Dynamics of haematopoietic stem cell differentiation

In this practical worksheet, you will be asked a number of questions as you work your way through
the practical. These are really to keep you on track and check you understand what you do. The
answers to all the questions are at the bottom of this worksheet.

Introduction
Haematopoietic stem cells are the stem cells that provide all the circulating cells of the blood.
These include red blood cells- or erythrocytes, which lack a nucleus- and macrophages, as well as
most of the immune system (including B cells and T cells). The haematopoietic system is
incredibly active throughout our lifetimes, producing around 2 billion red blood cells per day.

These cells undergo stepwise differentiation. The system is a hierarchical one, meaning that cell
differentiation takes place progressively and that cells at each point of the differentiation pathway
have increasingly limited options for differentiation, as illustrated in Figure 1.

Figure 1. CLP, common lymphoid progenitor; CMP, common myeloid progenitor; EP, erythrocyte progenitor;
GMP, granulocyte–macrophage progenitor; GP, granulocyte progenitor; LMPP, lymphoid primed multipotent
progenitor; MacP, macrophage progenitor; MEP, megakaryocyte–erythrocyte progenitor; MkP,
megakaryocyte progenitor; MPP, multipotent progenitor; NK, natural killer. Taken from Cedar and Bergman
Nature Reviews Immunology 11, 478-488 (July 2011). [Feel free to read the review above if after the
practical you are interested in digging deeper about how the haematopoietic hierarchy works. However, it’s
well beyond the scope of this practical and the course as a whole.]

You have met some of these cell types before in your CTO2 lectures (L16- Cell differentiation 4-
Andrew Jarman), where you looked at the differentiation of the common myeloid precursor (CMP)
to erythrocytes and myeloid cells. Some questions in your practical assessment will ask you to
make connections between the material from this lecture and the practical you will do here.

Stem cells, by definition, are able to both proliferate indefinitely as stem cells, and also
differentiate. The haematopoietic stem cell (HSC) provides an excellent example of how a stem
cell system has to maintain itself throughout life. These cells must be able to both self-renew
(divide to give rise to more copies of themselves) and differentiate throughout the lifetime of the
individual. The cells they produce have a more limited capacity for cell division, and some of their
differentiated derivatives, like erythrocytes, have no ability to divide, not simply because they are
fully differentiated cells, but also because they lack a nucleus. For this reason, erythrocytes have
a finite lifespan because the proteins present are not replenished via transcription of genes in the
nucleus, and so they must be continuously replaced from precursors. In this practical, we will look
at the fact that the maintenance of the haematopoietic system can be described by a model that
has very straightforward rules governing the probability of each cell dividing, differentiating or
dying. We will also look at how the system is skewed by factors that alter the balance of cell
proportions. We will use a simulation that models how the numbers of cells changes with
successive cell divisions.

We are going to use a simplified hierarchy based on selected cells from Figure 1, as shown in
Figure 2. Notice here that any given cell at any point in the pathway is a potential control point to
set the numbers of cells in any one category.

Figure 2 A. The same hierarchy as in Figure 1, but with the cells we will study outlined with red boxes. B.
The simplified hierarchy we will study here.
Part 1. Using a simulation to model cell numbers
To access the simulation, navigate to http://ctoweb.mvm.ed.ac.uk/practical_cell_div/cell_div.html

You should see a screen with a number of parameters at the righthand side, which have been set
to default values, as in the screenshot in Figure 3:

Figure 3. Screenshot of default simulation parameters. Note that the righthand side of the figure is just the
same window, scrolled to the bottom to show all the default parameters.

This simulation models how the numbers of cells alter over several rounds of cell division (called
steps), given a set of simple rules (the Simulation Parameters). The graph on the right hand
side allows you to track the numbers of the different cell types over time as you alter the simulation
parameters. The box at the bottom gives a readout of some of the numerical values plotted on the
graphs.

As shown in Figure 3, for each step, or cell division, each of the cell types has:

a probability of dividing (p_div; in the default online version, this is set to 0.5, ie a 50% chance of
dividing), and
a probability of dying (p_die; set to zero in the default version).

Also, for the cells that undergo differentiation (HSC, CMP and CLP), there is a probability of them
either dividing to give rise to more of themselves (e.g. CMP to CMP+E or CMP+M), or to give rise
entirely to differentiated cells (e.g. CMP to E+E or M+M). These have been set to a probability of
0.5 in your initial version. Finally, remember that erythrocytes cannot divide, and so there is no
(p_div) for this cell type.

Now try running the simulation 5 times with the default setting (click ‘run simulation’) After 5 runs
of the simulation, answer the following question

Q1. What are the relative proportions of E, M, B and T?

a. roughly equal
b. quite different

Q2. Why are the numbers not the same each time?
a. The simulation is slightly faulty, and these faults get amplified with more cell
divisions.
b. The system relies on the probability of a cell undergoing division or differentiation.

Q3. What type of relationship between number of cells E, M, B, and T and number of
divisions best describes the graph?
a. Linear
b. Logarithmic

Q4. And what about HSC, CLP and CMP?

a. Linear
b. Logarithmic

Think about why the cells might behave in this way. In your practical assessment, you will be
asked for an explanation.

Now try adjusting the following parameters:

1. Number of steps (ie rounds of cell division), keeping all other parameters the same.
Q5. What is the effect of decreasing the number of steps on the proportions of E, M, B and T
cells?
a. Proportions become more homogeneous
b. Proportions become less homogeneous

Now reset the number of steps to 50 and adjust

2. Probability of cell type CLP differentiating to B
Q6. What is the effect of increasing the probability of cell type CLP differentiating to B cells
(try going stepwise with CLP->B+B probability of 0.1,0.2 etc)?
a. Number of B cells increases
b. Number of B cells decreases

Q7. What is the effect of the above alteration of parameters on the number of T cells?
a. Number of T cells increases
b. Number of T cells decreases

Q8. What is the effect on HSC, CMP, E and M cells?

a. No effect
b. A small effect

Now reset the probability of CLP->BB to 0 and adjust

3. probability of cell type CLP dying.
Q9. What is the effect of increasing the probability of death of CLP?
a. just decreases CLP numbers
b. Decreases numbers of CLP and its descendants

Now reset the probability of CLP dying to 0.

You’ve learnt how manipulating various parameters of the system can affect the numbers of cells
generated after 50 divisions.
Notice that up to now, the numbers of cells has continued to increase over the whole period of the
simulation.
Q10. Is this what would normally happen in an adult organism?
a. Yes
b. No

Have a play with the parameters to see what will allow the system to reach a steady state, ie
where the numbers of cells are not constantly increasing. Remember to run the simulation at least
5 times to be sure how reproducible the effect of your alteration is.

Hint and cautionary note: To do this, you will need to increase the number of divisions.
However, once the simulation generates too many cells, it freezes because it can’t handle too
many computations. If/when this happens, just logout and in again, and the default parameters
should reappear, the system will reset and you can carry on. Since this is the case, think about
how you can generate fewer cells yet eventually reach a steady state. Feel free to use the
discussion board to share amongst yourselves parameters that work, but also think about why they
work!

The adult haematopoietic system has many more erythrocytes than other cell types. This is
because the blood’s minute-by-minute function is to carry oxygen to the tissues, so we need
enough oxygen-carrying capacity to keep the whole organism alive. The immune functions of the
blood need to be variable- ie we need to be able to produce large numbers of certain types of
immune cells during infection, but in the uninfected state, there is no need to carry around large
numbers of B cells, T cells and macrophages in the circulation.

Your task is now to produce, in steady state, the following proportions of cells:

Erythrocytes 90%
Macrophages 1%
B cells 6%
T cells 3%

Note that these values are an oversimplification: erythrocytes represent 99.9% of circulating cells
in the blood; the other cells are much rarer. However, many macrophages, B cells and T cells are
not circulating in the bloodstream but are held in sites like the lymph nodes, thymus and other
tissues, and therefore the proportions have been raised to account for this.

Feel free to confer with your classmates to work on a solution to this challenge. Post your
suggested solutions (ie a set of parameters which can be reproduced by anyone) to the Learn
discussion board. Small prizes will be awarded to the student/students who arrive at the
parameters giving the closest values to the stated proportions and post them on the Learn
discussion board before 5pm on Monday 23rd November.

Note that there will be more than one way of solving this.

Part 2. Manipulating erythrocyte levels.

The 7-times Tour de France winner Lance Armstrong was recently stripped of all his winning titles
and banned from competitive cycling for doping (see Figure 4)
Figure 4. News headlines in New York Times, October 2012.

One of the banned substances that he later admitted to taking was erythropoietin, a diffusible
molecule that acts on CMPs to increase their probability of differentiating to erythrocytes, which
greatly increases the blood’s ability to carry oxygen to the muscles, and therefore increases
performance and allows better recovery after intense exercise. You can now model the effect of
EPO on haematopoiesis by resetting the parameters to the default values and increasing the
probability of CMP to CMP+E to 0.7, while decreasing the probability of CMP to CMP+M to 0.3.

Q11. What effect does this have on any other cell types?
a. none
b. also an effect on CLP and its derivatives

Q12. If, instead of altering the balance of CMP to E versus M, you increase the probability of
CMP to E+E, would this have the same effect as the above alteration?
a. Yes
b. No
In fact, the truth lies somewhere between these two scenarios. EPO not only alters the balance of
CMP to E versus M, but also increases the likelihood of CMP differentiating to E. Think about what
the simulation tells you would happen to anyone taking EPO over a long period.
(A note for your interest, and in case anyone is thinking of increasing their stamina this way:
although the simulation doesn’t model this, the lethal effect of too much EPO is that the numbers of
erythrocytes increase so much that the blood is too thick to get through the blood vessels and the
heart gives out)

Part 3. Modelling Acute Myeloid Leukemia

In the UK, around 8,600 people are diagnosed each year with leukaemia. Of those, just over 2,600
people have acute myeloid leukaemia (AML). In many cases, this disease is caused by a mutation
in a haematopoietic transcription factor, Runx1, which normally stimulates the transcription of
haematopoietic-specific genes. In AML patients, a chromosomal translocation fuses the DNA-
binding domain of Runx1 to a transcriptional repressor such as ETO. The net effect is that, in CMP
progenitors, the transcription of genes promoting differentiation is blocked, while cell division of
CMPs increases. You can partially model this by resetting the simulation to the default values
again, and then increasing the p_die of all cell types except the CMP to 0.1.

Q13. What is the effect on the CMPs relative to CLPs?

a. Proportions increase
b. Proportions decrease
Q14. what is the effect on E and M cells relative to T and B cells?
a. Proportions increase
b. Proportions decrease

You can see that this effect was achieved just by increasing the survival of CMPs relative to other
cell types, which is not exactly the same as the effect described above (increased proliferation of
CMPs and reduced differentiation). In the practical assessment, we will return to this issue.

End of the non-assessed part of the practical.

Answers to questions 1-14

Q1. What are the relative proportions of E, M, B and T?
a. roughly equal (correct)
b. quite different (wrong- check the parameters)
Q2. Why are the numbers not the same each time?
a. The simulation is slightly faulty, and these faults get amplified with more cell
divisions. (wrong- think about the text at the top)
b. The system relies on the probability of a cell undergoing division or differentiation.
(correct- In each case, any given cell is not certain to do the same thing)
Q3. What type of relationship between number of cells E, M, B, and T and number of
divisions best describes the graph?
a. Linear (wrong- not a straight line)
b. logarithmic (correct- a curve that increases in slope with linear increase in cell division)
Q4. And what about HSC, CLP and CMP?
a. Linear (correct- a straight line)
b. logarithmic (wrong- not a curve that increases in slope with linear increase in cell division)
Q5. What is the effect of decreasing the number of steps?
a. the proportions of E, M, B and T cells become more homogeneous (wrong- try
decreasing the steps to 10)
b. the proportions of E, M, B and T cells become less homogeneous (correct- this is
because the more times you toss a coin (with probability of 0.5 of landing heads or tails)
you have, the closer the outcomes will be to 50%)
Q6. What is the effect of increasing the probability of cell type CLP differentiating to B (try
going stepwise with CLP->BB probability of 0.1,0.2 etc)?
a. Number of cell type B increases (wrong, actually it decreases- think about why)
b. Number of cell type B decreases (correct- but why? You’ve increased the probability of
producing B, but decreased the relative probability of producing more CLP cells, which
reduces over time the number of CLP cells)
Q7. What is the effect of the above alteration of parameters on the number of T cells?
a. Number of T cells increases (wrong- check parameters)
b. Number of T cells decreases (correct- again, check the hierarchy and think about why?)

Q8. What is the effect on HSC, CMP, E and M cells?

a. No effect (correct- these lie upstream of the manipulation you performed, therefore
shouldn’t be affected)
b. A small effect (wrong- this is just the variation between trials you noticed earlier. Try
running the same simulation several times)
Q9. What is the effect of increasing the probability of death of CLP?
a. just decreases CLP numbers (wrong)
b. Decreases numbers of CLP and its descendants (correct)
Q10. Is this what would normally happen in an adult organism?
a. Yes (wrong- otherwise we’d just carry on increasing our differentiated cells and eventually
explode)
b. No (correct- we need to lose cells at the rate that they are produced- ie reach a steady
state.)
Q11. What effect does this have on any other cell types?
a. None (correct)
b. Also an effect on CLP and its derivatives (wrong- the proportions are altered, but the
absolute numbers of CLP and derivatives are unchanged)
Q12. If, rather than altering the balance of CMP to E versus M, you increase the probability
of CMP to E+E, would this have the same effect?
a. Yes (wrong)
b. No (correct- think about why? If you increase the probability of CMP differentiating to E,
you alter the body’s ability to produce more CMPs and macrophages, and consequently its
ability to fight infection)
Q13. What is the effect on the CMPs relative to CLPs?
a. CMP proportions increase (correct)
b. CMP proportions decrease (wrong- think about what happens when you increase the
probability of death of all cell types except CMP)
Q14. And what is the effect on E and M cells relative to T and B cells?
a. E and M proportions increase (correct- think about why this is predictable)
b. E and M proportions decrease (wrong)