Autophagy (literally “self-eating”) is the body’s way of cleansing cells by recycling old or damaged components and is a process that appears to have a strong relationship to preventing disease and aging.  More technically, it is a highly regulated lysosome-dependent catabolic program used by the body to clear dysfunctional proteins, organelles and other structures that are typically first marked for the process by a protein called “ubiquitin.”  Primarily intra-cellular but also extra-cellular components are engulfed in autophagocytic vacuoles and degraded to constituent molecules, such as sugars, fatty acids and amino acids, which can be then be used as nutrition to produce ATP and/or provide building blocks for the synthesis of essential proteins.   There are different types of autophagy but for the most part they all serve this same purpose.  Mitophagy is a similar process that removes damaged or dysfunctional mitochondria.  There is also nucleophagy of material in the cell nucleus.


A number of types of autophagy are currently described.

  • Microautophagy: direct engulfment of cytoplasmic materials by invagination of lysosomal membrane.
  • Macroautophagy:  classic autophagasome mediated double membrane engulfment.
  • Chaperone Mediated autophagy: extremely selective target protein mediated (Hsc70) recycling that is activated in response to cellular stress.
  • Mitophagy:  autophagy related to mitochondria and their components
  • Also… ribophagy, lipophagy, xenophyagy, nucleophagy, clockophagy, lysophagy, etc.


Why is it important?
Autophagy is critical to cell function, regeneration and keeping the cell working properly.  Inhibition of autophagy by knocking out any of a variety of autophagy related genes invariably accelerates aging and disease.  Similarly, with aging, autophagy is downregulated (particularly in lymphocytes) leading to a forward feeding loop of accumulation of cellular debris and cellular dysfunction / dysregulation.   In contrast, enhanced autophagy delays aging and prolongs lifespan in a range of model organisms including yeast, nematodes, fruit flies and mice.  Lifespan studies in humans have not been conducted given the obvious challenges in conducting experiments in long-lived organisms, but it is generally assumed that regular induction of autophagy is a salubrious process, something we want to promote.  So how to do so?

How to promote autophagy?
First the biochemistry–skip ahead to the punchline if you prefer.  ULK1 is the enzyme that catalyzes production of phagosomes implicated in autophagy.  AMPK and SIRT activate ULK1, MTOR deactivates it.  If MTOR1 is inhibited by rapaymycin or starvation, autophagy is activated.  TFEB is considered the primary transcription factor for the synthesis of both lysosome and autophagosomes.   Again, AMPK boosts the activity of TFEB while MTOR inhibits it.  Spermidine, which declines with age, is a required co-factor in the translation of TFEB, so its supplementation is theorized (and shown) to stimulate autophagy.

Caloric restriction activates autophagy by stimulating AMPK and sirtuin1 as well as inhibiting mTOR.  When an autophagy related gene is knocked out or knocked down, the life-prolonging effect of rapamycin disappears, suggesting that autophagy plays an important role in the life-prolonging effect of rapamycin.

Simply not eating for a few hours and reducing nutrient levels in the blood starts the process of autophagy.  This is the rationale for time restricted feeding.  Starvation for longer periods such as two- or three-days initiates chaperone mediated autophagy, a targeted deep clean that modulates diverse functions such as glucose and lipid metabolism, DNA repair, cellular reprograming and the cellular response to stress.

Drugs to stimulate autophagy

  • Berberine and metformin both boost AMPK, thus phosphorylating and activating ULK1–> autophagy.
  • Rapamycin and spermidine inhibit MTOR, activating autophagy and spermidine, as already mentioned, both blocks motor and also is a necessary cofactor in the translation of TFEB.
  • Resveratrol and NR (nicotinamide riboside) theoretically target SIRT1, though evidence of their bioavailability after oral administration is questionable.
  • Ferulic acid may up-regulate SIRT1 expression and is derived from Angelica sinensis or “female ginseng” and used in traditional Chinese medicine.  It is also widely available as an additive in cosmetic skin creams.  Anthrocyanins in the diet are converted to ferric acid, providing a rationale for their health value.
  • Urolithin-A, a phytochemical derived from pomegranate has also been shown to boost SIRT1, as well as improve mitochondrial biogenesis.
  • Aerobic exercise has also been shown to stimulate autophagy.
  • It turns out these are all the usual suspects in the realm of anti-aging drugs and supplements, and while many of them have diverse actions, it may be that their action on the autophagy pathways is primary to their purported value as anti-aging interventions.  Yet the effect sizes of these interventions are not known, which is to say that we don’t really know to what extent autophagy will be triggered by taking any of these.

Can there be too much autophagy?  Perhaps yes.  While normal or even supernormal levels of autophagy promote cellular health, excessive or uncontrolled autophagy can initiate a type of cell death known as “autosis” (  While autophagy is believed to be helpful, cell death through autosis is inarguably harmful and to be avoided.  The clinical significance of autosis in the context of therapeutic autophagy is not clear.  I reached out to Beth Levine, the woman who did much of the original research on autosis but her email bounced back and it turns out she has tragically died of breast cancer.  In any case, in this setting of incomplete understanding, the best course is one of prudence and moderation.  In the words of Hippocrates, first do no harm.  Practically speaking, my policy is to cycle agents, to avoid excessive dosing and to carefully follow serologic markers during treatment.

Relationship between autophagy and weight?
There is an emerging understanding of a relationship between autophagy in the medial-basal-hypothalamus (MBH) and obesity mediated by the FK506 (tacrolimus) binding protein FKBP51, which appears to be a major upstream regulator of autophagy.  Could autophagy and more broadly, the intracellular environment in hypothalamic neurons be the functional memory that induces phenomena such as the physiologic set point?  Could agents that augment autophagy such as rapamycin reset the physiologic set point in humans, as it has in mouse models?  These are intriguing questions.

Epigenetic clocks and Reprogramming  

How do we measure aging?  Certainly, there is the passage of time and how many birthdays someone has had, but we know that aging proceeds more quickly for some people than it does for others.  The speed of aging depends on your genes but also various factors such as, for instance, whether you smoke, your diet and level of exercise, your stress levels, exposure to ionizing radiation, and many other variables.  A 50 year old executive will likely be biologically younger than a 50 year old homeless alcoholic who has had a hard life, to take an example.  Is there some other marker that indicates a biological age that is distinct from chronological age?  This question is critically important if we are to assess the effectiveness of anti-aging interventions and has been a matter of much debate.  Recently as our understanding of aging has evolved, scientists have focused on methylation patterns in the DNA as a likely proxy for aging.

These so-called methylation clocks presuppose a process of epigenetic aging in which the organism’s DNA is gradually ornamented with methyl groups and that this methylation pattern is a good indicator of how old someone really is.  These methylations happen at points in the genome where cytosine is bound to guanine through a phosphate group (CpG).  When these CpG sites are methylated, it changes the three-dimensional structure of the DNA and changes the way that DNA can be expressed. Many of these sites are located at gene promoters or regulatory sequences including enhancers. The theory holds that the loss of function with aging happens because the heavily methylated DNA is shaped differently and cannot be read (or transcribed) as easily as it once was.

These methylation patterns were discovered using deep computer learning analysis of tissue samples, by Steve Horvath and others made possible by affordable array-based technologies and massive amounts of publicly available human methylation data.  Horvath was able to identify 353 CpG methylation sites that predicted the progression of aging from embryonic stem cell to old age.   Of note, this is a very small fraction of the total number of CpG sites, which is itself a small fraction of the other methylations, histone acetylations and other changes that happen with age, so keep in mind that this is a highly focused view on a process that is immensely more widespread.  Since his pioneering work, many different age clocks have been described which employ a variety of clusters of CpG methylation sites.  It’s not clear if these clocks are measuring the same thing[i]

Are these methylation patterns a marker for aging or are they actually driving aging?  This is not a settled question. Many of these sites are located at gene promoters or regulatory sequences including enhancers, and so one theory is that methylation changes the .We do know that increased epigenetic age relative to chronological age has been linked to disease and mortality risk.  And removing methylations in mouse retinal cells (using Yamanake factors, see below) and reintroducing these cells into the retina using a viral vector reversed vision loss and cured blindness in mice.[ii]   The researchers believed that they were reprogramming the cells to an earlier (younger) state of their progression on the program of methylation.   Reprogramming means taking aged or otherwise faulty cells and tissue that have already grown and “differentiated,” and resetting them closer to an original state of induced pluripotency.  This is done using a group of transcription factors called Yamanake factors, discovered by the Nobel laureate Shinya Yamanaka.  These Yamanake transcription factors regulate the expression of DNA in such a way that the cell returns to a state where it can become any kind of tissue.  They are:

Oct-3/4:  transcription factors critical for maintaining pluripotency and preventing cell differentiation
Sox family:  important for sex determination but also maintenance of pluripotency
KLF4: regulation of apopotosis, tumor suppression
c-Myc: proto-oncogene

One of the actions of the Yamanke factors is to make possible the de-methylation of the same CpG groups that are theorized to drive differentiation and aging.  Myc is sometimes left out because it is associated with uncontrolled growth, leaving Oct/Sox/KLF aka “OSK” treatment.  OSK treatment in mice has been shown to reduce levels of methylation and also to restore function.  If reprogramming with OSK is pursued to its natural conclusion, cells return to pluripotency and can become cancerous.  However, if the reprogramming is done incompletely, then the cell simply becomes younger.  This is the aim of cellular reprogramming in a nutshell, and an enormous amount of money is being invested in this field ($3 billion for Bezos funded Altos labs last week) to capitalize on the technology.  The goal will be to safely do partial reprogramming, to avoid pluripotency and the risk for malignancy.  In a recent paper, they appear to have done just that with mice.

What is driving the methylations?  Is it just DNA damage, a stochastic and random process, as David Sinclair has proposed?  Or is it some programmatic process from birth to death?     The connection between metabolic activity and modification of histones is well documented, could metabolic activity be a driver of methylation?   We don’t know yet.  One observation supporting randomness is that methylation changes are more likely to occur at early and late replication sites in DNA.  In other words, when a cell divides and DNA is copied to create two daughter cells, the epigenetic changes are also copied.  The observation that the epigenetic changes are concentrated at specific sites relevant to the replication process suggests some mechanical issue rather than an external program driving methylation from embryogenesis to old age.  On the other hand, how would it be that all cells in the body (or at least a statistical majority) should change their epigenomic methylation patterns simultaneously?  The improbability of this occurrence argues for a central regulator.  Thus far, empiric evidence either way is scant.

Another question is whether reprogramming creates a truly rejuvenated cell in every sense of the word.  We know treatment with Yamanake does not remove all methylations, but only important subsets of them. Is there longevity for these cells?  How do they behave over time.  Do they still have all the replicative DNA damage accrued over time?  How are their mitochondria? These are questions that remain to be resolved.  Hopefully in the next decade we will see clinical applications that derive from this technology in human beings.  It will be expensive at first, but I’m sure at least one child with a previously incurable disease will be granted a normal life thanks to reprogramming.  And he or she will be free to age normally and watch the world gradually warm to uninhabitable levels and descend into geopolitical chaos, but that is a matter for another blog post.