Please give me some feedback on these ideas. There are probably many
errors of logic and my knowledge of biology is still scant, but maybe
there is something useful herein. Also please feel free to add
suggestions for experiments if any of these ideas seem promising.
I Continuous vs binary expression of proteins:
In the following I will assume that differential expression of a
particular protein between young and old cells occurs as a result of a
mutation to some gene (not necessarily the one coding for that protein).
This assumption may be false in several ways. One is that several
concurrent but not necessarily simultaneous mutations have to occur for
the protein of interest to have
differential expression. The other is that the cell's environment of
nutrients and hormones is changed, perhaps due to other cells (which
produce hormones or metabolize nutrients that go to this cell and the
extracellular matrix) having mutations or stoping replication due to
e.g. tellomere shortening and
lack of tellomerase. I will think of some experiments to distinguish
these possibilities later.
The experiments that have been done so far on differential expression
between old and young tissues (or senescent and young cells for in vitro
systems) mostly look at relative amount of mRNA from a large population
of
cells of the same type (One purported exception that I need to double
check is albumin expression in certain liver cells which supposedly
shows continuous changes in expression in each cell). This does not
allow to distinguish between
mutations that result in a binary (high/low) expression of proteins, vs
mutations that result in a continuous expression of proteins. Such a
distinction is important because if it is determined experimentally that
the continuous class correlates with ageing (in a particular cell type,
or even better, in all cell types) then many mechanisms are eliminated
from candidacy for root causes of ageing. Actually I think there are
very few non lethal mutations that can cause a continuous change in
protein expression per cell. These include (correct me if I'm wrong)
mutations to
1. chromatin coding genes(coding for histones/scaffold
proteins,methylases and acetylases). Chromatin structure can affect mRNA
synthesis
through both throughput (ease of transcription initiation by RNA
polymerase and transcription factors) and possibly rate (I think RNA
polymerase might interact with histones, not just DNA, as it moves
along the DNA).
2. enhancer or promoter proximal DNA (TATA, CG and others). Effects on
ease of transcription initiation.
3. hnRNP coding genes, which may be able to function at lower efficiency
in transport out of nucleus and preventing secondary structures from
hapenning if only a few are missing.
4. rRNA coding DNA, since there are many copies of this gene, not just
two per diploid cell. Also, a ribosome may be defective and translation
can still proceed, albeit at a lower rate (?)
5. mtDNA, since there are many? (how many) mitochondria per cell.
6. nDNA coding for mtDNA repair and/or replication enzymes (same reason
as 5).
Of course if the result of the experiment is that differential protein
expression is binary, one learns less since most of the above mutations
can also cause binary expression (e.g. the histone may not be produced
at all). It would be good to induce each of these separately and
intentionally and see the effect on expression of the protein being
studied, to verify that indeed they produce continuous changes. Actually
I don't think (if aging occurs because of DNA damage) that protein
expression can vary discontinuously in each cell because, as I show in
the next section, unrepairable damage in non proliferating cells can
only occur by knocking out not only a certain gene coding for a certain
protein, but all repair pathways for that type of damage, and then other
genes that are essential for cell viability will become unrepairable.
But this would mean that unless one of the six cases above occurs, lots
of cells would start dying with age, not just stopping expression of a
few proteins, which is not observed.
Now the question is how does one look at protein expression from 1
cell at a time, and one answer is PCR, but for a protein (whose gene has
already been sequenced so that primers can be obtained) known to be
differentially expressed between young and old cells, such as a collagen
proteins,
or interferon gamma. This can be done for several different tissue
types. It would be better to do this for hair producing cells which have
(I think) an easily identifiable (single cell specific) phenotype which
differs between young and old ( i.e. a bald spot, or a white hair). One
would look primarily for melanins in the case or greying of hair(?).
FACS can also be used, but I think this is limited to proteins which
bind to the cell membrane. When it was done for albumin, the result was
continuous change in expression, supporting the above arguments (but I
thinkk it was done at one time, not as a function of time, looking at
differences between stem and more differentiated cells).
If aging occurs as a result of changes in nutrients or hormones then it
can also be either abrupt or continuous. Abrupt change might be ruled
out for the arguments above. Continuous change in hormone supply will
cause abrupt change in a cell's protein expression unless it affects one
of the six gene classes above (a hormone may come into the cell at
various time intervals and in various doses but at any one time when the
cell is measured, it would either signal transduce and turn on
transcription of genes or not). If we look at the CAUSE (as opposed to
the effect) of continuous change in hormone production then we don't
learn much because it can be due to abrupt changes in protein production
at the cell level which are continuous on the population level (e.g.
some cells are senescent/non proliferating and some not), not just
continuous changes in protein production .
II DNA repair in replicating vs non replicating cells
Although DNA repair is less efficient in non replicating
cells, there is more time to do it before something is no longer
recognizable as damage (as Syndey Shall pointed out), which is what
happens in replicating cells that
actually fixate a mutation. In a non replicating cell, the only way to
knock out a gene with random damage (due to toxic environmental or
metabolic oxidant species) is if there is only 1 (or no) repair pathway
for that
type of damage (or one enzyme shared between two or more pathways), and
if a gene coding for an enzyme in that pathway (or the shared
enzyme)also gets the
same type of damage. Redundancy makes it almost impossible to knock out
a gene. First consider the simple situation in which different repair
pathways do not share enzymes. Consider a simple example with genes A, B
and C which code for
repair enzymes which can repair damage of type 1,2 and 2 respectively
(with the help of other proteins).
If gene D (which produces some protein necessary for cell function) gets
type 2 damage, (we'll then call it D2), then it can be
repaired by either E2B or E2C (the enzyme from gene B or the enzyme from
gene C). If we have either A2,B2,C2 or D2, then still no problem since
there is still E2 available (either E2B or E2C) to repair the damage. It
is extremely unlikely that we have simultaneous B2 and C2 (simulataneous
meaning B2 happens before E2C has a chance to repair it).
Also no problem if D1 happens, since it can be repaired by E1A. However,
if A1 happens, then it can't be repaired, and eventually D1 will happen
and not be repaired. Unrepairable damage (the no E case above) is also
possible, but I think it is unlikely to be related to ageing, because if
unrepairable damage happens to a gene that produces a protein needed for
cell viabilty that cell will die, and the random death of a population
of cells does not have much to do with ageing.
Also in the case where an enzyme is essential for two repair pathways,
and if those two pathways repair the same type of damage and no other
repair pathway exists for that type of damage, it is possible to knock
both of these out with a single mutation (of the same type as the repair
pathways) to the gene coding for that enzyme, and in that case
redundancy doesn't help. So we should look for enzymes that are shared
between repair pathways, as these are candidates for aging-relevant
mutations.
Now the clincher: for diploid organisms, even if there is only 1 repair
pathway (or 1 shared enzyme) for a specific damage type, there are 2
genes coding for it, one
on each chromosome, with the exception of the X for males and females,
and the Y for males. My guess is that there is no repair enzyme coding
genes on Y chromosomes (females would
live shorter lives than males). Indeed, a repair exonuclease has been
found on the X chromosome (of humans, I haven't checked other animals
yet). I don't think it is known if this exonuclease is active in
mitochondria (this should be able to be determined experimentally).
Maybe aging starts with mutation to this
repair enzyme and everything follows from that. I suppose it is
possible that simultaneous mutations though rare, do happen, and this
kind of system could be simulated (by Monte Carlo) to see if it exhibits
the same linear differential expression observed as a function of time
(although it is not known, as I said in the first section above, whether
there are just more cells being knocked out, or each cell continuously
expresses more or less protein. Experimental resolution of this would
give the simulation something to compare against)
The situation in replicating cells is totally different. There the trick
is to either repair the DNA before a mutation is fixed, or to have a
good apoptosis/selection mechanism and replace the cell (and in the rare
case of a better cell, let evolution select for that cell) . Redundancy
happens at
cellular level, instead of enzyme level (as it might in non-replicating
cells).
III. Selection in sperm and egg in order to keep the germline immortal
Presumably the embryo is young because some selection occured in sperm
and/or egg for a cell without (or with less of) the damage that causes
aging (I won't
consider now the other possibility, that the hormones/nutrients which
may be different in an aged individual compared to a young individual,
and cause differential protein expression in most cells, are not
necessay for sperm or egg function).
Selection for good mtDNA may occur efficiently for eggs since a few
mitochondria are
amplified to many in oogenesis (thus widening, or even bimodaling the
fitness distribution) and this has been proposed a while back and is
known as
the mtDNA bottleneck theory. The bottleneck theory does not explain how
or when selection for eggs low in damaged/mutated mitos occurs. An
obvious example is after fertilization, the high damage fetus
spontaneoulsy aborts. Complete speculation: a selection for good
mitochondria may also occur in either meiosis 1 or 2. TEM observations
suggest two mitochondrial populations prior to meiosis 1 (R Jansen and K
de Boer Mol Cell Endoc 145 (1998) 81-88) . Possibly these segregate into
polar body and oocyte ?) and I'm wondering if these are just selected at
random or not. One population appears to be replicating DNA while the
other is not.
By the argument in section
II, it is not sufficient (for aging to occur in non proliferating cells)
to have damaged DNA (mt or nuc), but a unique repair pathway (or unique
shared enzyme in multiple repair pathways) has to be damaged as well. If
this is the case, then selection on good repair genes in eggs (which
will be passed to non proliferating cells) should happen. Also
there is a role for selection on sperm as well, and indeed selection
does happen on things like motility and ability to penetrate the egg,
which are energy dependent (and
hence ATP and good mtDNA dependent). What is selected on sperm is not
mtDNA (which is not passed to the zygote), but nucDNA coding for mtDNA
repair enzyme(s). And by the argument in the previous section it should
be (perhaps among other genes) for the gene on the X chromosome coding
for the exonuclease used in repair. More wild speculation: possibly
selection in the egg occurs also on the X chromosome during meiotic
division 1: the inactive X may have less damage and be used as a "spare
tire" and segregate to the egg (in duplicate) while the active X
chromosome may segregate to the first polar body. This should also be
easy to test experimentally (I realize that usually inactive genes are
repaired less efficiently, but maybe the Barr X is an exception).
Another consequence of these hypotheses is that an
in vitro culture of sperm (is it possible to culture sperm for long
periods?) will have more vigorous (after damage had a chance to happen
and be repaired more for X than for Y) and longer lived X type sperm
compared
to Y type sperm. In vivo I think we do see a slight selection of females
over males, which means that the time constant for damage is longer than
(but same order of magnitude) as the time sperm spend from meiosis to
fertilization (how long is that on the average?)