TGF vs BMP structure of smadi MHIDNA complex reveals distinctive rearrangements of a helix1

MH1 – DNA binding domain MH2 - Protein binding domain and transactivation domain 18 1.6 Multiple sequence alignment of MH1 domains of Smad1, Smad2, Smad3, Smad4, Smad5, Smad8, Smad6 and

CIS-REGULATORY ELEMENT SPECIFICITY OF TGF-β AND BMP TRANSCRIPTION FACTORS: STRUCTURE OF SMAD1 MH1/DNA COMPLEX REVEALS DISTINCTIVE

CIS-REGULATORY ELEMENT SPECIFICITY OF TGF-β AND BMP TRANSCRIPTION FACTORS: STRUCTURE OF SMAD1 MH1/DNA COMPLEX REVEALS DISTINCTIVE

DEPARTMENT OF BIOLOGICAL SCIENCES

NATIONAL UNIVERSITY OF SINGAPORE

2010

I would like to thank my supervisor Dr Prasanna R Kolatkar for giving me the

opportunity to work in his lab and for his support, open-mindedness and

encouragement throughout the project I would like to thank my co-supervisor Dr Kini

R Manjunatha for his critical comments and discussions.

I would like to acknowledge NUS and DBS for the graduate research scholarship;

A-Star and GIS for funding this project and for supporting me with financial assistance

till now

I would like to convey my heartfelt gratitude to Dr Ralf Jauch for his guidance and

help throughout this project I am grateful for his encouragement, suggestions and critical comments and also for being a huge inspiration I would also like to thank him for providing me an opportunity in organizing a crystallography workshop in Thailand

I am also thankful for the guidance and kind support of Dr Paaventhan Palasingam

and also for his help in the structure determination

I would like to sincerely acknowledge Dr Robert Robinson for his assistance during

the synchrotron visit to Taiwan and also for supporting me during the crystallography

workshop in Thailand I would like to thank Albert, Martin and Raj for their

assistance with the various lab equipments in IMCB I would also like to acknowledge

Dr Shyam Prabakar and Sun Wenjie for providing us with the co-operativity

formulae

I would like to convey my deepest gratitude to my close friends/colleagues in my lab

especially; Kamesh for his invaluable help and for being a pillar of support throughout

and Calista and Siew Hua for all the fun and support both in and out of the work place.

I would also like to thank all my amazing lab-mates Marie, Saravanan, Pugalenthi,

Leong Yu and Pan Hong for making the lab a very cheerful and energetic place to work in I would also like to thank my support system during my QE; Karthik,

Ayshwarya, Vishnu, Sravanthy, Manesh, Sheela, Gauri and Siva.

I am deeply grateful to aunt Chandrika and family for all the support, affection and pampering during my stay I would like to thank Parveen Prem for supporting me all

these years and for setting the wheel in motion for my PhD I would like to thank my

brother and his family for always taking care of me I am deeply indebted to my mother for her love and affection and for being my real-life role model I would love to

express my deepest gratitude to my father for being my mentor, always being there to

support my decisions against tsunamis of opposition and for his unparalleled love and affection

Specificity of receptor – ligand interaction

99

12

12

Smad3BMP SmadsSmad1Smad5Smad8Common Smad (Smad4)Inhibitory Smads

Smad6 Smad7Aim and Scope of the Project

16

1622

2223282830303133333334

2.4 Electrophoretic Mobility Shift Assay (EMSA) 402.5 Derivation of co-operativity constant for a protein

homodimer binding to a palindromic DNA sequence

41

2.6 Dynamic Light Scattering (DLS) 42

2.10 Crystallization of Smad1 MH1 with DNA 44

CHAPTER 3 RESULTS & DISCUSSION I

Smad1 MH1: Expression to Crystal Diffraction

46

3.1 Protein expression and protein/DNA complex assembly 47

CHAPTER 4 RESULTS & DISCUSSION II

Smad1 MH1: Structure-Function Relationship

60

4.3 Smad1 exhibits ‘open’ domain-swapped conformation

and is thermodynamically destabilized

5.3 How do Smads discriminate between TGF-β and BMP

6.1

6.2

ConclusionFuture directions

REFERENCES

8891

APPENDIX B Active percentage estimation of recombinant Smad1 and

Smad3 MH1 proteins by plotting protein to DNA ratio

vs fraction bound

116

APPENDIX C EMSA of Smad1 MH1 and Smad3 MH1 titrated with

1nM SBE DNA element in triplicates and a representative gel showing boxes used for quantification

117

APPENDIX D The coordinates for the Smad1 MH1/SBE DNA

complex structure is deposited in PDB id: 3KMP

118

Smad1 is a downstream effector of the Bone Morphogenetic Pathway signaling pathway that binds regulatory DNA to execute gene expression programs leading to, for example, the maintenance of pluripotency in mice On the contrary, the Transforming Growth Factor- activated Smad3 triggers strikingly different programs such as mesodermal differentiation in early development Because Smad1 and Smad3 contain identical amino acids at the DNA contact interface it is unclear how they elicit distinctive bioactivities Here we report the crystal structure of the MH1 domain of Smad1 bound to a palindromic Smad binding element Surprisingly, the DNA contact interface of Smad1 is drastically rearranged when compared to Smad3 The N-terminal helix 1 of Smad1 is dislodged from its intramolecular binding site and adopts a domain swapped arrangement with a symmetry related molecule As a consequence, helix 2 kinks away from the double-helix disabling several key phosphate backbone interactions Thermal melting analysis corroborates a decompacted conformation of Smad1 and DNA binding assays indicate a lower overall affinity of Smad1 to DNA but increased cooperativity when binding to palindromic DNA motifs These findings suggest that Smad1 and Smad3 evolved differential qualities to assemble on composite DNA elements and to engage in co-factor interactions by remodeling their N-termini

LIST OF FIGURES

1.1 Overview of the TGF-β superfamily of signaling The cytokine binds

to the type I and type II receptors resulting in phosphorylation of

R-Smads and in the formation of active heterotrimers of R-Smad/Smad4

complexes The active heterotrimers enters the nucleus to activate gene

expression

5

1.2 Cytokines with sequence identity ranging from 25% to 45% show

overall structural similarity

(A) TGF cytokine structure (PDB id: 1KLA) (B) BMP6 cytokine structure (PDB id: 2R52) (C) GDF-5 cytokine structure (PDB id: 2BHK)

11

1.3 The cytokine-receptor complexes reveal the complex protein-protein

interactions leading to the specific signaling pathway

(A) Crystal structure of GDF-5/BMPR-I complex (PDB id: 3EVS) shown in side view and

(B) Top view of GDF-5/BMPR-I complex (PDB id: 3EVS)(C) Crystal structure of BMP-2/BMPR-I/Acvr2 complex (PDB id:

2H64) shown in side view (D) Top view of BMP-2/BMPR-I/Acvr2 complex (PDB id: 2H64)(E) Crystal structure of TGF-β3/TGFBR-I/TGFBR-2 complex (PDB id: 2PJY) shown in side view and

(F) top view of TGF-β3/TGFBR-I/TGFBR-II complex

(C) BMP-RI and TGFBR-I (PDB id: 2H64 and 2PJY) superimposed using Pymol showing similar fold but difference

in interaction regions (D) Acvr2 and TGFBR-II (PDB id: 2H64 and 2PJY) superimposed using Pymol showing similar fold but difference in interaction regions

15

1.5 Pictorial representation of the structural arrangement of domains in the

Smad family of transcription factors MH1 – DNA binding domain

MH2 - Protein binding domain and transactivation domain

18

1.6 Multiple sequence alignment of MH1 domains of Smad1, Smad2,

Smad3, Smad4, Smad5, Smad8, Smad6 and Smad7 (mouse)

18

1.7 Rooted phylogenetic tree showing the different subgroups based on the

MH1 domains of Smad1, Smad2, Smad3, Smad4, Smad5, Smad8,

Smad6 and Smad7 (mouse) The BMP specific R-Smads (Smad1,

Smad5 and Smad8) and TGF-ß R-Smads (Smad2 and Smad3) cluster

together

19

1.8 Multiple sequence alignment of MH2 domains of Smad1, Smad2,

Smad3, Smad4, Smad5, Smad8, Smad6 and Smad7 (mouse)

20

1.9 Unrooted phylogenetic tree showing the different subgroups based on

the MH2 domains of Smad1, Smad2, Smad3, Smad4, Smad5, Smad8,

Smad6 and Smad7 (mouse) The BMP specific R-Smads (Smad1,

Smad5 and Smad8) and TGF-ß R-Smads (Smad2 and Smad3) cluster

together

21

1.10 (A) Smad2 MH2 domain of containing three helix bundle (cyan)

and β-sandwich (orange) (PDB id: 1KHX) (B) Two monomers of Smad2 MH2 (cyan) in complex with one

Smad4 MH2 (yellow & red) forming a heterotrimer (PDB id:

1U7V)

25

1.11 Structure of MH1 domain of Smad3 bound to Smad Binding

Element (PDB id: 10ZJ)

(A)Overall structure of two Smad3 MH1 monomers bound to

the palindromic SBE DNA element

(B) Overall structure rotated through the X-axis by 90˚

26

1.12 (A)Smad3 MH2 domain of containing three helix bundle (pink)

and β-sandwich (purple) (PDB id: 1MJS)

(B) Two monomers of Smad3 MH2 (pink) in complex with one

Smad4 MH2 (yellow) forming a heterotrimer (PDB id: 1U7F)

27

Chapter 3

3.1 (A)Expression profile of Smad1 MH1 domain

(B) Elution profile of Smad1 MH1 (blue) run on a Superdex 75

gel-filtration column calibrated with molecular weight standards

(grey) Smad1 MH1 elutes as a single symmetric peak consistent

with a molecular weight of ~16 kDa

(C) 12% SDS-PAGE analysis showing molecular-weight markers

(lane 1; kDa) and a pure Smad1 MH1 protein band (lane 2)

(D)The observed molecular weight of the pure Smad1 MH1

protein from the mass spectrometer

48

3.2 (A)Expression profile of Smad3 MH1 domain

(B) Elution profile of Smad3 MH1 (blue) run on a Superdex 75

gel-filtration column calibrated with molecular weight standards

(grey) Smad3 MH1 elutes as a single symmetric peak consistent

with a molecular weight of ~17 kDa

(C) 12% SDS-PAGE analysis showing molecular-weight markers

(lane 1; kDa) and a pure Smad3 MH1 protein band (lane 2)

(D)The observed molecular weight of the pure Smad3 MH1

protein from the mass spectrometer

49

3.3 (A)Elution profile of Smad1 MH1 (solid blue) and Smad1 MH1

with SBE (solid red) run on a Superdex 75 gel-filtration column

calibrated with molecular weight standards (grey)

(B) Elution profile of Smad3 MH1 (broken blue) and Smad3 MH1

with SBE (broken red) run on a Superdex 75 gel-filtration column

calibrated with molecular weight standards (grey) Both Smad1

MH1 and Smad3 MH1 elutes as a single symmetric peak

consistent with a molecular weight of ~17 kDa Smad1 and Smad3

MH1 domain in complex with SBE element also elutes as a single

symmetric peak corresponding to a molecular weight of two

monomers on the DNA (~ 45 kDa)

50

3.4 100 nM Smad1 MH1 domain was bound to 1 nM50-Cy5-labelled

15-mer SBE DNA (lane 1) The bound protein–DNA complex was

incubated with 2 mM unlabelled competitor DNA: 14-mer blunt

SBE, 15-mer blunt SBE (positive control for complete

competition), 16-mer blunt SBE, 17-mer TTAA overhang SBE and

54

16-mer SBE with spacer (lanes 2–6) An oligonucleotide with the mutated SBE sequence TCATCTGATTTATACT was used as a negative control (lane 7).

3.5 Smad1 MH1/SBE crystals obtained with

(A)A-T overhang DNA and (B) G-C overhang DNA grown at 18˚C (C) Microcrystals grown at 4˚C

(D)Rod-like crystals grown at 18˚C that diffracted to 2.7 Å resolution

55

3.6 Room temperature diffraction of crystals

(A)Crystal grown in 3% Propanol (additive) at 18˚C and (B) Crystal grown in 3% Propanol (additive) at 18˚C Crystals typically showed a diffraction of ~7-8 Å resolution

(C) Crystals grown in 3% Propanol (additive) at 15˚C and(D)Crystals grown at 18˚C in the presence of 3% Propanol as additive

(D)After annealing for 10 secs

Representative image of the 2.7 Å dataset of Smad1 MH1/ SBE crystal

Chapter 4

57

59

4.1 (A)Multiple sequence alignment of the MH1 domains of mouse

Smads prepared using T_coffee and shaded with boxshade (http://www.ch.embnet.org/software/BOX_form.html) Secondary structure elements as seen in the Smad1 structure are indicated above the alignments and stars mark pairing β-sheets The TGF-β/Smad specific α1/α2 insertion in Smad3 and Smad2 between helix α1 and helix α2 is marked with a red box The highly

65

4.3

4.4

conserved amino acid residues Arg74, Gln76 and Lys81 in the

β-hairpin (β2-β3) contacting specific DNA nucleotides are marked

by open circles

(B) The sequence of the 17 mer SBE DNA with TT-AA overhangs

used for crystallization The nucleotides which were not modeled

in the structure are not numbered while the other nucleotides are

numbered as submitted in the PDB The SBE palindrome

GTCTAGAC is boxed and shown in blue

Ramachandran plot of Smad1 MH1/SBE structure (3KMP) plotted

using Molprobity

(A)A stereo view of the overall structure of Smad1 MH1 with two

monomers of Smad1 MH1 shown as cartoon bound to palindromic

SBE DNA in stick representation α-helices are colored in blue,

β-sheets in red and loop regions in green with their respective labels

All structural figures were prepared using pymol The composite

omit map of the DNA is contoured at 1.0 σ

(B) Cross-section of the Smad1 MH1 with SBE DNA (Figure 4.3A

rotated by 90˚) revealing two glycerol molecules and two zinc

divalent ions

(C) Details of the zinc coordination site including Cys64, Cys109,

Cys121 and His126 amino acid residues The electron density

(2Fo-Fc) is displayed at 2σ

(D)The alternate rotamers of the His79 residue showing one of the

rotamers facing the glycerol and the other facing the DNA with

(2Fo-Fc) electron density contoured at 1.0 σ The other amino acids

Ser 78 and Lys 32 stabilizing the glycerol on either side are

contoured at 1.0 σ The electron density of glycerol molecule is

contoured at 0.7 σ

(A)Domain swap between adjacent symmetry related Smad1

MH1/SBE complexes coloured in green, yellow and blue The

Smad3 MH1/SBE complex (1OZJ) coloured in black is super

imposed with the blue Smad1 MH1/SBE The domain swap region

is boxed to highlight the difference in helix1 (α1) between Smad1

MH1 and Smad3 MH1

(B) The hinge loop between the helix1 α1 and helix2 (α2) of the

Smad1 structure coloured in green with the composite omit map

contoured at 1.0 σ is superimposed with the Smad3 MH1 (PDB_id:

10ZJ) in black The TGF-β specific α1/ α2-hinge consisting of

66

67

73

5.1

Gly21 at the sharp turn of the α1/ α2-hinge, Glu22 and Gln23

(Smad3 numbering) are labeled in blue and displayed as sticks

(A)Circular dichroism spectrum showing similar absorption

spectrum of Smad1 MH1/SBE (black) and Smad3 MH1/SBE (red)

complexes with peaks at 210 nm and 222 nm consistent with the

crystal structure

(B) Melting curves of Smad1 MH1/SBE in black and Smad3

MH1/SBE in red from 25˚C to 95˚C displaying a lower melting

point of Smad1 MH1/SBE (58˚C) and higher melting point of

Smad3 MH1/SBE (67˚C)

(C) Thermofluor melting of Smad 1 MH1 and Smad 3 MH1 in the

presence of SBE DNA is plotted as a function of Temperature (˚C)

vs –d (fluorescence)/dT Smad1 MH1/SBE (black) shows a lower

melting point compared to the Smad3 MH1/SBE (red)

Chapter 5

(A)10% native gel showing the binding of Smad1 MH1 to 2500

nM (10nMCy5 labelled and 2490nM 15mer unlabelled) of SBE

palindromic DNA (5’TGAGTCTAGACATAC3’) Protein

concentrations used were 0, 2.44, 4.88, 9.77, 19.53, 39.06, 78.13,

156.25, 312.5, 625, 1250, 2500, 5000, 10000 and 20000 nM (from

left to right)

(B) 10% native gel showing the binding of Smad3 MH1 to 2500

nM (10nMCy5 labelled and 2490nM 15mer unlabelled) of SBE

palindromic DNA (5’TGAGTCTAGACATAC3’) Protein

concentrations used were 0, 2.44, 4.88, 9.77, 19.53, 39.06, 78.13,

156.25, 312.5, 625, 1250, 2500, 5000, 10000 and 20000 nM (from

left to right)

(C) 10% native gel showing the binding of Smad1 MH1 to 1 nM of

SBE palindromic DNA (5’TGAGTCTAGACATAC3’) Protein

concentrations used were 0, 5, 10, 25, 50, 100, 250, 500, 1000 and

2000 nM respectively (from left to right) The titrations were done

in 1.5 ml polypropylene tubes

(D)10% native gel showing the binding of Smad1 MH1 to 1 nM of

SBE palindromic DNA (5’TGAGTCTAGACATAC3’) The

protein concentrations used were 0, 0.61, 1.22, 2.44, 4.88, 9.77,

5.3

5.4

5.5

(A)A stereo view of protein-DNA interaction of Smad1 MH1

domain revealing Arg74, Gln76 and Lys81 making specific

nucleotide specific interactions with the A9, G10, A11 and C12 of

the SBE DNA The water molecules, W5 and W7 tethering chain B

and chain A to the DNA are also shown here

(B) Schematic drawing of the SBE DNA with the specific

protein-DNA contacts (bold black), phosphate backbone interaction (blue

bold italics) and water mediated contacts (blue regular)

(A)Stereo view of the orientation of helix α2 of Smad1 MH1 in

blue superimposed with Smad3 MH1 in black to show the

displacement with respect to DNA and the unique amino acid

residues of Smad3 MH1 making phosphate contacts with the DNA

(B) Electrostatic rendering of the DNA binding surface of Smad1

MH1 domain showing the ‘open’ form of helix1

(C) Electrostatic rendering of the DNA binding surface of Smad3

MH1 domain showing the ‘closed’ form helix1

(A)Bar plot showing the mean affinities of Smad1 and Smad3

MH1 domains binding to the single Smad element (SB) calculated

from four replicates

(B) 10% native gel showing 1nM SB DNA element

(5’AGTATGTCTCAGATGA3’) incubated with increasing

concentrations of Smad1 protein Protein concentrations used were

0, 0.61, 1.22, 2.44, 4.88, 9.77, 19.53, 39.06, 78.13, 156.25, 312.5,

625, 1250, 2500 and 5000 nM (from left to right) Vectors of

fractions bound and corresponding protein concentrations were fit

to equation 3

(C) 10% native gel showing 1nM SB DNA element

(5’AGTATGTCTCAGATGA3’) incubated with increasing

concentrations of Smad3 protein Protein concentrations used were

0, 0.61, 1.22, 2.44, 4.88, 9.77, 19.53, 39.06, 78.13, 156.25, 312.5,

625, 1250, 2500 and 5000 nM (from left to right) The fraction

bound was plotted against the concentration of protein using

R-plot Vectors of fractions bound and corresponding protein

concentrations were fit to equation 3

(A)Cooperativity of Smad1 and Smad3 on the palindromic SBE

element 10% native gel showing the binding of Smad1 MH1 and

80

81

82

86

Smad3 MH1 to 1 nM of SBE palindromic DNA (5’TGAGTCTAGACATAC3’) The protein concentrations used were 0, 0.61, 1.22, 2.44, 4.88, 9.77, 19.53, 39.06, 78.13, 156.25, 312.5, 625, 1250, 2500 and 5000 nM (from left to right) Experiments were performed in triplicates.

(B) Box-plot representing the cooperativity factor (ω=kd1/kd2) for Smad1 and Smad3 binding to the SBE palindromic element calculated as described in the Materials and Methods for 5-6 independent measurements including lanes where the weakest band contributed a fraction bound of at least 10% The p-value was derived by performing a Welch two-sample t-test using R

(C) 10% native gel with 1nM GC-BRE (5’CGCCTGGCGCCAGAGA) incubated with increasing concentrations of Smad1 and Smad3 MH1 proteins Protein concentrations used were 0, 0.61, 1.22, 2.44, 4.88, 9.77, 19.53, 39.06, 78.13, 156.25, 312.5, 625, 1250, 2500 and 5000 nM (from left to right) Experiments were performed in triplicates

LIST OF TABLES

2.1 The oligonucleotide sequences used for cloning Smad1 and

Smad3 MH1 domains and the sequence of the forward strand of

5’ Cy5 labeled oligos used for EMSAs

4.1 Crystallographic data refinement statistics 64

4.2 Smad1 MH1 DNA base-pair step parameters with the negative

rolle angle in the middle of the palindrome highlighted in blue 68

4.3 Smad3 MH1 DNA (10ZJ) base-pair step parameters with the

negative rolle angle in the middle of the palindrome highlighted

in blue

68

4.4 Hydrodynamic properties of Smad-MH1 domains in the

4.5 Comparison of contacts between the ‘open’ and ‘closed’ forms

ActR - Adrenocorticotropic hormone Receptor

ACVR - Activin A Receptor

ADSC - Area Detector Systems Corporation

Ala - Alanine

ALK - Anaplastic Lymphoma receptor tyrosine Kinase

AMH - Anti-Mullerian Hormone

E10.5 - Embryonic day 10.5

EDTA - Ethylenediaminetetraacetic acid

EMSA - Electrophoretic Mobility Shift Assay

GDF - Growth/Differentiation Factor

MAP - Mitogen-Activated Protein

MBP - Mannose Binding Protein

MH1 – Mad Homolog1

MH2 - Mad Homolog 2

MIS - Mullerian Inhibiting Substance

NCS - Non-Crystallographic Symmetry

NES - Nuclear Export Signal

NLS - Nuclear Localization Signal

OD - Optical Density

PAGE - Polyacrylamide gel

PCR - Polymerase Chain Reaction

PCT - Pre-crystallization trial

PEG - Polyethylene glycol

SARA - Smad Anchor for Receptor Activation

SBE - Smad Binding Element

SDS- Sodium Dodecyl Sulphate

SELEX - Systematic Evolution of Ligands by Exponential Enrichment

CHAPTER 1

INTRODUCTION Cell signaling

Our world is of water, like the sea, But the molecules sparsely spread, Not independent, not touching, But somewhere in between 1

Communication between cells through cell signaling is essential for the coordination of cellular activities among cells and it is essential for the formation of tissues and organs (1) Tissue homeostasis and repair as well as immunity depend on the ability of cells to respond to the changes in the microenvironment and errors in this delicate system lead to diseases like cancer and autoimmunity Intercellular communication can occur through:

1 Direct contact or juxtacrine signaling

2 Over short distances or paracrine signaling

3 Over large distances or endocrine signaling

Cell-to-cell communication and cellular response to environmental stimuli are mediated by signaling pathways that relay the event of a ligand-receptor interaction at the cell surface by altering the gene expression in the nucleus Classes of proteins called receptors are responsible for receiving the information from the environment and relaying it into the cells The ligands that bind to and activate these receptors comprise cytokines, growth factors, hormones or neurotransmitters The activated ligand-receptor complex relays the signal by interacting with other proteins leading to a chain reaction resulting in a physiological outcome The entire set of cell changes induced by receptor activation is called a signal transduction mechanism or pathway The physiological outcome of a signaling pathway may depend on the strength of the signal received Regulation of signaling pathways can also be achieved by feedback mechanisms, signal amplification and cross-talk with other pathways A thorough understanding of the cell signaling is important for the effective treatment of many diseases, the development of drugs and growing artificial tissues/organs

The external stimuli in a signaling pathway are conveyed inside the cells by a chain of biochemical interactions between proteins and an appropriate downstream effect of gene regulation is achieved by the activity of transcription factors Even though the DNA sequence preference or DNA motif recognition of the transcription factors are known, the mechanism of specifically binding to the cis-regulatory region (located on the same molecule of DNA) of a particular gene resulting in a expression change is not completely understood Furthermore, transcription factors belonging to the same family recognize similar DNA motifs yet lead to pleiotropic effects In order to completely decipher the signaling mechanism and to unravel the mysteries of specific transcriptional regulation, the DNA recognition mechanism of the transcription factors must be studied in detail (2)

One classical example of a signaling pathway is the TGF-β superfamily of signaling, where the downstream pleiotropic effects are mediated by a family of transcription factors with similar DNA binding domain

1.1 An overview of the Transforming Growth Factor-β (TGF-β) superfamily of signaling

Transforming Growth Factor- β (TGF-β) superfamily of signaling regulates a wide range of processes such as homeostasis, migration, proliferation and differentiation of cells (3) Cytokines are involved in eliciting the downstream responses in TGF-β superfamily of signaling Binding of extracellular ligands belonging to the TGF-β super family of cytokines like TGF-β, BMP (Bone Morphogenetic Protein), activin and nodal to distinct sets of type I and type II receptors leads to receptor oligomerization (4,5) The type I and type II receptors have serine/threonine kinase activities and the type I receptor phosphorylates and activates

downstream transcription factors called Smads (homolog of Sma and Mothers Against Decapentaplegic proteins) (6) The Smad transcription factor complex enters the nucleus and activates/represses downstream genes leading to changes in phenotype The concentration of Smads that enter the nucleus and the DNA elements that they bind to are critical for the proper phenotypic outcome of the pathway Figure 1.1 shows the simplified overview of the TGF-β superfamily of signaling

The important components of the signaling pathway namely cytokines, type I & type II receptors and Smads will be described in detail with emphasis on the TGF-β and BMP specific pathway components

1.1.1 Cytokines involved in TGF-β superfamily of signaling

‘Cytokines are low-molecular-weight regulatory proteins or glycoproteins secreted by white blood cells and various other cells in the body in response to a number of stimuli’ (7) Cytokines have important attributes such as pleiotropy, redundancy, synergy, antagonism and cascade induction (7) The extracellular ligands that belong to the TGF-β super family of cytokines are TGF-β, BMP, activin, nodal, GDF (Growth and Differentiation Factors) and AMH (Anti-Mullerian Hormone) They are all characterized by a distinct cysteine knot scaffold structure with disulphide bonds between highly conserved cysteine residues (8) (9,10)

TGF-β R

Cytoplasm

Expression/Repression

Figure 1.1: Overview of the TGF-β superfamily of signaling (11) The cytokine binds to the type I and type

II receptors resulting in phosphorylation of Smads and in the formation of active heterotrimers of

R-Smad/Smad4 complexes The active heterotrimers enters the nucleus to activate gene expression

1.1.1.1 TGF-β cytokines

The TGF-β cytokines are 25 kDa disulphide linked homodimers and the TGF-β family

of cytokines includes: TGF-β1, 2, 3 4 and 5 The isomers have 70%-80% sequence identity TGF-β cytokines are secreted by platelets, macrophages, lymphocytes and mast cells The TGF-β cytokine is made up of four helices and seven β-strands forming an unusually extended conformation resembling a slightly curled left hand (Figure 1.2A) (8) The dimerized form of the TGF-β cytokine has nine disulphide bonds in total, out of which eight disulphide bonds are intra-molecular and one intermolecular disulphide bond The TGF-β cytokine is dimerized in a head to tail orientation through the interchain disulphide bond at the centre The extended form

of the monomeric form lacks a hydrophobic core typical of globular proteins But it is characterized by the presence of two hydrophobic regions at the ends of the molecule, one of which forms a hydrophobic core in the dimerized molecule (12,13) The interface between the monomers in the dimer form is largely hydrophobic The dimerized TGF-β cytokine bind to TβR-I and TβR-II receptors to elicit specific downstream response (Figure 1.3E and Figure 1.3F) The target cells for TGF-β include monocytes, activated macrophages, epithelial, endothelial, lymphoid, hematopoietic and proliferating B-cells

TGF-βs are involved in proliferation and differentiation in embryogenesis, wound healing, immune response, inhibition of mitogenicity and induction of extracellular matrix (14-24) TGF-B1-/- mouse embryos die due to infiltration of macrophages and lymphocytes into lungs and myocardial thinning and failure of coronary vessel development in heart (15,17,19,23,25-32) Germline mutations in TGF-β1 have been reported in the Camurati-Engelmann disease of the skeletal and the muscular system (33-36) Polymorphisms and decrease in the level of TGF-β1 and 2 are found in sporadic human diseases like

atherosclerosis, hypertension, cardiomyopathy, osteoporosis, cleft palate development, breast, colon and prostrate cancers (37-39) Increase in the level of TGF-β1 and 2 are also associated with restenosis, lung and pancreatic cancers (40)

1.1.1.2 BMP cytokines

The BMP cytokines include BMP-2, BMP-4, 5, 6, 7, 8 and BMP-9 The type I receptors that are known to bind BMPs are BMPR-IA, BMPR-IB and ActR-I The BMP cytokines also bind to type II receptors such as BMPR-II, ActR-II and ActR-IIB (Figure 1.3C and Figure 1.3D) The overall structure of BMP cytokines is similar to the TGF-β family of cytokines (Figure 1.2B) Comparison of the protein structures of the different BMP cytokines (BMP2 and BMP6) immediately shows that the loop regions differ widely between the cytokines which may lead to differences in binding to type I receptors (9)

The different BMP cytokines are involved in a multitude of cellular functions including induction of ventral mesoderm, induction of cartilage and bone and induction of apoptosis Some of the BMP cytokines are also believed to be involved in initiation of gastrulation and patterning of mesoderm, based on expression studies and genetic studies (41) Noggin and Sclerostin are known inhibitors of BMP ligands involved in regulating the BMP signaling pathway (42-49)

BMP-2 and BMP-4 are candidate genes in osteoporosis (50,51) Germ-line mutation in BMP-15 is involved in the premature ovarian failure in humans Alterations in expression levels in BMP-4 and BMP-7 are involved in the development of breast and colon cancers (52-58) Polymorphism in BMP-10 is involved in cardiomyopathy (59,60) BMPs are also implicated in iron metabolic disorders (40)

or indirectly to follistatins and cripto to regulate the activin mediated signaling (62-64)

Activins are involved in induction of dorsal mesoderm, induction of erythroid differentiation and induction of follicle-stimulating hormone release They also regulate many hormones including pituitary, gonadal and hypothalamic hormones as well as insulin and they also function as nerve cell survival factors (61,65-67) Inhibin acts as a potent inhibitor of activin signaling in a subset of activin responsive cells (eg Pituitary gonadotropes) (68-70)

Overexpression of activin in the mouse epidermis improves wound healing and inhibition improves the quality of wound healing (61) Activin rich growth medium can be used to maintain undifferentiated hESCs (human Embryonic Stem Cells) without feeder cells for over 20 passages (71) Activin has an anti-tumorigenic effect on certain tumour cells (from gall bladder, prostrate and pituitary gland) though increased levels of activin are also associated with the development of human endometrial adenocarcinoma (72-75)

1.1.1.4 Nodal cytokines

Nodals belong to the TGF-β superfamily of cytokines and the expression of Nodal cytokines is restricted to chordates Nodal binds to activin type I receptor, ActRIB (ALK4) and type II receptors, ActRIIA and ActRIIB (76,77)

Nodal signaling is involved in mesendoderm induction, neural patterning and left-right axis determination in embryos (77) Lefty and Cerberus act as antagonists to maintain the Nodal pathway (78-83)

Disruption of the nodal gene in mice leads to a failure to form the primitive streak during early embryogensis, a lack of axial mesoderm tissue and an overproduction of ectoderm and extraembryonic ectoderm (84,85) Germ-line mutations in Nodal is associated with the

developmental disorder called situs-ambiguus (40)

1.1.1.5 GDF cytokines

The GDF cytokines are part of the TGF-β superfamily of cytokines and named from GDF-1 to GDF-15 The structures of GDF cytokines have a similar fold as that of TGF-β and BMP cytokines (Figure 1.2C) The GDF cytokines bind to ACVR1B/ALK4, BMPR1B/ALK6 (type I receptors) (Figure 1.3A and Figure 1.3B) and ACVR2/ActRII, ACVR2B/ActRIIb, BMPR2/BMPRII (type II receptors) They are primarily involved in the development of the nervous system, mesoderm induction and embryonic development (86) Cripto acts as co-receptors in GDF signaling

GDF5 germ-line mutations are associated with diseases like Hunter –Thompson and Grebe-type chondrodysplasia, brachydactyly type C and A2 (skeletal and muscular system) in humans (40) Increase in the level of GDF-15 is involved in prostrate cancer (87)

1.1.1.6 AMH cytokines

Anti-Mullerian Hormone (AMH)/ Mullerian inhibiting substance (MIS) belonging to the TGF-β superfamily of cytokines is secreted by sertoli cells of the testes during

embryogenesis (88) The AMH acts through the AMH-II receptors ALK-2 acts a type I receptor for the AMH pathway (89) AMH is involved in the regulation of production of sex hormones and apoptosis of the fetal mullerian ducts (90,91) Mutations in the AMH gene causes persistant mullerian duct syndrome, a rare form of male pseudohermophroditism (88,91)

Figure 1.2: Cytokines with sequence identity ranging from 25% to 45% show overall structural similarity (A) TGF cytokine structure (PDB id: 1KLA) (8)

(B) BMP6 cytokine structure (PDB id: 2R52) (9)

1.1.2 Type I and Type II receptors in TGF-β signaling pathway

The TGF-β ligands bind to type I and type II receptors to elicit further downstream effects A receptor is a protein molecule, embedded in either the plasma membrane or the

cytoplasm of a cell, to which one or more specific kinds of signaling molecules bind leading to

a signaling activity The type I and type II receptors are structurally similar, with cysteine-rich extracellular regions and kinase domains in the intracellular regions (92) The type I receptors have a unique GS domain, rich in glycine and serine residues in the juxtamembrane domain and it is important for signal transduction in the signaling pathway (93) Each unique cytokine binds to a specific pair of type I and type II receptors to elicit a response downstream The TGF-β ligands/cytokines bind to type II receptors with constitutively active serine/threonine kinase receptor (94) Type I receptor can bind to the TGF-β ligands only in the presence of type II receptors The type II receptor, then, phosphorylates the GS domain of the type I receptors, thus activating it (93) The specificity of signaling is usually determined by the type

I receptor and the specificity of the ligand binding is controlled by a combination of both type I and type II receptor (95) The type I receptor, in turn, phosphorylates the MH2 domain of transcription factors called Smads (3,96) Genomic instability in TGFBR II (type II receptor) is involved in atherosclerosis and BMPR1A (type I) loss leads to valve malformation resulting in cardiovascular diseases (97-99) Type I and type II TGFBR are also implicated in gastric, colorectal, gastric and breast cancers (100-105)

1.1.2.1 Specificity of receptor – ligand interaction

The type I and type II receptors have similar overall structure with a three finger toxin fold (106) (Figure 1.4C and Figure 1.4D) The cytokines bind to distinct pairs of type I and

type II receptors thus leading to the hypothesis that the different cytokine-receptor complexes may differ structurally Analysis of the TGF-β3 /TGFBRI/TGFBRII complex and the BMP-2/BMPRI/AcvR2 complexes shows that though the overall folds of the type I and type II receptors binding to the cytokines are similar, the contacts between the ‘fingers’ of the receptors with the cytokine are different (Figure 1.3C, Figure 1.3D, Figure 1.3E and Figure 1.3F) Also, in the TGF-β3 complex, there are extensive contacts between the type I and type II receptors which explains the higher specificity of receptor recognition in TGF-β signaling (107,108) (Figure 1.4B) In the case of BMP complex, the contact between the type I and type

II receptors is minimal thus explaining the multiple binding partners of type I and typeII receptors with the BMP cytokine (109-113) (Figure 1.4A)

Figure 1.3: The cytokine-receptor complexes reveal the complex protein-protein interactions leading to the specific signaling pathway

(A) Crystal structure of GDF-5/BMPR-I complex (PDB id: 3EVS) (114) shown in side view and (B) Top view of GDF-5/BMPR-I complex (PDB id: 3EVS)

(C) Crystal structure of BMP-2/BMPR-I/Acvr2 complex (PDB id: 2H64) (109) shown in side view (D) Top view of BMP-2/BMPR-I/Acvr2 complex (PDB id: 2H64)

(E) Crystal structure of TGF-β3/TGFBRI/TGFBRII complex (PDB id: 2PJY) (107,115) shown in side view and

(F) top view of TGF-β3/TGFBRI/TGFBR2 complex

Figure 1.4: Specificity of signaling is determined by the specific cytokine-receptor complex interactions (A) Surface rendering of BMP-2/BMPR-I/Acvr2 complex (PDB id: 2H64) (109)

(B) Surface rendering of TGF-β3/TGFBRI/TGFBRII complex (PDB id: 2PJY) (107,115)

(C) BMPRI and TGFBRI (PDB id: 2H64 and 2PJY) superimposed using Pymol showing similar fold but difference in interaction regions

(D) Acvr2 and TGFBRII (PDB id: 2H64 and 2PJY) superimposed using Pymol showing similar fold but difference in interaction regions

1.1.3 The Smad Family of transcription factors

Smads (homolog of Sma and Mothers Against Decapentaplegic proteins) are transcription factors which act as effectors of the TGF-β super family of signaling pathway Smads are grouped into three different classes termed regulatory Smads (R-Smads: Smad 1, 2,

3, 5 & 8), a common Smad (Co-Smad: Smad 4) and inhibitory Smads (I-Smad: Smad 6 & 7) (116) The R-Smads and Smad4 possess two distinct globular domains, Mad homology 1 and 2 (MH1 and MH2) that are connected by a linker region of variable length The phosphorylation

of the MH2 domain of R-Smads leads to the formation of homotrimer of three Smads or heterotrimer consisting of two R-Smads and one Smad4 molecule (117) The heterotrimeric complex translocates to and accumulates in the nucleus and associates with specific chromatin regions in an R-Smad and cell-type specific manner (118-120)

1.1.3.1 Structural Features of Smads

In general Smads have three main features namely the MH1 domain, the MH2 domain, and the linker (Figure 1.5) The MH1 domain consists of approximately 130 amino acids and is highly conserved in R-Smads and Smad4, but not in I-Smads In the basal state, the MH1 domain inhibits the transcriptional and biological activities of the MH2 domain The inhibitory effect is likely due to the interaction between these two domains The MH1 domain does not have a purely inhibitory function because it has DNA-binding activity in the activated state The selective recognition and regulation of target genes is enabled by the DNA binding activity of the MH1 domain (121,122) The multiple sequence alignment of the MH1 domain

of Smads reveals that they are ~70% identical and they are most diverse in the N-terminal end Also the unrooted phylogenetic tree show that the Smad6 and Smad7 MH1 domains are the

most diverse when compared to the R-Smads and Smad4 MH1 domains (Figure 1.6 and Figure 1.7)

The MH2 domain contains receptor phosphorylation sites (in R-Smads), has effector function, and is involved in several important protein-protein interactions The MH2 domain is about 200 amino acids long and contains a characteristic insert in the case of Smad4 The phosphorylated MH2 domains interact to form homo-oligomeric complexes (117,120) The MH2 domain also mediates the association of R-Smads with type I receptors, with Smad4 upon receptor-mediated phosphorylation, and with DNA-binding factors Within the nucleus, the MH2 domain is involved in protein-protein interactions and facilitates transactivation (123) Smad4 contains the nuclear localization signal for the heterotrimer to enter the nucleus Various mutations leading to loss of function of Smads have been mapped to both the MH1 and MH2 domains in pancreatic and colorectal cancers (124) The multiple alignment of the MH2 domain of all the Smads shows a very high amino acid identity (~80%) (Figure1.8) The unrooted phylogentic tree show that the TGF-β specific R-Smads, BMP specific R-Smads and the I-Smads cluster together Smad4 is quite diverse from the other R-Smads probably due to the insert found in the MH2 domain (Figure 1.9)

The linker region in Smads is highly variable in both size and sequence This region contributes to the formation of Smad homo-oligomers In R-Smads, the linker region contains MAP-kinase phosphorylation sites (125) Increasing evidence now points to the function of linkers in binding specific co-factors (126)

Figure 1.5: Pictorial representation of the structural arrangement of domains in the Smad family of transcription factors MH1 – DNA binding domain MH2 - Protein binding domain and transactivation domain

Figure 1.6: Multiple sequence alignment of MH1 domains of Smad1, Smad2, Smad3, Smad4, Smad5, Smad8, Smad6 and Smad7 (mouse) The 30 amino acid insert in Smad2 is highlighted using a black box

Figure 1.7: Rooted phylogenetic tree showing the different subgroups based on the MH1 domains of Smad1, Smad2, Smad3, Smad4, Smad5, Smad8, Smad6 and Smad7 (mouse) The BMP specific R-Smads (Smad1, Smad5 and Smad8) and TGF-ß R-Smads (Smad2 and Smad3) cluster together