD synthesis was determined together with brain DNA content. BMD values were largely reproducible: in the enriched condition, the acrophase occurred at 11 pm and exhibited a 27% increment with respect to the lowest value while in the impoverished condition the acrophase occurred at 7 pm and exhibited a much lower increment (10%). As to brain DNA content, its acrophase in the enriched condition also occurred in the dark period but with one hour delay from the BMD acrophase. Its increment over the lowest value was 18%. Conversely, in the impoverished condition, a pseudo-acrophase was detected at 3 am with an increment of only 6 % with respect to its lowest value (25).

Overall, the data underline the key role played by social and environmental stimuli in shaping and intensifying the circadian oscillations of BMD synthesis and brain DNA content.  The comparable oscillations of BMD and DNA content present in the enriched condition and the delayed appearance of brain DNA acrophase with respect to the BMD acrophase support the possibility that a genomic turnover may be also occurring in the adult rat brain.

We come now to a brief summary of recent data on the properties of mice BMD that regard its cytoplasmic localization, prevalence in glial cells, and origin from reverse transcription. The first of these features, initially hinted in the seventies (19) and demonstrated in the eighties (18), was confirmed in CD-1 mice subcutaneously injected with BrdU two hours before analysis. When BrdU specific antibodies emitting a green fluorescence challenged the brain homogenate and the related nuclear and mitochondrial fractions, the green fluorescence was exclusively present in cytoplasmic particles that could readily be distinguished from the blue colored nuclei stained by the DNA specific Hoechst dye. Only a very few nuclei appeared of a somewhat paler blue appearance that could indicate the presence of a colocalized green fluorescence.

Additional data regarded the colocalization of the green fluorescence of BrdU specific antibodies with the red fluorescence of anti-glial fibrillary acidic protein antibodies. Data were in agreement with the prevalent BMD presence in glial cells detected by autoradiographic analyses (7,8), but they also indicated the astroglial nature of the BMD-containing glial cells that are known to envelop brain synapses. In the squid, perisynaptic glial cells transferring newly synthesized RNA to presynaptic regions (26) have recently been shown to also transfer newly synthesized BMD (27).

The origin of BMD by reverse transcription was initially suggested by the higher rate of reverse transcription determined in fast learning rats than in slow learning rats, and by its comparable higher rate in learning rats than in control rats (28). The BMD origin from reverse transcription had been indicated by the temporary localization of newly synthesized BMD in cesium density regions known to harbor RNA-DNA hybrids, but data were not properly interpreted until recently (19). These observations led to analyses demonstrating the presence of hybrids between newly synthesized BMD and native RNA by the colocalization of the green fluorescence of BrdU specific antibodies with the red fluorescence of RNA-DNA hybrid-specific antibodies. Comparable data were also obtained from dot-blots of DNA and RNA purified from nuclear and cytoplasmic fractions of mouse brain treated with the same RNA-DNA hybrid-specific antibodies.

What perspectives are opened by this marginalized wealth of data that has so suddenly appeared? The first consideration that comes to mind concerns the consolidated belief in genome stability. This concept is unexpectedly gaining a dynamic dimension previously hidden but apparently present in all cells (29,30), but notably in brain cells (31). DNA is definitely modulated by the cell metabolic life which spurs its synthesis when cell activity gets high. On these occasions, DNA segments providing inadequate support to cell responses are replaced by new DNA segments. Is this turnover only mediating an increase in the number of DNA segments, as Pelc believed (6), or is it also including novel changes as BMD may suggest? If so, how is the novelty encoded? Preliminary evidence supports the quantitative hypothesis  but also its qualitative extension. Future experiments should aim at a rigorous examination of both possibilities.

What is at risk is the pervasive but unconvincing neoDarwinian theory of biological evolution based on the random DNA mutations in germ cells that natural selection should allow. From the little bit we know, the genomic turnover triggered by cell activity requires that the new DNA is reverse transcribed from the template RNA selectively present during the cell response. The entire fleeting information to be encoded in metabolic DNA is held by the cell response. Is the template RNA sufficient to preserve all that is needed for storing novel traits or additional information is coming from somewhere else? Electromagnetic waves have been shown to act as template of DNA (32), and brain cells with their intercellular fluid are immersed in a comparably fleeting electromagnetic field. It would be of interest to compare BMD sequences with those of template RNA. If sequence analysis will detect novelty in both, the deprecated but illuminating switch to a neo-Lamarchian mechanism will become likely. Of course, future studies will have to examine the transfer of novel BMD segments to germ cells and the progeny, quite a long journey that may possibly be preceded by epigenetic modifications.

Nonetheless, what seems already in our hands concerns the molecular biology dogma claiming the monodirectional flow of information from randomly modified DNA to proteins. Together with the now obsolete Weissmann’s dogma negating the information flow from somatic cells to germ cells, biology has been compelled to accept the notion that environmental information could reach somatic cells to allow adaptation but not the holy seat of the double strands. The situation looks like to have been upturned: the adapted cell phenome may be transferring to the double strands the new information it has produced. Perhaps the time is approaching that genomic impression may be standing on the same plane of genomic expression since phenome is including the genome in its persisting spiraling with the environment, a long-expected step (33).

From the less demanding but equally relevant medical point of view, BMD involvement in memory processing (18) is also likely to open intriguing perspectives. They are centered on the possibility that instructive BMD changes may be detected during the pathogenetic course of a wide range of health disorders studied in animal models, including the chronic neurodegenerations. If increments in cell activities are being coded in metabolic DNA, notably BMD, maladaptive cell responses leading to more serious derangements may hopefully be detected at early enough times to interfere with forthcoming modifications by acting on the presumed causes possibly by suitable modifications of the organism’ diet () and living conditions.



  1. De Marianis B and Giuditta A (1978) Separation of nuclei with different DNA content from the subesophageal lobe of octopus brain. Brain.Res. 154:134-136.
  2. De Marianis B et al. (1979) Excess DNA in the nuclei of the subesophageal region of octopus brain. J.Comp.Neurol. 186:293-300.
  3. Giuditta A (1983) Role of DNA in brain activity. Handbook of Neurochemistry, 2nd edition. Lajtha A ed. (Plenum Press) 5, pp. 251-276.
  4. Bushman DM and Chun J (2013) The genomically mosaic brain: aneuploidy and more in neural diversity and disease. Semin. Cell. Dev. Biol. 24, 357-69.
  5. Giuditta A et al. (2017) Hyperdiploid DNA from octopus brain is enriched in AT sequences. Rend. Acc. Sc. fis. mat. Napoli, LXXXII, 149-152.
  6. Pelc SR (1972). Metabolic DNA in ciliated protozoa, salivary gland chromosomes, and mammalian cells. Int. Rev. Cytol. 32, 327-355.
  7. Reinis S and Lamble RW (1972). Labeling of brain DNA by 3H-thymidine during learning. Physiol. Chem. Phys. 4, 335-338.
  8. Reinis S (I972). Autoradiographic study of 3H-thymidine incorporation into brain DNA during learning. Physiol. Chem. Phys. 4, 391-397.
  9. Giuditta A et al. (1978). Effect of electroshock on thymidine incorporation into rat brain DNA. J. Neurochem. 31, 83-987.
  10. Scaroni R et al. (1983). Synthesis of brain DNA during acquisition of an active avoidance task. Physiol. Behav. 30, 577-582.
  11. Perrone Capano C et al. (1982) DNA turnover in rat cerebral cortex. J.Neurochem. 38:52-56.
  12. Giuditta A et al. (1995) The sequential hypothesis of the function of sleep. Behav.Brain Res. 69:157-166.
  13. Giuditta A (2014) Sleep memory processing: the sequential hypothesis. Front. Syst. Neurosci. 8, 219. Doi: 10.3389/fnsys.2014.00219.
  14. Giuditta A et al. (1985) Effect of sleep on cerebral DNA synthesized during shuttle-box avoidance training. Physiol.Behav. 34:769-778.
  15. Langella M et al. (1992) The sequential hypothesis of sleep function. IV. A correlative analysis of sleep variables in learning and non-learning rats. Physiol.Behav. 51:227-238.
  16. Ashapkin VV et al. (1983). Selective DNA synthesis in the rat brain induced by learning. Biokhimija 48, 355-362.
  17. Ambrosini MV et al. (1985) Brain DNA synthesis during learning. Riv.Biol. 78:441-444.
  18. Giuditta A et al. (1986) Synthesis of rat brain DNA during acquisition of an appetitive task. Pharmacol.Biochem.Behav. 25:651-658.
  19. Rutigliano B and Giuditta A (2015) The unexpected recovery of misplaced data on brain metabolic DNA. Rend. Acc. Sc. fis. mat. Napoli, LXXXII, 99-106.
  20. Viola-Magni MP (1966). An analysis of DNA loss and synthesis in the rat adrenal medulla nuclei upon cold stimulation. J. Cell Biol. 30, 213-25.
  21. Tongiani R and Viola-Magni MP (1969). Differences in adrenal medulla nuclear DNA content among rats of different strains following intermittent exposure to cold. J. Cell Biol. 42, 452-459.
  22. Pelc SR and Viola-Magni MP (1969). Decrease of labeled DNA in cells of the adrenal medulla after intermittent exposure to cold. J. Cell Biol. 42, 460-468.
  23. Grassi Zucconi G et al. (1988) Circadian rhythms of DNA content in brain and kidney: effects of environmental stimulation. Chronobiol. 15:195-204.
  24. Grassi Zucconi G et al. (1988) Circadian oscillations of DNA synthesis in rat brain. Brain.Res. 447:253-261.
  25. Grassi Zucconi G et al. (1990) Environmental stimuli modulate the circadian rhythm of [3H- methyl] thymidine incorporation into brain DNA of male rats. Behav.Brain Res. 41:103-110.
  26. Eyman M et al. (2007) Local synthesis of axonal and presynaptic RNA in squid model systems. Eur. J. Neurosci. 25: 341-350.
  27. Cefaliello C et al. (2015) Newly synthesized DNA in squid nerve terminals. Rend. Acc. Sc. fis. mat. Napoli, LXXXII, 61-64.
  28. Salganik RI et al. (1983). Probable role of reverse transcription in learning: correlation between Hippocampal RNA-dependent DNA synthesis and learning ability in rats. Neurosci. Lett. 36, 317-322.
  29. Stroun M et al. (1967). Metabolic DNA in heart and skeletal muscle and in the intestine of mice. Nature 216, 716-717.
  30. Gahan PB et al. (2008). Metabolic DNA as the origin of spontaneously released DNA? Ann. N.Y. Acad. Sci. 1137, 7-17.
  31. Giuditta A et al. (2017) Brain metabolic DNA in memory processing and genome turnover. Rev. Neurosci. Jan 1; 28: 21-30. doi: 10.1515/revneuro-2016-0027.
  32. Montagnier L et al. (2015) Transduction of DNA information through water and electromagnetic waves. Electromagn Biol Med.34(2):106-12. doi: 10.3109/15368378.2015.1036072.
  33. Giuditta A (1982) Proposal of a “spiral” mechanism of evolution. Riv.Biol. 75:13-31.