15 replies to this topic

[jaydfox]

One of the big points of contention with SENS has been that of nuclear DNA repair. Aubrey de Grey has proposed WILT for dealing with the problem of nuclear DNA mutations. Simply put, Aubrey's position is that cancer is such an overwhelming problem with regards to nDNA mutations that the rest of the DNA essentially gets a free ride. Appealing to mathematics, the level of DNA repair necessary to maintain genetic fidelity for 70 years on average before even one cell out of trillions becomes cancerous, is a level that seems stupdenously efficient.

As animals have grown larger, lifespans have increased. DNA repair has presumably been part of this increase: a human couldn't live 85 years with DNA repair rates only sufficient to stave off cancer for a few months in flies or years in mice. In fact, as animals have increased in size by several orders of magnitude, one would expect the DNA repair rates to be nearly flawless.

As I was writing a reply to noam in another thread regarding the following quote, I began to wonder about whether this is as true as it seems on the surface.

link to entire quote

if we: 1. Remove extrinsic death factors for a vertebrate, and 2. Remove its ability to die of cancer, then the amount of time (after its old extrisic life expectancy period) that will take him to lose that certain amount of cells which will make it dysfunctional, will be relatively very long.

The gist of it is, an organism can lose a large fraction, say 1%, of its cells, and still survive. In a human body of 100 trillion cells, this could mean a trillion defective cells. Yet if the DNA repair rates are good enough to prevent even one metastatic cancer cell for 70 years, then certainly it must be good enough to prevent a trillion defective cells for several lifetimes' worth.

I started approaching this idea from the top side: what's good enough for 70 years isn't adequate for 90 years, let alone 140. It's a basic tenet of aging in higher organisms, that for the most part, age-related diseases (and by association, the underlying damage) increase near-exponentially. Reliability theory adds a subtle kink: the damage might accrue linearly (or faster, or perhaps even slower), but the overall effect is a roughly polynomial increase in the incidence of disease, essentially exponential over the period of interest.

So, while non-cancer-related genes might be in very good shape for 70 years, we might still expect a pathological level of mutations in less than 140 years.

But, as I was approaching this from the top, I started wondering about the approach from the bottom. Depending on the cancer, about 4-8 genes need to mutate for cancer to metastatize. I'm fuzzy on the number, but that's what I seem to recall.

Well, for simplicity, let's just call it 6. Six genes need to mutate in a single cell and a few years later (a small fraction of its lifespan, and hence a selective pressure), the organism is dead.

Well, we need a further qualification. It appears that not all cells are equally liable to develop into cancers. Stem cells and other mitotic cells seem to be the main culprits, which makes sense. They may already have telomerase active, which reduces by one the number of mutations needed. Additionally, they replicate their DNA, which exposes their DNA to more potential mutations.

So let's say that in some small fraction of the body's cells, five mutations are needed. I'll just pick a number, say a billion, or 10**9. Well, as it turns out, the mutation rate (of unrepaired, undetected mutations) that can be tolerated is the fifth root of a billion, since five genes need to mutate. Why? Because you multiply that mutation rate by itself for each gene that must mutate within a single cell.

And 1/10**1.8 turns out to be about 1.6%. That's right, for stem and other mitotic cells, it's a 1.6% mutation rate for each cancer-related gene, in a (roughly) 70-year period. Using 10 billion or 100 billion cells as our starting point hardly changes things, reducing this rate to 1% or 0.6%, respectively. Using the full 100 trillion cells and six genes (e.g. adding telomerase back in), we get (10**12)**(1/6), or 1%, so it's not much better for the rest of the body.

Assuming that all genes get this same "free ride", then we should expect that in a 70-year-old, nearly 1% of all genes in all cells should be mutated in some way.

This is staggering. Note that I didn't say 1% of all cells would have mutations (which in and of itself would be worrying). No, 1% of each of the 25,000 genes, or about 250 genes per cell, with only a small standard deviation. The overwhelming majority of cells in the body would have at least several dozen mutated genes. This certainly doesn't seem like the "free ride" we were promised. What happened? Using this math, the vast majority of cells in a 25-year-old's body would already contain a dozen or more mutations, and a significant percentage would already contain scores.

Certainly these numbers seem too high, and I'd say this is because DNA repair is better than it needs to be to fight off cancer. Cancer's a walk in the park compared to having 250 mutated genes in almost every cell in the body.

But the principle can't be ignored: mathematically speaking, cancer doesn't buy our genes a free ride. The rate of undetected, unrepaired mutations that can be tolerated before cancer becomes pathological is on the order of 1% of every gene in each and every cell. That's a horrible repair rate. Cancer might seem like it's done us a favor, but I don't buy it. Evolution needed better ammunition to fight cancer than just DNA repair, and it found it in the elaborate system of checks and balances, tumor suppressors, etc., that require a single cell to develop multiple specific mutations, and probably in a fairly strict if not absolute order.

On top of this, evolutionary theory would seem to dictate that pathological systems are only made at best slightly less pathological than other systems. Because of the (near) exponential rise in pathology, a disease with a 1% incidence at one age might have a 90% incidence just a fraction of a lifespan later. Non-cancer pathology arising from nuclear DNA damage is probably at best a couple decades behind cancer, and we should not assume that fixing cancer will end all our worries about DNA damage.

With that said, I think that Aubrey has still taken a prudent course of action, because, thanks to the principle of escape velocity, WILT doesn't need to buy us more than probably two or three decades. In fact, WILT, if ever used in human treatments, will probably only be a stopgap measure for those who already are at very high risk of imminent cancer or already have cancer. For those with low risk (less than 5% or 10%), I seriously doubt it would be worth the risks. But it's a nice idea that, if nothing else, is good for putting to rest the arguments that cancer can't be stopped.

Edit: Updated "I seriously it would be worth the risks", adding "doubt"

[ag24]

Vijg gets numbers around 10 mutations per cell from experimental observation, so you're only an order of magnitude out. But in determining how bad that is, you haven't taken into account that we have two copies of most genes and missing one copy of most genes is harmless.

[jaydfox]

Heh, I left work right after writing this, and as I was thinking about it, I realized I did my math wrong. 100 trillion is 10**14, not 10**12 (I was thinking 10 plus 2 for the hundred, I guess...).

So for 100 trillion cells and six mutations, we have (10**14)**(-1/6), which is a little under 0.5%, not 1%. But as you can see, with so many mutations, the number of cells gets diluted in the end anyway...

Using an alternate, more pessimistic set of numbers, let's go with the lower figure of four mutations. For 100 trillion cells, that gives (10**14)**(-1/4), or 0.03%. That doesn't sound like much, but multiplied by 25,000, it would still mean an average of 8 mutated genes in each an every cell by age 70.

Going with 100 billion stem/mitotic cells and three mutations gives (10**11)**(-1/3), or 0.2%, in the same ballpark. Any way you slice it, cancer hasn't given us much of a free ride. If cancer affected humans as it does mice, requiring fewer mutations, then yes, cancer would have given the rest of our genes a free ride. But evolution took another, more efficient path, and nDNA still has to fend for itself. Cancer may be the predominate form of DNA damage that is exerting a selection pressure, but other non-cancer effects aren't guaranteed to be far behind.

[jaydfox]

Vijg gets numbers around 10 mutations per cell from experimental observation, so you're only an order of magnitude out. But in determining how bad that is, you haven't taken into account that we have two copies of most genes and missing one copy of most genes is harmless.

Heh, that's embarrassing.

[jaydfox]

Hmm, factoring in the redundant copies of genes, we effectively have a reliability system. Let's say a subset of 1,000 genes are crucial to any given cell, and let's go with the lowest rate derived above, 0.02%. Each of those 1,000 genes represents a redundant system with two copies, so we get 0.0002**2 as the probability of a given gene going kaput in a given cell. Multiple that probability by 1,000, and we get a probability of 0.004% that any given cell will critically lose both genes. That's not so bad. Using the other low figure of 0.03% (which roughly matches Vijg's numbers, i.e. his 10 and my 8) gives an overall probability of 0.01% that any one cell would go. Again, not so bad. However, note that that figure will grow quadratically, so lifespan will begin to be affected not too long after the cancer selection pressure would have kicked in but admittedly probably at least another lifespan later. Hmm, interesting...

Using the more pessimistic rate of 0.5% (i.e. six mutations) gives 2.5%, which is a reasonable threshold for non-cancer mutations to exert their selective pressure on par with cancer, if not earlier.

[jaydfox]

an overall probability of 0.01% that any one cell would go. Again, not so bad. However, note that that figure will grow quadratically, so lifespan will begin to be affected not too long after the cancer selection pressure would have kicked in but admittedly probably at least another lifespan later.

I keep messing up orders of magnitude. I saw 0.01%, but somehow in my mind I was using a figure of 0.1% when I said "at least another lifespan later". Another lifespan would take 0.1% up to 0.4%, or basically half a percent, approaching a threshold of pathological levels. But with 0.01%, even at three lifespans of age (say 210 in humans), the levels would only reach about a tenth of a percent, 0.1%, still probably below a critical threshold. So if these numbers hold up, then perhaps I have finally convinced myself that non-cancer DNA mutations won't matter on the timescales Aubrey quotes. Since these numbers also seem to line up with Vijg's, that's at least one source of experimental validation for the numbers/formulas I used, with 4 DNA mutations in 10 to 100 trillion cells or 3 mutations in 1 to 100 billion cells, and 1,000 critical genes.

[noam]

The gist of it is, an organism can lose a large fraction, say 1%, of its cells, and still survive. In a human body of 100 trillion cells, this could mean a trillion defective cells. Yet if the DNA repair rates are good enough to prevent even one metastatic cancer cell for 70 years, then certainly it must be good enough to prevent a trillion defective cells for several lifetimes' worth.

Actually, the DNA repair/replication mechanisms are suppose to keep the genome (more or less) intact during the extrinsic-governed life span of the organism.

The extrinsic-governed life span for humans today is about 1000 years (based upon the chances of 20 y/o who lives in the western world to die at this age), but it only happened recently. Before the modern era, the extrinsic-governed life span gave humans ~30 years of lifespan.

So, if you WILT a person at age 30, I think there is no argue that if you leave other types of damages (apart from nDNA mutations) aside, that person will be able to live ~ 3-4X its old extrinsic-governed life span (i.e 90-120 years).

We are not built to last 70 years, we are built to last 30 years. At age 75, our genome might already be damaged to a high degree, maybe not life threatning, but still to a high degree.

Furthermore, as jaydfox said, we evolved mechanisms to fight cancer cells. Apart from intracellular mechanisms (ala p53/63/73 etc.), we also have the immune system doing the job of getting rid of cancerous cells. My professor is saying each day every one of us is developing many potential cancerous cells, which are efficiently killed by the immune system.

I'm not sure how efficient our DNA repair/replication system really has to be, if you factor out all these cancer fighting mechanisms. It is possible our body "decided" that instead of improving DNA repair to deal with cancer prevention (above the level needed to defend the rest of the genes) , it is better to simply evolve special mechanisms to deal with these cancerous cells, that's because our body is built from trillions of cells, and losing ~1% won't do much of a difference, so it's better to just deal with cancer cells separately.

[jaydfox]

I'm not sure how efficient our DNA repair/replication system really has to be, if you factor out all these cancer fighting mechanisms. It is possible our body "decided" that instead of improving DNA repair to deal with cancer prevention (above the level needed to defend the rest of the genes) , it is better to simply evolve special mechanisms to deal with these cancerous cells, that's because our body is built from trillions of cells, and losing ~1% won't do much of a difference, so it's better to just deal with cancer cells separately.

Yes, I agree that DNA repair was not pushed exclusively by cancer. DNA repair has to be at least good enough to keep functioning genomes in ~99% of cells during the typical lifespan. It also has to be at least good enough to prevent death by cancer. But with a 10,000-fold increase in cell count between mice and men, as well as a 10-fold increase in lifespan (staying to powers of 10), evolution was in no position increase DNA repair by even a decent fraction of what was required.

Evolution took the path of least resistance, which in this case was to fight cancer by means other than DNA repair. But DNA repair still needs to be at least good enough to prevent death from cell loss/dysfunction, which puts a selection pressure against getting even fancier at fighting cancer than our bodies already are. Were our bodies to contain 10,000 times as many cells as they presently do, with another factor of 10 in lifespan, evolution would once again seek additional checks and balances in dealing with cancer, and would not merely resort to better DNA repair. But those checks and balances wouldn't help us in our present form, because DNA repair would still need to be at least good enough to prevent death from cell loss/dysfunction. If we had the cancer fighting system of such a large and long-lived creature, we'd still die at a few hundred years of age at most with our current DNA repair systems (even assuming we address the other strands of SENS), and probably much sooner. Or more practically, a mouse with the cancer fighting systems of a human (i.e. a mouse that essentially would never get cancer, not even if it lived 120 years) would still die only a few months later than his normal cousins, or at best a few years to a decade later with all strands of SENS but no change in DNA repair. DNA repair rates in mice are sufficient for mice lifespans.

We probably won't know for sure for a couple more decades, but it still seems to me now that cancer hasn't given nDNA a free ride. But we're talking about orders of magnitude: a free ride would be if we could live in excess of 700 years on average if we could eliminate cancer but otherwise do nothing for DNA repair. 140 to 210 years would not be a free ride, but it would be way more than sufficient for WILT or some other effective cancer strategy to help effect escape velocity.

But if it's only 90-100 years (versus 70 on average), then DNA repair will remain an important topic to address in addition to a cure for cancer. So in this sense, the practical sense, the numbers matter, and for that we need not arguments (which can accurate to within an order of magnitude or a little better), but hard data.

Of course, the big point I haven't addressed is that if curing cancer would still lead to death from DNA damage only a couple of decades later, would this still be true if we also cleaned up the other 6 points of SENS, especially mtDNA damage and intracellular junk. Presumably these other items might add to nDNA damage rates in some way, perhaps through toxicity or oxidants or stress. So perhaps fixing the other six SENS targets would slow the DNA damage rate, buying our genomic integrity-limited lifespan an extra decade or two, which is already, say two decades longer than our cancer-limited lifespan. That could be 30-40 years extra, enough to effect escape velocity.

[jaydfox]

So, if you WILT a person at age 30, I think there is no argue that if you leave other types of damages (apart from nDNA mutations) aside, that person will be able to live ~ 3-4X its old extrinsic-governed life span (i.e 90-120 years).

By the way, I didn't want you to think I just ignored your comment. It was quite insightful. I usually just think of lifespans in terms of intrinsic mortality (~85 in humans), but it's interesting to think about the extrinsics factors and how the affected evolutionary pressures.

But I think 30 is an underestimate, because our species has seen a near doubling of intrinsic life expectancy in the past few million years (rapid evolution, too rapid for fancy new systems, so it must be the basics like DNA repair, anti-oxidants, mitochondrial electron transport leakage, that sort of thing), based on comparisons with our primate cousins (chimps, apes, etc.). So this would equate to an extrinsic-governed lifespan of 15 years for our ancestors, which I think is low considering when puberty is and that the children would need their parents alive to help them reach puberty or at least become minimally self-sufficient (~8 years old?). Hence our ancestors probably had lifespans closer to 20 years, I would guess. I've heard a figure of 18 somewhere, so this seems more reasonable. Taking into account the doubling (or more) of our lifespans, I think at least 35-40 would be a more realistic figure for extrinsic-governed lifespan.

But with 40 as an extrinsicly governed lifespan, that would mean that we survive cancer-free for about 1.5 lifespans, and in the absense of extrinsic factors, we can survive about 2.0 lifespans. Where WILT and SENS can take us (without DNA repair), we'll hopefully find out soon.

[olaf.larsson]

Vijg gets numbers around 10 mutations per cell from experimental observation, so you're only an order of magnitude out. But in determining how bad that is, you haven't taken into account that we have two copies of most genes and missing one copy of most genes is harmless.

Please, Does anyone of you have a link to this paper? Would you like to post references to some other good papers about mutation estimations in nucleus as well as in mitos.

Thank you

[ag24]

Vijg has been publishing on this for a decade, so there are loads of refs. This one has full text free online:

http://www.ncbi.nlm....l=pubmed_DocSum

[noam]

Regarding the quality of DNA repair in Yeast, this is an interesting article: An age-induced switch to a hyper-recombinational state

They show that the rate of DNA-damage-accumulation in the Yeast strain that they checked, is drastically higher with age. (They don't say if this actually is the major cause of death for the strain, but it's a possible conclusion, especially because they also checked for a fob1 mutant strain, and they obsereved the same results as with the w.t, so the effect has nothing to do with ERCs).

They explain it using the following interesting postulation:

"Aging mother cells accumulate damage proteins over time (25), which effectively eliminates the normal function of a gene product required for genome integrity. This defect appears to thwart normal DNA damage detection, because, unlike young cells repairing an induced double strand brake (26), mother cell divisions producing a daughter with LOH (Loss Of Heterozygosity), lacked noticeable cell cycle delays or arrests.".

So, if they are correct, what category will you include this phenomena ?, nDNA mutations ?, intracellular junk ?.

To me it looks like it belongs to the intracellular junk category, and this raises the interesting (and somewhat disturbing) possibility, that intracellular junk does not have to be highly abundant in the cell, in order to create a catastrophic consequence. It can be some small amount of junk protein, with high affinity to some important enzyme (related to DNA repair detection, in this case), and so it will not be easy to detect and isolate this junk, in order to find the right enzyme to brake it.

The only reason we were able to detect ERCs as a possible cause for yeast aging, is because they are extremely abundant in old yeast cells, and have a very dominant appearance in a 2 dimentional electrophoresis analysis of the yeast DNA.

Another consequence, is that the Yeast DNA repair efficiency, might allow a replicative life span never seen before, just by SENSing a few problems... (but it might not be that easy for the reason mentioned above).

Edited by noam, 05 January 2006 - 07:22 PM.

[ag24]

This looks much more like a mutational event to me -- "Once an LOH event occurs in a pedigree, additional LOH is observed at a higher frequency for the duration of the motherÕs lifespan. This suggests that as mother cells age, there is a switch from a state with a low spontaneous rate of LOH to a state of increased genomic instability." Junk, abundant or not, would be expected to cause a gradual increase in rate of LOH.

[apocalypse]

How does this all fit with non-cancer related DNA damage? There are substantial changes in gene regulation as we get older. It is as if the developmental program initiated at conception does not stop at puberty but continues until death. It would be naive to consider that such exquisitely developed mechanisms of differentiation just stop functioning at this stage. Particularly when we observe a gentle, gradual but highly systematic process.

A famous cell biologist (Hayflick) once described this process as stochastic and compared it to the gradual breakdown of a car. But in extending his metaphor it does not account for when a door handle decides to become a shiny, brand new carburetor which continues to grow at an alarming rate until it overwhelms the function of the car (cancer). On the contrary, such an observation implies tremendously powerful control systems designed to constrain the incredible vitality inherent in all cells.

Given that control systems such as apotosis (cell suicide) and senescence exist it does not require a stretch of the imagination to consider that phenoptosis (organism suicide) may also be an evolved biological strategy. We know that the programmed absence of the telomerase repair mechanism facillitates a cell division clock in somatic cells that has been selected for by evolution. Can we extend this reasoning further and consider that other mechanisms of repair have been similarly constrained so as to limit lifespan? If we do, then genomic DNA repair is a logical candidate and it would imply that cancer is but a side effect of a deliberate underlying process of genomic disintegration.

Indeed, the moment I saw the issue of limited resources this conclusion became highly plausible to me(the data indicating programmed aging in lower species such as C.elegans, further cemented this possibility in my mind). The reasoning goes: A population has in essence to divide resources amongst the various organisms that compose it. If within a certain region a particular species began to slowly develop greater lifespans beyond what's required for its particular niche, it would in essence be taking away resources that could be used to build more individuals of the next generation, in addition fully mature/grown organisms that did not manage to reproduce or had done so prior would be competing with the immature organisms of the same species for the limited resources available, and should in essence compromise the survival of these despite whatever genetic advantages these may've that would've allowed the species to prosper further. In essence the species' ability to adapt by replacing the old generations with new generations with fresh variation would be compromised and it would be less able to adapt to changes then other competing ones. This effect should increase the further off the avg organisms within a particular species got from the lifespan that would be required for its particular niche, selective pressure in opposition growing stronger.

Yes, I agree that DNA repair was not pushed exclusively by cancer. DNA repair has to be at least good enough to keep functioning genomes in ~99% of cells during the typical lifespan. It also has to be at least good enough to prevent death by cancer. But with a 10,000-fold increase in cell count between mice and men, as well as a 10-fold increase in lifespan (staying to powers of 10), evolution was in no position increase DNA repair by even a decent fraction of what was required.

Evolution took the path of least resistance, which in this case was to fight cancer by means other than DNA repair. But DNA repair still needs to be at least good enough to prevent death from cell loss/dysfunction, which puts a selection pressure against getting even fancier at fighting cancer than our bodies already are. Were our bodies to contain 10,000 times as many cells as they presently do, with another factor of 10 in lifespan, evolution would once again seek additional checks and balances in dealing with cancer, and would not merely resort to better DNA repair. But those checks and balances wouldn't help us in our present form, because DNA repair would still need to be at least good enough to prevent death from cell loss/dysfunction. If we had the cancer fighting system of such a large and long-lived creature, we'd still die at a few hundred years of age at most with our current DNA repair systems (even assuming we address the other strands of SENS), and probably much sooner. Or more practically, a mouse with the cancer fighting systems of a human (i.e. a mouse that essentially would never get cancer, not even if it lived 120 years) would still die only a few months later than his normal cousins, or at best a few years to a decade later with all strands of SENS but no change in DNA repair. DNA repair rates in mice are sufficient for mice lifespans.

We probably won't know for sure for a couple more decades, but it still seems to me now that cancer hasn't given nDNA a free ride. But we're talking about orders of magnitude: a free ride would be if we could live in excess of 700 years on average if we could eliminate cancer but otherwise do nothing for DNA repair. 140 to 210 years would not be a free ride, but it would be way more than sufficient for WILT or some other effective cancer strategy to help effect escape velocity.

But if it's only 90-100 years (versus 70 on average), then DNA repair will remain an important topic to address in addition to a cure for cancer. So in this sense, the practical sense, the numbers matter, and for that we need not arguments (which can accurate to within an order of magnitude or a little better), but hard data.

Of course, the big point I haven't addressed is that if curing cancer would still lead to death from DNA damage only a couple of decades later, would this still be true if we also cleaned up the other 6 points of SENS, especially mtDNA damage and intracellular junk. Presumably these other items might add to nDNA damage rates in some way, perhaps through toxicity or oxidants or stress. So perhaps fixing the other six SENS targets would slow the DNA damage rate, buying our genomic integrity-limited lifespan an extra decade or two, which is already, say two decades longer than our cancer-limited lifespan. That could be 30-40 years extra, enough to effect escape velocity.

Don't forget that unlike mice and their mad breeding large population cycle, humans have been living in not so large pockets for the majority of history(and even more-so as you go into prehistory). Along with larger size and longer lifespans the number of offsprings decreases, iirc, and so too does population size in this comparison. If ndna mutations accumulated too much the avg viable progeny of the species(as nonviable ones would be culled prior to conception, soon after or by selection thereafter) would slowly decrease in fitness as the generations went by in small tribes/pockets/populations. Natural Selection would choose the fittest amongst these but it can only tolerate a certain rate of dmg in such small pop.s before it's simply choosing amongst those that were the least worse off rather than those that were even fitter then prior generations(organs with countless genes and regulatory regions demanding even higher fidelity due to higher lvls of precise organization and extreme complexity would also put pressure on it, such as a large and complex brain.). We can see this today even with the present lvl of selection, and with post-civilization pop.growth, lots of detrimental and especially slightly detrimental mutations are still present at large. Even a small group of long living individuals with few offsprings in virtually isolated locales can last for thousands of years(if not indefinitely), impressive when you consider the countless high fidelity demanding regulatory elements required for the development of such complex organisms, along with the protein coding genes.