Due to the word limit of 3000, key subtleties/rationale/insights
obtained from our data were emitted. Furthermore, this made the paper
essentially incomprehensible for anyone not in the field of small-RNAs/sperm
biogenesis (I think). Even in its original length when we submitted to Cell,
some reviewers had trouble understanding rationale behind key experiments. To
supplement scientific discussion about our exciting findings in this paper, I
took the initiative to elaborate on some of these missing pieces. This is not
meant as a summary of what we wrote in the paper, but an extension of what we
found and its implications. Obviously I am not the most qualified author to do
so. Alas, everything written here represents my personal opinion, and in no way
reflects my co-authors/colleagues’ views of the science itself. If I
misrepresented either our or others’ science, or events that have transpired,
please accept my sincerest apologies here.
Prelude
Can
experiences of the parents be passed on to their kids biologically (not through
teaching, commonly referred to as environment), and not through DNA (or
classical genetics)? Parental diet and other stresses have long been postulated
to impact offspring development. Famously, the Dutch Hunger Winter provided
important epidemiological evidence that dietary restriction to the parents
could influence their kids in biological ways that don’t appear to be simply
genetic (https://en.wikipedia.org/wiki/Dutch_famine_of_1944). This phenomenon is termed epigenetic inheritance
(For detailed review of diet and epigenetic inheritance, please see “I’m eating
for two..” Rando and Simmons 2015 Cell).
In
2010, our lab published a paper (Carone et al. 2010 Cell) reporting that male
mice fed a low protein diet sired offspring that have altered liver metabolic
phenotype, mainly in the cholesterol biosynthesis pathway. Paternal epigenetic
inheritance was still a controversial (actually it is still controversial, a
lot of people don’t believe it occurs at all), but our paper and others
solidified the role that fathers can play in altering offspring phenotype
epigenetically. Other labs reported, in addition to low protein diet, other
diets such as high fat diet, and social stresses such as social defeat, can
influence offspring phenotype. One of my personal favorites is early childhood
trauma, specifically separating them from their mother (Gapp et al. 2014, Nature Neuroscience).
Surprisingly, the authors removed all doubt that this
information was passed on from the sperm by using In Vitro Fertilization (IVF)
– mice developed from IVF using sperm from mice that suffered trauma also showed behavioral and metabolic phenotype.
Now
this is all very intriguing, but how is the information passed on? Or like my
friend Juliano likes to say, everyone and their fucking cousin want to know the
mechanism (inside joke). Interestingly, the Gapp study indicates that the
information is stored in small RNAs that are carried by the sperm. This is a
very exciting possibility, but a little more about sperm to understand why this
is so groundbreaking.
The Sperm – an amazing
cell
Every
cell type has their particular awesomeness, but I think sperm is definitely one
of the most awesome. We already know quite a lot about sperm. Mature sperm is a
haploid cell, and during its development from a diploid germ cell, sperm not
only loses half its DNA but becomes devoid of most organelles that all
eukaryotic cells contain. All sperm is conventionally thought to retain are
mitochondria in the mid-portion of the sperm tail (mid-piece), and obviously
the haploid nucleus, its main cargo. During this process, everything that makes
a normal cell is degraded – ribosomes (translation of proteins), endoplasmic
reticulum, etc. The mitochondria are extremely important since the sperm needs
energy to swim through the female reproductive tract to reach the oocyte (egg).
Once there, sperm also carry enzymes on the front of its head (the acrosome,
which undergoes reactions to activate these enzymes called capacitation) that
allows it to penetrate the tough exterior of the oocyte. It is now believed
that after the sperm head breaks through the cellular membrane of oocyte, it
kind of fuses with the oocyte and expels its content – the nucleus, and also
the midpiece, containing the mitochondria. Since we all only carry our mother’s
mitochondria, this means that the male mitochondria are specifically targeted
for degradation (this is really cool! I included this because it’s awesome, but
also illustrates the complexity of fertilization and early stages of mammalian
life). This also means that nothing is supposed to be contributed by sperm
other than its haploid genome. Then, if any epigenetic information is to be
passed on, this information must reside somehow in the genome itself, or
changes to DNA methylation or histone marks.
A
few papers have reported changes in DNA methylation and histone modifications,
but these changes are surprising for two major reasons. First, during
spermatogenesis DNA is repackaged from histones into protamines. While some
studies suggest that histones are retained in certain regions of the genome,
the regions change depending on how you treat the sperm for deep-sequencing and
the field is still very much in flux. Studies from Antoine Peters and Bradley
Cairnes labs suggest that histones are retained in promoter regions of genes
that are involved in development. Therefore the marks of these histones, which
can theoretically be modified during spermatogenesis in response to
environmental changes, can influence phenotype in the offspring. However, data
from our lab suggest that histones are probably not retained in promoters, but
instead in gene desserts. In addition, these histone marks must become diluted during
cleavage of the zygote. How then, can they impact gene expression later into
the offspring’s life? Histone marks are definitely not copied faithfully in a
developing zygote… This leads to the second point. DNA methylation is globally erased
during spermatogenesis, and once again during zygotic activation, in addition
to histone marks. While there are regions in the genome that escape
demethylation, these regions are usually also gene desserts, and are thought to
not contribute to gene regulation. Additionally, how can environmental stress
impact differential demethylation in the germ-line? The testis is thought to be
separated from the rest of the body (soma-germline barrier), famously
postulated by August Weismann.
If
we entertain the possibility that the sperm carries potential carriers of
epigenetic information other than the well-known DNA methylation and histone
modifications, what can these be? Little is known about what other contents of
the sperm, if any, are absorbed into the oocyte. Even less is known about how
these contents can be altered during the maturation of sperm, a critical period
where changes in the environment could theoretically influence the development
of sperm, which can then lead to changes in payload, and therefore information
passed on to the offspring. But even if we understand what these magic
epigenetic contents are, and how they become influenced by the environment as
the sperm matures, we know nothing about how changes to these contents can
influence development of the offspring. The sperm head is tiny compared to the
oocyte (the entire sperm with the very long tail measures 1/3 of the oocyte),
so what can it carry with its limited space that can influence anything in the
oocyte?
Small RNAs in sperm?
To reiterate: since everything is degraded as the sperm
matures, what other contents can conceivably be epigenetically modified during
sperm maturation, and passed on into the oocyte, eventually influencing the
development of the zygote (fertilized oocyte)? Small RNAs have long been
understood in other organisms to carry epigenetic information.
Small-interfering RNAs (siRNAs) in bacteria fed to worms can be transmitted
many generations to silence genes, and their pathways have been very well laid
out by efforts of Dr. Craig Mello, nobel laureate from our institute, and many
other amazing scientists. Small RNAs also play very important role in
epigenetic inheritance in plants, famously purported by a giant in the field
Dr. Rob Martienssen. This is where the link to small RNAs in the Gapp paper was
so intriguing. Obviously this possibility was realized much earlier than the
Gapp paper (circa 2008/2009?, maybe earlier), and people in the lab have been
trying to sequence small RNAs in sperm of low protein vs. control fathers since
2010 (I think).
Interestingly,
sperm contain very little intact RNA, but surprisingly abundant amounts of
small RNAs. This is not that surprising given that RNAs are degraded during
sperm maturation, right? Well, it doesn’t appear that these small RNAs are
simply degradation products. First, there are abundant micro-RNAs, small RNAs
that are known to serve biological function in moderating gene expression. In
addition, there is an over-representation of “tRNA-halves” with characteristic
length in our deep-sequencing results, we judge to be at least 10,000-50,000
copies of the most abundant tRNA-half species. These so called “tRNA-halves”/tRFs
map to the 5’ of tRNAs, particularly those that are more abundant in most
cells, such as tRNA-Glycine and tRNA-Valine. In the literature, tRNA-halves are
characteristic cleavage products by a protein called Angiogenin in mammalian
cell lines, particularly during stress. The stress part we think is a red
herring, but supports the claim that tRFs are not random degradation products.
However, whether these tRFs in mature sperm are cleavage products of Angiogenin
is unknown, and is actively under investigation in the lab.
On tRF biogenesis
The origin of tRFs in mature sperm, regardless of the
protein that makes them, is one of the most controversial parts of our paper,
and there is debate even within our lab as to where tRFs in mature sperm come
from. As sperm transit through the male reproductive tract from the testis,
where it is not yet motile, to the cauda, where it becomes motile and “mature”,
it transits through a very long winding tubule/organ called the epididymis. The
epididymis is a very active organ in that it secretes tons of vesicles that
appear to fuse with the developing sperm to help/nuture it to maturity, a
process well worked out by Robert Sullivan and others. Actually, this is
controversial because it was shown in a very old paper that sperm could mature
in an artificial epididymis that is devoid of vesicles, and that all the sperm
really needs is time, which the epididymis provides with its slow flow from the
testis to the cauda. This controversy aside, many in the field believe that
vesicles shed by the epididymis, particularly exosomes, or epididymosomes, carry
important proteins to aid the sperm maturation process.
In
our paper, we isolated small-RNAs from epididymosomes and deep-sequenced them.
Intriguingly, the payload of epididymosomes at various regions of the
epididymis very closely mirrors that of the sperm isolated from those regions.
Either this is a gigantic coincidence or epididymosomes can fuse with sperm and
deliver its cargo, including small-RNAs to sperm as it transits through the
epididymis. An alternative hypothesis is simply that we are not washing the
sperm well and that epididymosomes are stuck on them. This doesn’t seem to be
the case, since stringent washes and treatment with lysis buffer that all but
destroys everything except sperm, leaves the small-RNA profile of sperm pretty
much the same. Another way to argue for our case is that if small-RNAs are
delivered by epididymosomes, there should not be much of these tRFs, for
example, in testicular sperm, those that have yet to enter the epididymis.
Well, we couldn’t yet isolate enough testicular sperm for small-RNA sequencing
so we did the next best thing, isolating late spermatids and also small-RNA
deep-sequencing whole testis. Here, we found very low levels of tRFs,
suggesting that tRFs indeed arise after sperm exit the testis. Lastly, we
reconstituted epididymosomes and immature sperm and saw that the sperm actually
“picked up” small-RNAs from the epididymosomes. All these data together suggest
that the epididymosomes at least partially contribute to the small-RNA payload
of maturing sperm.
However,
there are a few interesting caveats/details we omitted due to word limit.
First, all these experiments don’t rule out the possibility that tRFs and other
small RNAs are cleaved in the sperm itself during maturation, and not only from
epididymis fusion. In fact, data from our lab and other papers that have come
out show that certain RNases exist in the acrosome and the midpiece, like
Dicer, an important micro-RNA cleaver. In addition, it appears that tRFs go
from next to zero to over 60% of sperm small-RNAs between a late spermatid and immature
sperm in the initial segment of the epididymis. To me, this argues largely
against the hypothesis that epididymis fusion contributes the majority of
small-RNA payload in sperm, and that tRFs are mostly generated during the last
stages of sperm elongation as the sperm enters the testicular lumen. To verify
this, we have to sequence small-RNAs from testicular sperm. Eagerly looking forward
to this data from us and others!
Why do we care about tRF
biogenesis in sperm?
Whether tRNAs are cleaved in situ or fused to sperm via
exosomes is a debate for later. However, it seems clear that exosomes from the
epididymis do indeed carry tRFs and do eerily track tRF payload in maturing
sperm. As mentioned, the epididymis is a highly secretory organ that
communicates with the rest of the body. This is then, intriguingly, the place
where stresses to the parents can be “sensed” and metabolically coded/transferred
to the sperm, then to the offspring! Is this the mammalian soma-to-germline
transfer location that has eluded biologists for decades? This claim, not
directly made in the paper but eluded to, is the most controversial, but
interesting part of our paper.
Our data shows that certain tRFs and micro-RNAs,
particularly tRF-Glycine-GCC and let-7 family
micro-RNAs, clearly differ in mature sperm isolated from fathers fed different
diets. Manipulating levels of these tRFs, particularly tRF-Glycine-GCC, in the
oocyte significantly influences expression of a set of genes during
zygotic-genome activation at the two-cell stage. We also see changes in these
genes when we manipulate tRF levels in embryonic stem cells, an easier model to
work with and also giving us an additional line of evidence that tRFs regulate
the expression of these genes during early development. Serendipitously, these
genes that change expression have been previously shown to impact gastrulation
– specifically the number of cells that become the placenta, a key modulator of
metabolism in the development offspring.
Recap
In summary: diet impacts tRFs in sperm (through the
epididymis, we think), which appear to influence expression of certain genes in
the developing zygote that can influence its metabolic phenotype later in life.
The data presented in our paper may not be conclusive in certain places, but it
raises very interesting questions and possibilities, and represents a rich
resource for developmental biologists interested in small-RNAs in developing
sperm and how changes in these small-RNAs could influence gene expression in
the zygote.
We
have amassed a tremendous amount of high quality deep-sequencing data,
including hundreds of single-embryo RNA-sequencing, small-RNA libraries from
sperm of various developing stages, and ribosome-profiling data from embryonic
stem cells. In addition, our paper presents data from not only mice but also
bull, lending evolutionary significance to the results. Finally, our paper
presents a unifying theory for paternal epigenetic inheritance!
Angry rant: On word
limits
Who
doesn’t want to publish in the prestigious journals Science, Nature, and Cell?
Who knew how hard it was to do so? Well I had some inkling of how hard it could
be, but the actual process from our perspective was much more difficult than
reasonable expectation, and what we ended up with felt like a sell-out, frankly
speaking. Of course you’re saying, you published in Science why are you
complaining. To paraphrase my boss, dude, our paper ended up being only 3000
words long! That means each word cost the National Institute of Health a few
thousand dollars (assuming a million dollars per year)!!! That also means the
work of twenty people over three-four years, ideas that have developed over a
decade by one of the smartest people I know (my boss), cannot be summarized in
more words than a middle-school book report. We did not even have enough space
to explain why we did the experiments. Yes one could explain themselves in the
supplementary information, but who really reads that? (Honestly, I don’t even
know if the reviewers read that in much detail, and for our paper to be
described as descriptive by a particular reviewer? Isn’t the idea of a
scientific paper to “describe” science?) Anyways, if this is what it takes to
publish for that word at the end of everyone’s CV – Science – I don’t think
it’s worth it. Who knows, in an age where nobody has enough attention span to
read more than 3000 words at a time, it probably doesn’t matter anyways…
No!
Science, the namesake of the journal we published in, shouldn’t be limited by
the attention span of the reviewers or the editor or the journalist that is
reporting the “groundbreaking” science. Particularly since we are in the electronic
age, word limits seem particularly ridiculous. I’d personally rather publish in
a journal that doesn’t have word limits, thank you very much. Too many times I
have read very important papers in my field and was left with a bad taste in my
mouth, by how much detail was lacking to fully comprehend the science, by how
much great work have been skipped over and piled in the supplementary, by how
jam-packed the figures are the fonts in them give me a headache to attempt to
read. The process is flawed, and some are trying to fix it. For example, I love
elife! But to truly change how big journals operate, we need to abolish the all
but too common practice of looking at the only the journal names on someone’s
CV. We need to take power away from big journals, and make them small. In
economics this is called monopolies. Science, Nature, and Cell would have been
sued for competition lawsuits. They cannot have a monopoly over “impactful
science”! The toxic cycle of big journals controlling the content of published
materials should be abolished!
