Advances in complex systems July 3-7, 2017

1. DYNAMICS OF
DEVELOPMENTAL FATE
DECISIONS
Luís A. Nunes Amaral
Northwestern University
Biological Physics Seminar

2. Good to get some historical perspective
2

3. From ‘Lords of the fly’
3
‘… experimentalists trust and value most highly results
that they can put to use productively in their own
experimental work.’

4. A bit of history
4
Thomas Hunt Morgan (September 25, 1866 – December 4,
1945), an American evolutionary biologist, geneticist,
embryologist, won the Nobel Prize in Physiology or Medicine in
1933 for discoveries elucidating the role that the chromosome
plays in heredity.
Morgan began to study the genetic characteristics of the fruit
fly Drosophila melanogaster and demonstrated that genes are
carried on chromosomes and are the mechanical basis of
heredity. These discoveries formed the basis of the modern
science of genetics. As a result of his work, Drosophila became
a major model organism in contemporary genetics.

5. From ‘Lords of the fly’
5
‘Morgan’s discovery of the autocatalytic property of
large-scale breeding – the breeder reactor – was a
turning point in Drosophila’s natural history.’

6. The wonder of development

7. An egg is the organic vessel
containing the zygote in which an
animal embryo develops until it can
survive on its own.

8. From the egg
8
Human fertilized egg
0.1 mm
0.5 mm
20 mm

9. To a multicellular organism
9
Spider mite: 0.4 mm
Human: 1,700 mm Chicken: 200 mm
Fruit fly: 3.5 mm

10. Formalizing development process

11. Waddington’s epigenetic landscape
11

12. Hematopoietic stem cells
12

13. Caenorhabditis elegans
13
https://www.youtube.com/watch?v=M2ApXHhYbaw

14. C. elegans
14

15. Mechanisms for development
15

16. Conservation of important genes
16

17. How do these mechanisms work?

18. Cell specialization
18

19. Cell communication
19

20. Inductive signaling
20

21. Signaling pathways are highly conserved
21

22. Morphogenic gradients
22

23. Patterning
23

24. Patterning in Drosophila
24

25. Drosophila melanogaster
25

26. Patterning of the complex eye

27. 27 https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3427658/
Studies of […] eye mutants have provided key insights
into the areas of cell fate specification, lateral inhibition,
signal transduction, transcription factor networks, planar
cell polarity, cell proliferation and programmed cell death
just to name a few.

28. The structure of the complex eye
28
Each complex eye comprises 475-490 ommatidia.
The core of the adult ommatidium contains 8 photoreceptor neurons, 4 lens secreting
cone cells and 2 primary pigment cells.
Each ommatidium shares 6 secondary pigment cells, 3 tertiary pigment cells and 3
mechano-sensory bristle complexes with its surrounding neighbors.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3427658/

29. Ommatidia formation
29 https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3427658/
All cells that are not
incorporated into the pre-
clusters undergo a single
additional round of mitosis.
This second mitotic wave is
required to generate the
remaining cells needed to
produce mature ommatidia

30. Ommatidia formation
30

31. R8 recruitment
31 http://dev.biologists.org/content/137/14/2265.full

32. Recruitment of other photo-receptors
32

33. Recruitment of other photo-receptors
33 http://dev.biologists.org/content/137/14/2265.full
Bistable expression of Yan and Pointed
regulates the transit to differentiation
Concentration
Time
Pointed (Activator)
EGFR Signaling
Yan Pointed
Yan
(Repressor)
Multipotent State Differentiating State
Genes
Differentiation
Yan
Genes
Differentiation
Pnt
7
Remake plot

34. Recruitment of other photo-receptors
34
0.5
1.0
1.5
2.0
anConcentration(AU)
Concentration
Time
0.5
1.0
1.5
2 .0
0.5
1.0
1.5
2 .0
− 2 0
0.0
PntC
PntConcentration(A.U.)
PntConcentration(A.U.)
D E
Time (h)
-10 0 10 2 0 3 0 40 5 0
Multipotent Cells
R2/R5 Neurons
tedConcentration(AU)
R3/R4 Neur
Concentration
Pointed
Time
Multipotent
Yan
Differentiated
Yan
Pnt

35. How where these results obtained?

36. Null Mutants!
Staining using antibodies

37. Summary of previous lecture
37

38. Summary of previous lecture
38

39. Summary of previous lecture
39
0.5
1.0
1.5
2.0
YanConcentration(AU)
Concentration
Time
0.5
1.0
1.5
2 .0
PntConcentration(A.U.)
D
Time (h)
-10 0 10 2 0 3 0 40 5 0
Multipotent Cells
R2/R5 Neurons
R3/R4
Concentration
Pointed
Time
Multipotent
Yan
Differentiated
Yan
Pnt

40. The human team
40
Sebastian Bernasek
Chem. Engineer
Neda Bagheri
Electr. Engineer
Rich Carthew
Biologist
Justin Cassidi
Biologist
Nicolas Pelaez
Biologist
Ilaria Rebay
Biologist

41. Dynamics of cell fate determinants
41 Peláez et al., eLife 4, e08924 (2015)

42. The first challenge
42 Qi et al., Int Conf Signal Process Proc. 2013, 670-674 (2014)

43. Performance of segmentation algorithm
43 Qi et al., Int Conf Signal Process Proc. 2013, 670-674 (2014)
Original image Gold standard
Our segmentation Comparison

44. Our images are more challenging
44 Peláez et al., eLife 4, e08924 (2015)

45. Flies that express Histone-RFP and Yan-YFP
45 Peláez et al., eLife 4, e08924 (2015)

46. So, what did we find?

47. This is the expectation!
47
0.5
1.0
1.5
2.0
YanConcentration(AU)
Concentration
Time
0.5
1.0
1.5
2 .0
PntConcentration(A.U.)
D
Time (h)
-10 0 10 2 0 3 0 40 5 0
Multipotent Cells
R2/R5 Neurons
R3/R4
Concentration
Pointed
Time
Multipotent
Yan
Differentiated
Yan
Pnt

48. This is the observed outcome for Yan
48
Approximately
exponential decay
Peláez et al., eLife 4, e08924 (2015)

49. This is the observed outcome for Pnt
49 Unpublished (2017)

50. Let’s look at Yan in more detail

51. Process is insensitive to absolute Yan levels
51

52. Rates of decay of Yan levels
52

53. Variability across cells highest at ramping up
53

54. What is going on between Yan and
Pnt?

55. Comparison of levels within same cells
55 Unpublished (2017)

56. We have not yet developed modeling
approach, but feel we now have the
data to attempt to validate models

57. Why this may be better
Yan and Pnt are very powerful transcription factors. Probably
not safe to have high levels of either for long!
System is re-usable: Whether particular fate is chosen or not,
levels of Yan and Pnt return to baseline.

58. Why are gene regulatory networks so large?
58 Unpublished (2017)

59. Because otherwise things breakdown…
59
miR-7 +miR-7 -
Unpublished (2017)

60. But what if we slow things down?
60
Glucose restriction
Unpublished (2017)

61. Lots of good things happen!
61 Unpublished (2017)

62. What if we slow things down by
reducing number of ribosomes?

63. Lots of good things happen!
63 Unpublished (2017)

64. Even when you remove a big system!
64 Unpublished (2017)

65. Modeling approach
65 Unpublished (2017)
D is activation level of gene in DNA
R is number of mRNAs available for translations
P is number level of proteins
Assumptions:
D takes continuous value between 0 and
some maximum
R and P are integer but large enough that can
be seen as continuous
Formation
Degradation

66. Feedback control
66 Unpublished (2017)

67. Redundant feedback control
67 Unpublished (2017)

68. Challenges of analysis of results
68 Unpublished (2017)
We are not modeling any particular system
System is almost certainly nonlinear
Change in metabolic rate will likely affect different parameters
differently
We cannot know values of any parameters
Parameters are effective anyway

69. Addressing challenges
69 Unpublished (2017)
Scan many possible parameter values (at least order an order
of magnitude change for each parameter value)
Consider linear and nonlinear version of equations
Consider different possibilities for how change in metabolic rate
will affect different parameters

70. Analysis of simulation results
70 Unpublished (2017)

71. Modeling approach
71 Unpublished (2017)

72. Redundancy allows for speed
Gene regulation network redundancy enables organisms to
increase metabolic rate while avoiding developmental
mistakes.
Shorter developmental times, increase fitness by reducing
time organism is helpless.
Provides driving force for increasing complexity.

73. PROMPTING IS CURRENTLY
THANK YOU!
NORTHWESTERN MEMORIAL