The Cell Cycle

If you're not familiar with the cell cycle, here's a very brief summary for you. Basically, the cell cycle is
the "program" for cell growth and cell division (proliferation). There are 4 broad phases of the cell cycle:
G1 (and G0), S, G2, and M.

The G1 (Gap 1) phase is characterized by gene expression and protein synthesis. This is really the only
part of the cell cycle regulated primarily by extracellular stimuli (like mitogens and adhesion). Anyway,
this phase enables the cell to grow and to produce all the necessary proteins for DNA synthesis. Why is
this important? Well, it primes the cell to enter the next phase: S (Synthesis) phase.

During the S phase, the cell replicates its DNA... so it now has 2 complete sets of DNA. This allows the
cell to divide into two daughter cells, each with a complete copy of DNA. But, before the cell can do this,
it needs to enter the third phase of the cell cycle: the G2 (Gap 2) phase.

During the G2 phase, the cell again undergoes growth and protein sythesis (it needs enough proteins
for 2 cells!)... priming it to be able to divide. Once this is complete (by the way, there are many
"checkpoints" along the way!), the cell finally enters the fourth and final phase of the cell cycle: the M
(Mitosis) phase.

During the M phase, the cell splits apart (called cytokinesis) into two daughter cells. Now, the cycle
has been completed! What do the cells do know? Well, there are two choices... it can either start the cycle
again by entering G1, or it can become quiescent by entering G0.

Okay, there's a very brief introduction to the cell cycle. Now, can you see the inherent problem with
this cycle? Well, if it wasn't controlled, your cells would continue to grow and divide... over and over
again! So, there are a number of proteins that regulate and control the cell cycle. They tell the cell when
is the proper time to grow and divide, and they stop the cell when the time's not right. That's what the
rest of this page will be about... proteins involved in controlling the cell cycle. The main families that I
will discuss are the cyclins (especially cyclin D and cyclin E... and little about cyclin A), the cyclin-
dependent kinases (especially CDK4, CDK6, and CDK2), the CDK inhibitors (especially p16, p21, and
p27), and the tumor supressor genes (especially Rb and p53). Because of the complexity of these
pathways, I'll divide them into two separate pages: the Rb pathway and the p53 pathway.

Of course, one of the main clinical interests of cell cycle control is cancer. Cancer can be very briefly
described as uncontrolled cell growth and proliferation (as well as metastasis, or the invasiveness of
cancerous cells into other tissues). Therefore, research in understanding cell cycle control has many
implications for cancer, especially for the development of therapeutics. Thus, throughout this page, as I
describe each of the above proteins and their roles in controlling the cell cycle, I will also try to mention
their abnormal role (either gain of function or loss of function) in cancer.
 Rb Pathway
Now, during most of the cell cycle, the cell does not require any extracellular signals. However,
during the G1 phase, the cell does require these signals... up to the restriction point (R). R is near the
end of the G1 phase, just prior to entry into S phase, so this is an important point in regulation in the cell
cycle. That's why the cell requires extracellular signals to tell it if it should proceed through R and
continue with the rest of the cell cycle (once it passes R, it is mostly independent of any other extrinsic
inputs).

So, what are these signals? A good example are growth factors, polypeptides secreted by cells that are
recognized by other cells (in some cases, the same cell will recognize the growth factor, this is called an
autocrine loop). How does the cell recognize growth factors? By using growth factor receptors, of
course! Growth factor receptors are single transmembrane proteins called receptor tyrosine kinases
(RTKs) because they contain tyrosine kinase domains in the cytoplasmic portion of the receptor. So,
when they bind growth factors, the RTKs dimerize and transphophorylate critical tyrosine residues in
the cytoplasmic portion of the other receptor.

Now these phosphorylated tyrosines (Y-P) act as docking sites for proteins containing SH2 (src-
homology 2) domains. SH2 domains bind to Y-P, which in this case brings the proteins to the cell
membrane. Why would this be important? Well, some of the proteins that are recruited (see below) need
to be at the membrane to carry out their function. So, they are normally cytoplasmic, then when the
receptor is activated, they translocate to the membrane. Translocation and cellular localization are
critical determinants in signal transduction, so you'll see these themes popping up frequently. What are
some of these proteins? Great question, and here are three examples:

Phosphatidylinositol-3-kinase (PI3-K): This protein (actually 2 proteins, one p85 regulatory subunit
and one p110 catalytic subunit) translocates to the membrane, binding to Y-P of the receptor via the SH2
domain on p85. Once at the membrane, the p110 catalytic subunit can phosphorylate the D3 position of
the inositol ring of phosphatidylinositides (a type of lipid in the cell membrane). This phosphorylated
lipid can then bind to the PH (pleckstrin homology) domain of cytoplasmic proteins, recruiting them to
the cell membrane and leading to their activation. One important protein in this pathway is the pro-
survival protein Akt.

Phospholipase C-gamma (PLC-gamma): This protein is recruited to the membrane by binding Y-P
residues via its SH2 domain. Again, this protein needs to be at the membrane to carry out its function
since it hydrolyzes phospholipids (phosphatidylinositol-4,5-bisphosphate in particular) into
diacylglycerol (DAG) and inositol-1,4,5-trisphosphate (IP3). DAG then goes on to assist in the
activation of PKC while IP3 releases calcium from intracellular stores (from the endoplasmic reticulum
or sacroplasmic reticulum). Calcium is involved in the activation of a variety of proteins, including the
classical PKCs and calmodulins.

Grb2-SOS complex: Grb2 is an adaptor protein containing 1 SH2 domain (which binds to Y-P
residues) and 2 SH3 (src-homology 3) domains. These SH3 domains bind to proline-rich regions of
other proteins, especially one called SOS. SOS is a guanine nucleotide exchange factor (GEF) for the
small G-protein Ras. Now, since Ras is tethered to the membrane via a farnesyl group, SOS has to get to
the membrane to activate it (see below).
 Well, since Ras is so important in passing R, I want to spend a little more time on it. So, when SOS
activates Ras, it exchanges the bound GDP for GTP. When Ras binds GDP, it is inactive. However,
when it binds GTP, it changes its conformation and can now interact with downstream effectors,
including RalGDS, PI3-K, and Raf. So Ras, like other G-proteins, operate like "molecular switches":
GDP = off and GTP = on. Below is a figure from a recent Nature paper illustrating the structure of Ras
bound to SOS.

Boriack-Sjodin et al. (1998). Nature 394: 337-343.

The N-terminus of SOS is colored blue while the catalytic domain is colored green. The region in cyan are the conserved
regions of most Ras GEFs. On Ras, the switch 1 region is colored orange, while the switch 2 region is colored red. These regions
are important for stabilizing the guanine nucleotide and interacting with downstream effectors. They change confomation
whether they're bound to GDP or GTP, and the downstream effectors can only recognize the switch regions in the GTP-bound
conformation. Pretty cool, eh?

Now, Raf is a serine/threonine kinase that is recruited to the membrane by binding to active (thus,
GTP-bound) Ras. Once at the membrane, Raf is activated (this is still being worked out, but you may
want to check out my PAK page to see how PAK plays an important role! Interestingly, PAK has been
shown by Mark Marshall's lab to phosphorylate serine 338 of Raf1, which they suggest is important in
its activation. Akt has recently been demonstrated to phosphorylate serine 339 of Raf1, which inhibits its
activation. And, lastly, Src has been shown to phosphorylate tyrosine 340, which also is involved in the
activation. So, within these three residues [S338, S339, and Y340] you have a variety of ways to
phosphorylate and regulate Raf activation!). Active Raf then phosphorylates and activates MEK. MEK
(MAPK and ERK Kinase) is a dual specificity kinase since it can phosphorylate both threonine and
tyrosine residues, which is does on ERK. ERK (Extracellular Regulated Kinase , also called MAPK for
Mitogen Activated Protein Kinase) is a serine/threonine kinase that then phosphorylates cytoplasmic
proteins, like p90RSK, or translocates into the nucleus where it phosphorylates transcription factors,
such as the TCF family (Ternary Complex Factors, like Elk1, Sap1, and Sap2) and c-Myc family of
transcription factors. The activation of these various transcription factors (either directly by ERK or
indirectly by cytoplasmic proteins such as p90RSK) then activate the transcription of responsive genes,
one of which is critical in passing R. This is cyclin D gene. Thus, the ERK activation cascade (Ras->Raf-
>MEK->ERK) is a critical pathway, stimulated by growth factors, in the G1 to S phase

So, now we have cyclin D expression. What does this do? Well, the cyclins are just what they sound
like, they cycle. So, depending on which part of the cell cycle the cell is in, different cyclins will be
expressed. Below is an example of the cycling of the different cyclins depending on the phase of the cell
cycle.

Sherr, C.J. (1996). Science 274: 1672-1677.

So, you can see the cyclin D levels are increased during G1 phase (due to mitogenic stimulation of the
ERK activation cascade), and then decrease again once the cell divides (at the end of M phase). Similarly,
the other cyclins (E, A, and B) levels are also regulated depending on the phase of the cell cycle. So, what
are cyclins? What do they do? Well, cyclins bind to CDKs (cyclin-dependent kinases). For the cyclins
mentioned above, there are 3 different CDKs: CDK4/6, CDK2, and CDK1 (also called cdc2). These, as
their name implies, are kinases that are involved in phosphorylating different components of the cell
cycle machinery. They are themselves regulated by phosphorylation (by CAKs, CDK-Activating
Kinases) and de-phosphorylation (by cdc25, a tyrosine phosphatase).

Now, back to cyclin D. Cyclin D binds to CDK4 and CDK6. Usually, you'll just see this written (as I
did above) CDK4/6. Once bound, the CDK is now active and can phosphorylate it s substrate: Rb. Rb is
a classic tumor suppressor protein, meaning that it suppresses the cell cycle. It binds to a family of
transcription factors, called the E2F family. When bound to these E2Fs, it represses transcription of E2F-
responsive genes (such as DHFR, dehydrofolate reductase, and TK, thymidine kinase, both necessary
for DNA replication; as well as cyclin E and cyclin A, both necessary for cell cycle progression).
Anyway, Rb can also bind to and recruit HDACs, or histone deacetylases.
 Now, remember, DNA is wrapped around histone octomers! That means that the transcriptional
machinery does not have good access to the gene. In order to permit this access, the association between
DNA and the histones needs to be loosened. This occurs when histone acetyltransferases (HATs, like
CBP/p300) acetylate positive charges on histones, thus weakening the electrostatic interaction between
DNA and the histones (normally, the negative charged backbone of DNA has a strong association with
positively charged residues on the histones. Since acetyl groups are neutral, the negatively charged
DNA isn't bound so tightly to the histones anymore [a side note: histones can also be phosphorylated as
well. Recently it was demonstrated that p90RSK (remember? the protein activated by ERK?) can
phosphorylate histone H4. Is this significant? Well, being that phosphates are negatively charged, like
DNA, it may even open up the chromatin even more!]). Anyway, Rb recruits HDACs to remove the
acetyl groups from the histones, thus once again restoring the tight association of DNA with the histone
octomer. By preventing the transcriptional machinery access to the E2F-responsive genes, Rb is able to
suppress cell cycle progression.

A recent paper from Douglas Dean's lab down at Washington University (Zhang et. al. (1999). Cell 97:
53-61.) utilized a dominant negative E2F mutant (a "dominant negative" is a mutant protein that inhibits
the normal protein in the cell) to determine which is necessary for cell cycle progression: release of
suppression or E2F-dependent transcription. How can you tease these apart? Well, Zhang et. al. used
the DNA binding domain of E2F... this would block access to both the Rb-E2F complex and E2F itself to
DNA. So, this mutant releases the suppression of the Rb-E2F complex (since it cannot bind to DNA and
recruit the HDACs), AND it blocks E2F-dependent transcription (since E2F cannot bind to DNA). What
did they find? In cases where cell cycle progression is normally arrested (e.g. by p16INK4a), they were
able to get cell cycle progression (as well as expression of "E2F-dependent" genes like DHFR and TK).
This work suggests that in some cases, all you need is a release of the Rb-E2F-dependent suppression
(and not E2F-dependent transcription) for cell cycle progression. Pretty interesting, don't you think?

DePinho, R. (1998). Nature 391: 533-534.

 So, this is where cyclin D comes in. Again, Douglas Dean's lab published another Cell paper (it must
be nice, eh?) demonstrating the intramolecular interactions that inactivate Rb (Harbour et. al. (1999).
Cell 98: 859-869.). First, the active cyclin D-Cdk4/6 complex phosphorylates the C-terminus of Rb.
These phosphorylated residues then fold back onto Rb, associating with a positively-charged patch of
lysines (makes sense, right? The negatively charged phosphates would associate with positively
charged lysines...). Why is this special? Well... this patch of lysines just so happens to be at the HDAC
binding site on Rb! That means when the phosphorylated C-terminus folds back onto Rb, this would
displace HDAC and release the transcriptional repression!

Is that all? No way! Rb is still bound to E2F, right? So, how do we get Rb to release E2F? Now, that's
where the cyclin E-Cdk2 complex comes in. It's been known for some time that Rb needs to be
phosphorylated by both cyclin D-Cdk4/6 and cyclin E-Cdk2... in that order. Why this order? Well, the
cyclin E-Cdk2 complex binds to the C-terminus of Rb... but phosphorylates the middle (called the
"pocket domain") of the protein. So... if cyclin D-Cdk4/6 phosphorylates the C-terminus, it then folds
over onto Rb, which would then position cyclin E-Cdk2 perfectly to phosphorylate the pocket domain of
Rb! Once this pocket domain is phosphorylated by cyclin E-Cdk2, E2F is then released and can then start
to activate the transcription of E2F-dependent genes! Isn't this really cool? I would HIGHLY recommend
this paper to everyone out there... it's great! A diagram is depicted below summarizing what I stated
above...

Harbour et al. (1999). Cell 98: 859-869.

Okay, this picture didn't come out

as well as I hoped, so let me take
you through it. In the first panel,
you can see Rb in its "inhibiting
state". The pocket domain is
designated by the "A" and "B", the
C-terminus is seen coming off like a
tail, and the lysines are the "k"s
along the side. You can see HDAC
binding to "B" region, and E2F
binding to the pocket domain. In
the second panel, the C-terminal tail
has been phosphorylated by cyclin
D-Cdk4/6, and now folds over Rb
to associate with the positively
charged lysine patch... this then
displaces HDAC and removes the
associated transcriptional
repression. In the third panel,
cyclin E-Cdk2 binds to the C-
terminal tail and phosphorylates
the pocket domain (they
determined that S567 was the
critical residue here). This
phosphorylation disrupts the
pocket domain structure and
releases E2F! I hope you can read it
enough to make sense out of it!
 How is cyclin E regulated? Well, I've seen a couple of hypotheses that explain this regulation, and
they may not be mutually exclusive.

First, cyclin E levels have been suggested to be regulated by Rb-E2F. So, when cyclin D-Cdk4/6
phosphorylates Rb, this leads to a partial inactivation (release of HDAC?), allowing cyclin E
transcription. Cyclin E then complexes with Cdk2 and leads to the additional phosphorylation of Rb in
the pocket domain and its subsequent complete inactivation (by releasing E2F). In other words, some
authors have shown the cyclin E does "cycle". That it is low in G0 and in early G1, and that near the end
of G1... it becomes induced.

The second hypothesis (again, these are not mutually exclusive events and probably occur together in
some way) is that cyclin E activity is more regulated by the CKIs, or Cyclin-dependent Kinase
Inhibitors. In this model, cyclin E is complexed to Cdk2, and does not really cycle so much. Well, how is
this complex regulated? Well, one of the CKIs ( more below!) called p27 binds to and inhibits this
complex. Then, upon mitogen stimulation, two things occur: one, cyclin D levels are increased and two,
p27 levels are degraded. How are cyclin D levels related to p27 inhibition of cyclin E-Cdk2? Well, its
believed that p27 (and p21) are "assembly factors" for cyclin D-Cdk4/6 complexes (check out a nice
article from Sherr's lab [Cheng et al. (1999). EMBO J 18:1571-1583.] for more information!). Where does
this p27 come from? That's right! From the cyclin E-Cdk2 complexes! So, by p27 leaving cyclin E-Cdk2
complexes (to bind to the cyclin D-Cdk4/6 complexes), cyclin E-Cdk2 complexes become active and can
now phosphorylate Rb (which again, can only occur after cyclin D levels are increased). The second
thing that happens is that p27 levels are also decreased upon mitogenic stimulation. How might this
occur? Well, believe it or not... cyclin E-Cdk2 complexes (when not bound to p27) can actually
phosphorylate p27... leading to its subsequent degradation by the ubiquitination-proteosome pathway!

So, what about these CKIs? There are two main families: the INK4 family and the Cip/Kip family. The
INK4 family consists of p15INK4b, p16INK4a, p18INK4c, and p19INK4d. Basically, the INK4 family
binds to and inhibits Cdk4/6 (INK4= INhibitors of cdK4). By binding, it does not allow cyclin D to bind,
and thus prevents the activation of Cdk4/6. Under conditions of stress or other times when a cell should
NOT undergo the cell cycle, the INK4 CKIs are expressed and prevent Rb phosphorylation.

The other family of CKIs consist of p21Waf1/Cip1, p27Kip1, and p57Kip2. These CKIs (unlike the
INK4 family) bind to and inhibit cyclin E-Cdk2 and cyclin A-Cdk2. By binding, these CKIs lead to the
inhibition of Cdk activity and thus prevents ability to stimulate cell cycle progression. While p21 levels
are induced during periods of stress (see my p53 page for more information), p27 is constitutively
expressed when the cell is quiescent (G0 and G1 phases of the cell cycle) and then is significantly
reduced at R and remains low until the end of the M phase (see the above cycling diagram). Remember,
p21 and p27 have been demonstrated to also act as assembly factors for cyclin D-Cdk4/6 (i.e. they're not
always inhibitory!).

Sherr, C.J. (1996). Science 274: 1672-1677.

Here then is a summary of the restriction point, R. Rb binds to an represses E2F while also binding to
and recruiting HDACs to repress transcription. Cyclin D-Cdk4/6 and cyclin E-Cdk2 phosphorylate and
inactivate Rb, releasing HDAC (which releases the transcriptional repression) and releasing E2F (which
then stimulates the transcription of genes necessary for cell cycle progression). INK4 CKIs bind only to
Cdk4/6, inhibiting their activation by cyclin D. Cip/Kip CKIs bind to and inhibit the cyclin E- and
cyclin A-Cdk2 complexes, while acting as assembly factors for cyclin D and Cdk4/6. When inhibitory,
these CKIs inhibit the phosphorylation of Rb and thus leads to cell cycle arrest. One last point... R is not
only regulated by mitogenic signals, but also by adhesion. Signaling from integrins then also play an
important role in the hyperphosphorylation of Rb as well as in the expression of cyclin A, Cdk2 activity
and p27 degradation. Thus adding another layer of complexity on this model. Why don't you check out
my Integrins page for details!

Now, one of the hallmarks of cancer is the inability to regulate R. Therefore, as you could probably
imagine a very common mutation found in cancer (probably the third most common mutation) is in the
Rb gene. When Rb is mutated and not functional, you lose most of the regulation of R, and thus the cell
cycle is unable to be properly regulated. Other common mutations in cancer include over-expression of
cyclin D, constitutively active Ras, and decreased p16INK4a expression (in fact, this is the second most
frequently disrupted gene in cancer next to p53, again check out my p53 page for details!). With an
understanding of the pathways regulating cell cycle progression from G1 to S phase, we can more
completely understand how mutations in these genes lead to unregulated cell growth... cancer.
 p53 Pathway
Another critical protein in regulating the cell cycle is the tumor suppressor protein p53. In fact, p53 is
the most frequently disrupted gene in cancer, illustrating its importance. What is p53 and why is it
important? Well, p53 is a DNA-binding protein involved in regulating the expression of genes involved
in cell cycle arrest (like p21WAF1/CIP1, check out my Rb page if you missed this!) and apoptosis (such
as Bax check out my Apoptosis page!). Apoptosis is basically cellular suicide. So, as you can probably
guess from its functions , p53 recognizes when something in the cell has gone wrong and either tells the
cell to stop growing (so the cell can repair any damage that has been done) or if all else fails, tells the cell
to kill itself (to prevent unregulated cellular growth... cancer) .

Now, p53 protein levels are normally kept very low within the cell. However, once stimulated (see
below), its protein levels are rapidly increased along with its half-life, while its mRNA levels remain
relatively unchanged. What does this suggest? That the regulation of p53 at the protein level (not DNA
or RNA level) is critical in its activation. An important negative regulator of p53 at the protein level is
Mdm2, which is actually a p53-responsive gene. So, can you see the negative feedback loop here? P53
becomes activated (see below) and then increased Mdm2 levels. Mdm2 then inactivates p53, turning it
off. As you can imagine, if you want to increase p53 protein levels within the cell, you are going to have
to inhibit Mdm2.

Giaccia, A.J. and Kastan, M.B. (1998). Genes and Development 12: 2973-2983.

This is a schematic of the domains of p53 including many of the modification (phosphorylation and acetylation)
sites involved in its stabilization, activation, or repression. See text for more details!

How does the cell inhibit Mdm2? Well, it depends on the stimulation. For example, DNA damaging
agents (like radiation) induce the activation of kinases (such as ATM and DNA-PK) that can
phosphorylate a critical serine residue in the Mdm2-binding domain of p53. So, when p53 is
phosphorylated here, it can no longer bind to Mdm2. Then, this is able to relieve Mdm2-mediated
inhibition of p53, right? Why would DNA damaging agents activate p53? Well, if your DNA is
damaged, you don't want to replicate it, right? If you did, you might produce cells with deleterious
mutations, which might then lead to cancer. So, p53 recognizes when the cell has sustained DNA
damage and halts the cell cycle so the cell can repair the damage, or in many cases, just tells the cell to
kill itself (apoptosis).

Another mechanism to inhibit Mdm2 is by oncogenes, constitutively active mutant proteins that
continually tell the cell to grow (such as Ras). Why would oncogenes activate p53? Again, you don't
want cells to grow uncontrollably, right? That's cancer! So, p53 recognizes when this happens and halts
the cell cycle. However, oncogenes do not lead to the activation of ATM or DNA-PK, in fact, oncogenes
don't even lead to the phosphorylation of p53 in the Mdm2-binding domain! So, how do oncogenes
inhibit Mdm2? By inducing the expression of a tumor suppressor protein called p19ARF. Now, this is a
really interesting protein (and relatively new!), so I'll spend a bit more time talking about it below.
 de Stanchina et al. (1998). Genes and Development 12: 2434-2442.

Remember in my Rb page that two Cdks are activated near the end of G1 and in S phase: Cdk2 and
cdc2 (Cdk1). Well, once the cell has begun to cycle (and has passed the restriction point), you don't want
p53 to halt the cycle. So, these two Cdks can also inhibit p53. How? Well, Cdk2 and cdc2 are thought to
keep p53 in the cytoplasm when the cell is not in G1. Remember CDK2 activity (by binding to cyclin E
or cyclin A) is increased at the end of G1, which can then phosphorylate p53 and get it out of the
nucleus so it does not interfere with DNA synthesis.

Giaccia, A.J. and Kastan, M.B. (1998). Genes and Development 12: 2973-2983.

So, is p53 relevant in cancer? Definitely! In fact, p53 is the MOST commonly disrupted gene in cancer.
So, you know it has to be important. Now, is it necessary for normal cell cycle progression? p53 is a
checkpoint. It recognizes when something has gone wrong (for example the DNA has been damaged or
the cell is being stimulated by an oncogene) and immediately halts the cell cycle to prevent the cell from
becoming cancerous. So, if you lose p53, your cell loses this important checkpoint and may go on to
become a cancerous cell. Not only are p53 mutations found in cancer, but so are overexpression of
Mdm2 (which would act to inhibit p53) as well as the loss of p19ARF. Remember in my Rb page when I
said that p16INK4a is frequently disrupted in cancer? Well, remember, p19ARF is also at that same
locus! And, since most of the disruptions that occur at this locus are deletions, it eliminates both
p16INK4a and p19ARF! Now you can probably see why this is the second most disrupted locus in
cancer. By eliminating these two tumor suppressor proteins, you lose the regulation (in part) of the Rb
and the p53 pathway.
 Apoptosis
Look down at your hands, probably one of them is on your computer's mouse right now. Go ahead, look. Unless you are
suffering from a genetic disease called polydactyly, you will see 5 fingers, right? And, if you look down to your feet, you will
5 toes. However, when you were a developing fetus, you did not start out having 5 fingers and toes. No, instead, you had
hands and feet that looked more like paddles. But, during development, the cells in the tissue in between your soon-to-be
fingers and toes died. And, because they died, you are now able to manipulate objects very carefully, wear rings, and even
hitchhike (although I would have to strongly recommend not hitchhiking these days, you never know what weird person
might be picking you up!).

Your brain. When you were born, your brain had many more neurons (brain cells) than it does now. Does that mean you
were smarter when you were born? Of course not! However, your brain has to wire itself properly, and it will use your
experience to do it. So, over the first few years of your life, your brain was busy making and breaking connections with other
neurons (called the pruning period). Those neurons that did not make enough connections, or more importantly, the right
connections, would die. Therefore, you are left with a brain in which all the neurons are wired correctly! As an aside, each
one of your neurons is connected to about 10,000 other neurons! Pretty amazing circuitry! And, totally unlike computers (I
hate when people compare computers to brains), those connections (called synapses) are more or less plastic. That means,
with experience, new synapses can be made or old synapses can be lost. And, to provide even more complexity, individual
synapses can become stronger or weaker! Now, you show me a computer that can do that!

Do you like to lay out in the sun? I do. But, then again, I grew up in Los Angeles and Phoenix (Arizona) where the sun is
shining all year long. However, ultraviolet (UV) light from the sun can cause DNA damage, especially in the cells in your
skin. So, what happens if your cell cannot fix the damage? Well, that cell will kill itself! Really! Why? If it stayed alive and
tried to undergo the cell cycle, it might become cancerous. So, to protect the rest of your body, your cell will give up its life.
Pretty utilitarian, don't you think?

These are all examples of programmed cell death, called apoptosis. This is different from another
form of cell death, called necrosis. What is the difference? As a simple analogy, necrosis can be thought
of as murder, while apoptosis can be thought of as suicide. So, in one case (necrosis) the cell does not
want to die, but is killed by some external factor. In the other case (apoptosis), the cell chooses to die.
Because of these differences, cell death can be determined by looking at the tissue: necrosis causes
inflammation, apoptosis does not. Why wouldn't apoptosis cause inflammation?

Remember, cells undergo apoptosis in many cases as a result of the developmental program or to
protect the rest your body from harm. Think of your body as a Utilitarian community (John Stuart Mill
would be proud!). So, your cell is willing to give up its life for the better of the community (your body).
In doing this, it does not want to cause inflammation, right? So, apoptosis is characterized by many
morphological features that eventually lead to the nice packaging of the cell into little packages that can
be easily phagocytized by other cells. These features include: membrane blebbing, cytoplasmic and
nuclear condensation, DNA fragmentation , and formation of apoptotic bodies (the little packages).

Now, there are many signaling pathways that may lead to apoptosis. These pathways converge onto
other pathways leading to the commitment to die. That is where I would like to pick up. Instead of
discussing the many signaling pathways (I will mention a couple of course!), I want to concentrate on
the aspects of apoptosis that are shared by most cells. That is, once a cell commits to die (how it gets to
this point may differ), the program to cell death is pretty similar. So, on the rest of this page, I will talk
briefly about how researchers were able to use model systems (the worm in this case) to better
understand apoptosis in mammalian cells. Then, on the other pages, I will talk about two major classes
of proteins involved in apoptosis in most cells: the Bcl-2 family and the caspase family.

So, I mentioned about the worm in the previous paragraph, right? Now, you might be thinking, what
worm? Well, geneticists and developmental biologists love model systems that are easier to work on and
understand than mammalian systems. Then, by understanding these model systems, the ideas can then
be applied to mammalian systems to understand them. Common model systems include yeast (single
cell eukaryotes), fruit flies (Drosophila), nematode worms (C. elegans), and zebra fish. In the case of
apoptosis, it is C. elegans that helped researchers understand apoptosis in mammalian cells. Basically, C.
elegans has 4 major genes that regulate programmed cell death (their programmed cell death is not
really apoptosis since it is not characterized by all the features mentioned above, which occur in
mammalian cells): ced-9, egl-1, ced-4, and ced-3. Ced-9 is a protector (prevents programmed cell death),
while egl-1, ced-4, and ced-3 are killers (promotes programmed cell death). You can see how these
proteins fit together below.
 Hengartner, M. (1998). Science 281: 1298-1299.

Ced-9 is a protein that binds to the outer membrane of the mitochondria, and also binds ced-4 and ced-3. This keeps ced-3
inactive and the cell is alive. However, when egl-1 comes in, it displaces the ced-4/ced-3 complex from ced-9. Once free, the
ced-4/ced-3 complex can oligomerize and become active. Once this complex becomes active, it will kill the cell.

Now, this had been worked out pretty well in C. elegans, so the question was: are there mammalian
homologues to these proteins? Indeed there are. Ced-9 represents the pro-survival Bcl-2 mammalian
proteins, while egl-1 represents the pro-apoptotic Bcl-2 proteins. The ced-4 homologue is thought to be
a protein called Apaf-1, while the ced-3 homologue is the caspase family of mammalian proteins. So, I
will address the Bcl-2 family of proteins (both pro-survival and pro-apoptotic) on one page, and I will
discuss the caspase family (and Apaf-1) on the other page. Here is a figure below so you can see the
homology of these pathways in C. elegans compared to mammals. Anyway, I hope you enjoy these
pages! Please be sure to check out my other cell biology pages and sign my guestbook. I love to hear
what you guys think of my page and how I can make it better! Thanks again for coming by!

Adams, J. & Cory, S. (1998). Science 281: 1322-1326.

 Bcl-2 Family
Okay, now that we have gone over a brief intro into apoptosis, I can continue my discussion with the
Bcl-2 family of proteins. As you remember from my last page, Bcl-2 proteins are the mammalian
homologues to the C. elegans pro-survival protein ced-9 and pro-apoptotic protein egl-1. And, indeed,
Bcl-2 family members can be both pro-survival and pro-apoptotic. The pro-survival family members
that you might often hear about (and will be discussed in this page) are Bcl-2 and Bcl-XL. The pro-
apoptotic family members that I will discuss (and, again, you will hear about often in papers) are Bax,
Bad, and Bid.

Adams, J. & Cory, S. (1998). Science 281: 1322-1326.

So, above are the classification of the different Bcl-2 family members with their domain structures.
Bcl-2 and Bcl-XL have all 4 domains (BH1, BH2, BH3, and BH4), the Bax has only BH1, BH2, and BH3
(no BH4), and Bad and Bid are part of the BH3 sub-family, so therefore only have the BH3 domain. Also
illustrated are different phosphorylation sites (indicated by the black arrows on Bcl-2, and by the way ,
these are negative regulation sites), the dimerization domain, and interestingly, the pore-forming
domain

Adams, J. & Cory, S. (1998). Science 281: 1322-1326.

Now, the crystal structures of the Bcl-2 family proteins have really helped in understanding the
function of these members. For example, it was illustrated that the BH1, BH2, and BH3 domains form a
"pocket" that bind the BH3 domain of another family member. This helped illustrate how these proteins
can form homo- and heterodimers. Above, on the left, you can see the crystal structure of the BH1, BH2,
and BH3 domains forming this pocket and binding the BH3 domain of Bak. This led to the idea that the
pro-apoptotic Bcl-2 family members (like Bik) can bind to the pro-survival Bcl-2 family members (like
Bcl-XL) to displace Apaf-1. This is shown above on the right. What is Apaf-1? Well, I will go into more
detail in my page discussing the caspase family of proteins, but let me just say that it initiates the
caspase cascade to induce apoptosis.
 Okay, so this brings me to how Bcl-2 proteins protect or initiate apoptosis. Now, there may be other
ways, but I will discuss the initiation of the caspase cascade, i.e. the activation of Apaf-1. Apaf-1 requires
two co-factors to become active: ATP and cytochrome C (Cyt C). So, ATP is already in the cell ( you may
also see the requirement of dATP), but what about Cyt C? Cyt C is normally in the intermembrane space
of the mitochondria and participates in the electron transport chain and oxidative phosphorylation.
However, if Cyt C gets out into the cytoplasm , it binds to Apaf-1 (along with ATP) leading to the
activation of Apaf-1. So, how does Cyt C get out of the mitochondria? Great question! That is where the
Bcl-2 proteins come into play. It has been shown that the pro-survival Bcl-2 proteins block the release of
Cyt C from the mitochondria, while the pro-apoptotic Bcl-2 proteins (especially Bax) lead to the release
of Cyt C and subsequent Apaf-1 activation.

So, this is where crsytallography has been helpful again. The crystal structure of the Bcl-2 proteins is
very similar to the pore-forming proteins of bacteria. Are you thinking what they were thinking when
they discovered this? Perhaps, the Bcl-2 proteins are forming pores in the outer membrane of the
mitochondria and letting Cyt C to escape into the cytoplasm! Does any evidence support this?
Definitely. So, many of the Bcl-2 family members have membrane anchors in the C-terminus (check the
figure at the top of the page), and this has been shown to target many Bcl-2 family members to the outer
membrane of the mitochondria. So, they are definitely in the right place. The current thinking, then, is
that the pro-survival Bcl-2 proteins prevent the release of Cyt C from the mitochondria by forming
homodimers (although this is not fully understood yet) and by forming heterodimers with pro-
apoptotic family members (preventing Bax from forming pores, for example). However, if there is a
shift in the balance of pro-survival and pro-apoptotic family members, then Bax can then form
homodimers, form pores in the mitochondria, and release Cyt C. Does this make sense? I hope so! Pretty
cool, eh?

Chau, D.T. & Korsmeyer, S.J. (1998). Annual Review in Immunology 16: 395-419.

Now, there are various models of how the ratio of pro-survival (e.g. Bcl-2) and pro-apoptotic (e.g.
Bax) inhibit or initiate apoptosis (above). One model says that Bcl-2 inhibits apoptosis, and Bax relieves
this inhibition to initiate apoptosis. Another model suggests that Bax induces apoptosis, and Bcl-2
inhibits this induction. And another model says that independently, Bcl-2 inhibits apoptosis, and Bax
induces it. The most accurate model (not depicted above, sorry!) is a combination of all three. That is,
Bcl-2 inhibits apoptosis independently of Bax, AND inhibits the induction of apoptosis by Bax. Further,
that Bax induces apoptosis independently of Bcl -2, AND relieves the inhibition by Bcl-2. Pretty complex,
eh?

So, if the ratio of pro-survival to pro-apoptotic Bcl-2 proteins is so important, how is this regulated? I
thought you would never ask! I will give an example involving transcriptional regulation first. So, have
you checked out my p53 page yet? Anyway, p53 is a DNA binding protein that regulates the expression
of genes involved in cell cycle arrest (like p21WAF1) as well as apoptosis. What kind of apoptosis-
related genes? Bax! That's right! Our old pro-apoptotic friend! So, when p53 is activated (by DNA
damage, stress, oncogenic stimulation... see my p53 page for details on p53 activation), it induces the
expression of Bax, throwing off the ratio of pro-survival to pro-apoptotic Bcl-2 proteins. Now, there is an
excess of Bax at the mitochondria, leading to the formation of more Bax-Bax homodimers (since there is
not enough Bcl-2 proteins to titrate out the Bax anymore), leading to the release of Cyt C, and the
activation of Apaf-1. However, there is another mechanism to shift the balance between pro-survival
and pro-apoptotic proteins as well!
 Chau, D.T. & Korsmeyer, S.J. (1998). Annual Review of Immunology 16: 395-419.

The second way of shifting this balance is regulating another class of pro-apoptotic Bcl-2 proteins (the
BH3 sub-family) at the post-translational level. One example is Bad. So, Bad only has a BH3 domain.
Now, if you remember, the pocket of Bcl-2 proteins bind BH3 domains, right? So, Bad can dimerize with
Bcl-2 via its BH3 domain. What does this mean? If Bcl-2 is bound by Bad, it cannot bind Bax, right?
Therefore, this leads to the formation of Bax-Bax homodimers, etc., etc., etc. So, how does the cell
regulate Bad? Via a common post-translational modification: phosphorylation. A pro-survival protein
called Akt (sometimes called PKB) is activated by PI3-K. Akt then phosphorylates Bad. When Bad is
phosphorylated, it is now bound by a protein called 14-3-3 that sequesters Bad in the cytoplasm.
Therefore, Bad is unable to bind to Bcl-2. How is Bad released to bind to Bcl-2? Well, first it has to be de-
phosphorylated, and it was recently demonstrated that calcineurin (a calcium-dependent phosphatase)
can de-phosphorylate Bad, causing it to dissociate from 14-3-3, bind to Bcl-2, and induce apoptosis.
Pretty cool mechanism!

A second example involving post-translational modification is another BH3 sub-family member: Bid.
So, another way to induce apoptosis is via death receptors (pretty ominous name!), like Fas. So, the
activation of Fas by Fas ligand (FasL) leads to the activation of a caspase at the plasma membrane
(which is very different from the other caspases that I will talk about on my other page) called caspase-8.
Caspase-8 cleaves the inactive form of Bid into two parts. The one part that contains the BH3 domain is
now active! Now, active Bid translocates to the mitochondria and induces apoptosis. How? Well, a
recent paper in the Journal of Cell Biology (Desagher et al. 1999. JCB 144: 891-901.) has demonstrated
that when Bid binds to Bax (remember, Bax also has the pocket to bind BH3-containing proteins), it
causes a conformational change in Bax that leads to Cyt C release from the mitochondria. Another
really cool mechanism!

Okay, I hope you enjoyed my little discussion on Bcl-2 family proteins. It is really a cool area, and very interesting! Please
read my page discussing the caspase family, too. That way you can get the full effect of how apoptosis is regulated in
mammalian cells!
 Caspases
Now you have a basic background in apoptosis and Bcl-2 family proteins (if you have not yet read my
Bcl-2 page, you may want to go there now before reading this one). Now I will continue my discussion
with the caspase family of proteins. These are the mammalian homologues of the C. elegans protein
ced-3. Basically, caspases are cysteine proteases (have a critical cysteine residue in their active site...
mutate this cysteine, and you loose activity!) that cleave after aspartic acid residues. So, these are very
specific proteases.

Caspases can be divided into two main classes: initiator and effector caspases. Initiator caspases (like
caspase-9) are the upstream activators of the effector caspases (like caspase-3). Effector caspases are the
executioners in the cell, they cleave the proteins that actually induce apoptosis in the cell. These
cleavages lead to the morphological features that I mentioned earlier: membrane blebbing, cytoplasmic
and nuclear condensation, DNA fragmentation, and the formation of apoptotic bodies.

Raff, M. (1998). Nature 396: 119.

So, if these are such killer proteins, they must be tightly regulated, right? Definitely! But, on the other
hand, they need to be ready for quick action, so they are not regulated at the transcriptional level.
Rather, they are regulated at the post-translational level. That's right, the proteins are sitting in the
cytoplasm ready to be activated! Pretty scary, eh? How are the regulated? Well, they initially exist as
immature pro-caspases. In other words, they need to be processed to be activated. In this case, they need
to be cleaved. There are three basic domains in the immature form: the pro-domain, the large subunit
(p20), and the small subunit (p10). The initiator caspases have a large pro-domain, since these are
regulated by proteins other than caspases. The effector caspases have small pro-domains since they are
directly regulated by other caspases. In other words, the pro-domain is important for protein-protein
interactions.

Indeed, the large pro-domain interacts with other proteins in the cell, containing CARD domains. For
example, the initiator caspase-9 interacts with the CARD domain in Apaf-1 (mentioned in my Bcl-2
page). Or, the initiator caspase-8 interacts with similar domains in death receptors (like Fas). These
interactions lead to the cleavage and activation of the initiator caspases (mentioned below), which then
go on to cleave the effector caspases.

Thornberry, N.A. & Lazebnik, Y. (1998) Science 281: 1312-1316.

 What happens when the pro-caspases are cleaved? Well, the large subunit and the small subunit come
together in a heterodimer. Now, this is another time where crystallography has come in handy. It was
found that two heterodimers come together to form a tetramer! Above, you can see the family of
caspases in (A), the crystal structure of an active caspase tetramer in (B), and a schema tic of how the
pro-caspase is cleaved and comes together to form the active tetramer in (C).

So, what is the basic pathway of caspase activation? Let me tell you a little story about a researcher
named Xiaodong Wang at the University of Texas Southwestern Medical Center. Why? His lab
amazingly deciphered many of the important proteins and steps involved in the activation of the
caspase cascade in mammalian cells using hard core biochemistry. First, his lab set up a cell-free
apoptosis system. They found that if you extracted cytosolic extracts from normal HeLa cells and added
dATP, you could induce apoptosis (as measured by the cleavage of the effector caspase-3). So, they then
when on to purify the necessary co-factors in the HeLa cytosolic extracts (S-100 fraction) to produce 3
Cell papers.

In the first paper (Liu et al. 1996. Cell 86: 147-157.), they found they could divide the S-100 fraction into
two fractions, each containing a necessary co-factor they called Apaf-1 and Apaf-2. In this paper, they
went on to purify and identify (from 100 liters of HeLa cell extracts!) Apaf-2: cytochrome C. Indeed, as
mentioned in my Bcl-2 page, Cyt C is released from the mitochondria and is necessary to induce
apoptosis (discovered first by Wang's lab).

In the second Cell paper (Zou et al. 1997. Cell 90: 405-413.), they realized they could split the Apaf-1
fraction into two more fractions, each containing a necessary co-factor. They called these Apaf-1 and
Apaf-3 (since they already called Cyt C Apaf-2). They then went on to purify Apaf-1 (again, from 100
liters of HeLa cell extracts!) , and discovered a novel protein they simply called Apaf-1. Interestingly,
this appeared to be the ced-4 homologue that researchers have been trying to identify for a while. They
further showed that Apaf-1 can bind Cyt C and this is necessary to induce apoptosis in their cell-free
system.

In the final Cell paper (Li et al. 1997. Cell 91: 479-489.), they went on to purify Apaf-3 (yes, they used
100 liters of HeLa cell extracts again!). They found this to be the previously identified initiator caspase-9.
Now, if they really had identified all the necessary factors to induce apoptosis, they should be able to
reconstitute the cleavage of caspase-3 with purified Cyt C, Apaf-1, caspase-9, and dATP. And, indeed
they did! Wow! Now, that is some really hard core biochemistry! But, you have to have a lot of respect
for this lab. They have really done some amazing research! Oh, by the way, remember my little
discussion of Bid in my Bcl-2 page? That was also discovered by Wang's lab in the same way (Luo et al.
1998. Cell 94: 481-490.)!

Thornberry, N.A. & Lazebnik, Y. (1998) Science 281: 1312-1316.

So, with this background behind us, what is the pathway leading to the activation of the caspase
cascade? Okay, as I mentioned in my Bcl-2 page, the ratio of pro-apoptotic to pro-survival Bcl-2 proteins
regulates the release of Cyt C from the mitochondria. Cyt C (and ATP or dATP) binds to Apaf-1, leading
to its activation. Apaf-1 can then oligomerize, which by binding to the pro-domains of pro-caspase-9,
can bring pro-caspase-9 proteins in close proximity to one another. In this situat ion, pro-caspase-9 can
then cleave each other, leading to the formation of the mature caspase-9 (tetramer with two large
subunits and two small subunits). The mature caspase-9 can then cleave and activate effector caspases
(like caspase-3), which can then go on to cleave and activate other effector caspases or other cellular
substrates. Pretty complex mechanism, eh?

Slee et al. (1999) Journal of Cell Biology 144: 281-291.

So, what about caspases activating other caspases? Is this a complex cascade? You bet it is! The above
figure is from a recent article in the Journal of Cell Biology showing the complexity of this cascade. There
are initiator caspases (caspase-9) that cleave and activate effector caspases (pro-caspase-3 and pro-
caspase-7). Caspase-3 can then go on to cleave and activate other effector caspases (pro-caspase-6 and
pro-caspase-2). Caspase-6 can then go on to cleave and activate even more effector caspases (pro-
caspase-8 and pro-caspase-10). All of these effector caspases, in addition to cleaving and activating other
effector caspases, can cleave other cellular substrates (see below) that lead to the morphological features
associated with apoptosis. And... if you look back at the above figure, you can see that caspase-3, when
activated, can even feed back and cleave the initiator caspase-9! What happens when this occurs? Well,
it's a positive feedback loop, so it even further activates caspase-9 (which would then go on to cleave
and activate even more pro-caspase-3 and pro-caspase-7 proteins!). This insures that the caspase cascade
leads to irreversible cell suicide.

Thornberry, N.A. & Lazebnik, Y. (1998) Science 281: 1312-1316.

 Well, what about these other cellular substrates that I mentioned? The above figure gives three nice
examples. In the first example, the effector caspases cleave an inhibitor or an effector protein. An
example of this would be CAD (Caspase-Activated Deoxyribonuclease), and ICAD (Inhibitor of CAD).
When ICAD binds to CAD, CAD is kept inactivated. However, active effector caspases cleave ICAD
which then releases CAD. CAD can then cleave the DNA into fragments (forming the characteristic
DNA laddering of apoptotic cells). The second example illustrates that the effector caspases can also
cleave structural proteins, such as the nuclear lamins. Nuclear lamins maintain the integrity of the
nucleus, but when they are cleaved by the effector caspases, the nucleus condenses (another
characteristic of apoptotic cells). Finally, in the third example, effector caspases can cleave off the auto-
inhibitory domains of certain proteins. A good example of this would be PAK2 (see my PAK page!).
When the effector caspase cleaves off the auto-inhibitory domain of PAK2, PAK2 now becomes
constitutively active, playing a role in the membrane blebbing that is characteristic of apoptotic cells.