University of Southern California Dissertations and Theses

The crystal structure of APOBEC-2 and implications for APOBEC enzymes

(USC Thesis Other)

The crystal structure of APOBEC-2 and implications for APOBEC enzymes

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content THE CRYSTAL STRUCTURE OF APOBEC-2

AND IMPLICATIONS FOR APOBEC ENZYMES

by

Courtney Prochnow

A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(MOLECULAR BIOLOGY)

May 2008

Copyright 2008 Courtney Prochnow

ii
Acknowledgements

I want to acknowledge Xiaojiang who has been a wonderful advisor and has
also become a good friend. Thank you, Xiaojiang, for fostering a positive work
environment in the lab that has been both supportive and fun. Thank you also for
providing consistent support and insight with matters in and out of the lab.
This work could not have been accomplished so quickly without the
contributions of Dr. Ronda Bransteitter. Ronda played an integral role in every
aspect of this project including performing the deamination experiments, designing
mutations, having insightful and motivating discussions and surfing. Thank you for
teaching me when to push hard for the “big bingo” and when to slow down, save my
energy and relax and especially have fun. Your enthusiasm for science is
contagious! Apo2 we love you!!
Dr. Myron Goodman, our lab neighbor and collaborator, provided a
stimulating collaboration.
Thank you to the entire Chen lab for your help, suggestions and insightful
discussions and for making my time as a graduate student such a wonderful and fun
experience. Thank you also for providing entertainment at any hour of the day or
night.
Thank you last, but not least, to my family who has supported me in all of my
professional decisions.

iii
Table of Contents

Acknowledgements ii
List of Tables v
List of Figures vi
Abstract viii

Chapter 1 Background 1
1. Activation-Induced Cytidine Deaminase (AID) 3
2. APOBEC-3 Enzymes (Apo3A, 3B, 3C, 3DE, 3F, 3G, 3H) 5
3. APOBEC-2 8
4. APOBEC-1 9
5. APOBEC-4 10
6. Cytidine Deamination 10
7. Free-Nucleotide Cytidine Deaminase Structures 12

Chapter 2 Purification of APOBECs 14
1. AID Purification 14
2. AID Activity 19
3. APOBEC-3B, 3F and 3G Purification 19
4. APOBEC-2 Purification 20
5. APOBEC-2 Crystallization 23

Chapter 3 The Structure of Human APOBEC-2 27
1. General Structural Features 27
2. Monomer Fold 29
3. Dimer Formation 31
4. Tetramer Formation 33
5. Distinctive Properties of the APOBEC-2 Tetramer 35
6. Hairpin-Loop Switch-Mediated Zinc Coordination and Enzyme 37
Activation
7. Oligomerization 39
8. Discussion 40

Chapter 4 The Quest for APOBEC-2 Activity 42
1. APOBEC-2 Substrates 43
2. Going “Fishing” for a Co-factor 45
3. Does APOBEC-2 have an RNA Inhibitor? 46
4. Is APOBEC-2 Regulated by Post Translational Modifications? 46
5. The Conserved SWS Motif 50
6. Inhibition by the Regulatory Loop or Glutamic Acid Residue? 50
7. APOBEC-2 DNA Binding 52

iv

Table of Contents (cont.)

8. Discussion 53

Chapter 5 Structure Modeling of AID and Implications for AID 54
Biochemical Activity
1. APOBEC-2 and AID Conservation of Sequence and Structure 54
2. AID Mutants as a Functional Surrogate 56
3. APOBEC-2 Structural Insights for Hyper-IGM-2 Immunodeficiency 59
4. Discussion 61

Chapter 6 Comparative Modeling of APOBEC3G and Implications for 62
other APOBECs
1. A3G Modeling 63
2. Mapping of A3G Functional Residues to the Dimer Interface 67
3. Discussion

Chapter 7 Experimental Procedures 71
1. Cloning and Purification of APOBEC-2 71
2. Structure Determination and Refinement of APOBEC-2 71
3. APOBEC-2 Glutaraldehyde Crosslinking 72
4. Construction of AID Mutants 72
5. Deamination Reactions 73
6. APOBEC-2 DNA Deamination Substrates 74
7. Comparative Modeling of A3G 74

References 76

v

List of Tables

Table 3-1 Data collection, phasing and refinement statistics (MIR) 28

vi
List of Figures

Figure 1-1 APOBEC cytdine deamination domains 2

Figure 1-2 AID induces SMH and CSR on Ig genes to generate high-affinity 4
antibodies

Figure 1-3 Anti-viral action of APOBEC-3G 6

Figure 1-4 Cytidine Deamination 11

Figure 1-5 Structure of a free nucleotide cytidine and cytosine deaminase 12

Figure 2-1 Purified recombinant GST-AID from E.coli 18

Figure 2-2 Purified recombinant GST-APOBEC proteins from E.coli 20

Figure 2-3 Expression test of purified APOBEC-2 22

Figure 2-4 Purification and Crystallization of tAPOBEC-2 24

Figure 2-5 The Synchrotron and Bunnies 26

Figure 3-1 The structure of APOBEC-2 29

Figure 3-2 The APOBEC-2 monomer structure 30

Figure 3-3 The APOBEC-2 dimer structure 32

Figure 3-4 The APOBEC-2 tetramer interface 34

Figure 3-5 The unique features of the APOBEC-2 active site 36

Figure 3-6 APOBEC-2 zinc coordination 37

Figure 3-7 The hairpin-loop switch and zinc coordination 38

Figure 3-8 Time course of glutaraldehyde crosslinking with truncated 40
APOBEC-2 protein (22kD)

Figure 4-1 I saw a ghost! 43

Figure 4-2 APOBEC 2 potential phosphorylation sites near the active center 48

Figure 4-3 The L1 loop on APOBEC-2 containins the regulatory E60 51
vii
List of Figures (cont.)

Figure 5-1 Structural guided mutagenesis of AID impairs deamination activity 55

Figure 5-2 Mutant AID deamination activity 58

Figure 5-3 AID HIGM-2 mutations 60

Figure 6-1 Sequence alignment of truncated Apo2 and A3G cytidine deaminase 63
domains (CD1, CD2)

Figure 6-2 Model structure of an A3G monomer and dimer based on the Apo2 65
tetramer crystal structure

Figure 6-3 Model structure of an A3G dimer interface joining between two CD1 68
subunits

viii
Abstract
APOBEC-2 (Apo2) belongs to the Apolioprotein B (APOB) mRNA-editing
enzyme catalytic polypeptide (APOBEC) family of cytidine deaminases that modify
genes by deaminating cytosines in mRNA coding sequences and cytidines in ssDNA
(Franca et al., 2006). APOBEC deaminases have remarkably diverse functions
comprising antibody maturation, inactivation of viral genomes including Human
Immunodeficiency virus-1 (HIV-1) and Hepatitis B virus (HBV), inhibition of
retrotransposition, and RNA editing. The repertoire of APOBECs and their
respective activities are continuously expanding to include new functions and sub-
classes of the family.
The APOBEC family is composed of APOBEC-1 (Apo1), Apo2, AID, and
APOBEC-3 (3A, 3B, 3C, 3DE, 3F, 3G, and3H) and APOBEC-4 (Conticello et al.,
2004; Sawyer et al., 2004). Previous comparative modeling of APOBECs has been
performed using the structures of free nucleotide cytidine deaminases (fntCDA)
which deaminate free cytidine/cytosine bases which share considerable sequence
homology with APOBECs. However, a comprehension of the molecular
mechanisms of the APOBEC enzymes has been limited by the lack of 3-dimensional
structures of APOBEC proteins that deaminate large nucleic acid substrates.
Here, a body of work is presented that describes the 2.5Å crystal structure of
Apo2, which is the first X-ray analysis reported for an APOBEC enzyme. The Apo2
structure reveals an elongated tetramer that differs markedly from the square-shaped
tetramer of fntCDAs. In Apo2, two long a-helices of a monomer structure prevent
the formation of a square-shaped tetramer and facilitate formation of the rod-shaped
ix
tetramer via head-to-head interactions of two APO2 dimers. The Apo2 active sites
can accommodate large RNA/DNA substrates and the structure reveals a possible
“built in” mechanism for regulating substrate access and enzyme activity.
Extensive sequence homology among APOBEC family members allows us to
test Apo2 structure-based predictions using activation-induced cytidine deaminase
(AID). Furthermore, the Apo2 structure can be used as a template for comparative
modeling to gain both structural and functional insight into these important enzymes.
1
Chapter 1
Background
In the past four years, the Apolioprotein B mRNA-editing enzyme catalytic
polypeptide (APOBEC) family of cytidine deaminases has expanded from four
known enzymes (APOBEC-1, APOBEC-2, Activation Induced Cytidine Deaminase
(AID) and APOBEC-3G) to a total of eleven enzymes with very remarkable and
diverse functions. The seven new APOBEC proteins include: APOBEC-3A, -3B, -
3C, -3DE, -3F, -3H and APOBEC-4. APOBEC-1 is involved in lipid metabolism
and AID plays a key role in antibody maturation. Most all of the APOBEC3 proteins
are capable of restricting viral pathogens, such as the Human Immunodeficiency
Virus (HIV) and Hepatatis B Virus (HBV) and inhibiting the mobility of
retroelements.
All of the APOBEC proteins share a conserved cytidine deamination
signature motif [H/C]XE and PCXXC (where X denotes any amino acid) (Jarmuz et
al., 2002). Some of the APOBEC enzymes have a single deaminase domain (Apo1,
AID, Apo2, A3A, A3C, A3H, Apo4), whereas others (A3B, A3F, A3G, A3DE),
through what is considered to be a gene duplication event, have evolved to have two
deaminase domains (Fig. 1-1). Although both domains contain the signature cytidine
deaminase motif, the N-terminal catalytic domain 1 (CD1) of A3F and A3G are
shown to be catalytically inactive whereas the C-terminal catalytic domain 2 (CD2)
is catalytically active (Navarro et al., 2005). The reason for this difference remains
unknown.

Figure 1-1. APOBEC cytdine deamination domains. Some APOBEC proteins contain a
single deaminase domain, whereas others have two deaminase domains. All deaminase
domains contain the signature cytidine deaminase motif. The active site residues [histidine
(H), glutamate (E), cysteine (C)]are colored in purple.

APOBEC enzymes have a preference for the particular cytidine they
deaminate depending on the flanking nucleotide sequence where the target cytidine
is located. For example, A3G has a very strong preference for a CCC motif
whereby it deaminates the third cytidine of the triplet on ssDNA (Chelico et al.,
2006) and seldom deaminates other cytidines outside that motif. However, AID
favors cytidines in the hot spot motif of WRC (R = purine; W = A or T), but also
often deaminates cytidines outside that motif (Pham et al., 2003). The precise
mechanism for cytidine recognition is currently not well understood although it is
thought that particular amino acid residues for each individual APOBEC are
involved (Bransteitter et al., 2004).
2
3
While APOBEC enzymes appear to have advantageous biological functions,
aberrant expression of these proteins could lead to unregulated deamination and/or
deamination of inappropriate substrates resulting in tumorigenesis. The aberrant
expression and deamination activity of AID has previously been shown to result in B
cell lymphoma (Pham et al., 2005; Bodor et al., 2005). Also, overexpression of
Apo1 leads to liver dysplasia and hepatocellular carcinoma (Yamanaka et al., 1995).

1. Activation-Induced Cytidine Deaminase (AID)
AID induces somatic hypermutation (SHM) and class switch recombination
(CSR) in activated germinal center B cells resulting in the production of high affinity
antibodies (Longerich et al., 2006; Bransteitter et al., 2006) (Fig. 1-2). SHM is a
process whereby point mutations are targeted a region of the Ig gene, the variable
region (VDJ), which codes for the portion of the antibody that contacts the antigen.
The purpose is to enhance to the affinity of the antibody to the antigen so to more
rapidly clear the antigen from the body. The mutations are tightly regulated and
specifically targeted to a region on the Ig gene that is approximately 1 to 2 kb
downstream from the Ig gene promoter. The VDJ region of the Ig gene is the main
target within that region. The mutation rate is extremely high, a million fold higher
than the mutation rate of any typical somatic cell in the body. The process and
regulation of SHM is an extraordinary cellular feat. One could not possibly survive,
if this type of mutatagenesis in the cell was not so precisely regulated. AID initiates
SHM by introducing multiple dC → dU mutations in the VDJ region when the Ig
gene undergoes transcription. These mutations often become permanent or they may
be acted on by alternative repair mechanisms that allow for mutations at bases, dA
and dT to occur (reviewed in Bransteitter et al., 2006).

Figure 1-2. AID induces SMH and CSR on Ig genes to generate high-affinity
antibodies. a. A cartoon of how AID expression is induced. A B-cell encounters an antigen
and is activated by a T-cell which induces AID expression. b. A schematic of an antibody
which consists of variable (V) and constant regions (C). The variable region of the antibody
makes contact with an antigen. During SHM, this region acquires a high rate of point
mutations in order to make a tighter fit to an anitigen. The C regions determine the type
isotype of the antibody, for example, IgM or IgA. c. A schematic of the Ig gene elements.
The arrow indicates the promoter. The variable region is in red boxes (VDJ) and the
constant regions are represented as boxes labeled: γ3, γ1, ( γ2b and γ2a not shown), α or ε.
E3’ is an enhancer regulatory region important for CSR. SHM results in point mutations
made in the variable region (VDJ) extending approximately 1 to 2 kb downstream from the
promoter. d. Schematic displaying CSR of an Ig gene. CSR occurs in the switch (S) regions
(1 to 2 kb) upstream of the constant region that undergo transcription. CSR switching to
downstream constant domains [ γ3, γ1, α or ε] results in the conversion of IgM or IgD to
other isotypes IgG3, IgG1, IgA or IgE.

CSR involves a recombination event on the Ig gene that that swaps out one
constant region for one further downstream on the gene. The expression of different
constant regions allow for different isotypes (for example, IgM or IgA, etc.) of
antibodies to be expressed. These different isotypes determine where in the body the
4
5
antibody will take action. AID initiates CSR by generating dC → dU deaminations
in the switch regions located upstream of the constant regions (Fig. 1-2d). This
deamination event leads to the double-DNA breaks that are required for swapping
out the constant regions (Bransteitter et al., 2006). Similar to SHM, AID
deamination events occur during transcription of switch regions which are located
upstream of the constant region. How these mutations are exclusively targeted to
these switch regions and the VDJ regions is yet to be understood.
The antigen induced expression of AID is required for SHM and CSR.
Individuals who lack a functional AID protein caused by mutations in their gene that
codes for AID develop a rare immunodeficiency disease called Hyper-IgM-2
(HIGM-2) syndrome. HIGM-2 patients are immuno-compromised and have
elevated levels of IgM antibody and lack IgG, IgE, and IgA antibodies. Additionally,
their antibodies have a weaker affinity for antigens and cannot efficiently clear
infections. For this reason HIGM-2 patients suffer with severe and recurrent
inflammatory and autoimmune disorders (Durandy et al., 2006; Minegishi et al.,
2000). This disease occurs because AID is not initiating SHM or CSR.

2. APOBEC-3 Enzymes (Apo3A, 3B, 3C, 3DE, 3F, 3G, 3H)
Apo3 enzymes are unique to primates. Their most publicized and
remarkable functions involve the inhibition of Hepatitis B Virus (HBV) and
retroviruses, such as HIV, and retroelements. The APOBEC enzymes with double
deaminase domains, A3B, A3F and A3G, most efficiently inhibit HIV-1 replication
(Cullen et al., 2006) (Fig. 1-3). Unfortunately, the reason we are still witnessing a
world wide AIDS epidemic is because HIV expresses a protein called the viral
infectivity factor, VIF. Vif binds and targets APOBEC proteins for proteasome-

Figure 1-3. Anti-viral action of APOBEC-3G. A3G gets incorporated into budding HIV-
1 virions. These new virions containing A3G are non-infectious when they fuse with a
target cell. In the target cell, APOBEC3G can restrict HIV-1 by two different mechanisms.
The first (1.) A3G is released with the viral RNA. It can bind to the viral RNA and block the
accumulation of reverse transcripts. The details of this mechanism are still unknown, but
cytidine deamination is not involved. Alternatively, (2) A3G can deaminate multiple
cytidines on the viral cDNA. These multiple deaminations can lead to degradation of the
cDNA by cellular repair enzymes or result in an inactive provirus.

mediated degradation, thereby eliminating the APOBEC3 protein from the cell or
reducing its expression to levels that are ineffective against HIV. In the laboratory,
strains of HIV that lack the VIF protein are used to study how APOBEC3 proteins
inhibit HIV. In cells that are infected with such HIV strains (lacking Vif), the
6
7
APOBEC3 enzymes are incorporated into HIV virions (Fig. 1-3). When these
virions infect new target cells, an APOBEC3 protein is released along with the viral
RNA. The APOBEC3 protein can then inhibit the reverse transcription of the HIV
RNA. Alternatively, APOBEC3 proteins introduce multiple dC→dU deaminations
on the minus strand of viral cDNA formed during reverse transcription (Cullen et al.,
2006). These mutations result in degradation of the viral cDNA or a defective
provirus.
APOBEC3A (A3A) and APOBEC3C (A3C) consist of only one CDA.
While A3C is weakly active against HIV-1, A3A displays no HIV-1 antiviral activity
(Cullen et al., 2006). Beyond HIV-1, A3G inhibits other retroviruses such as, simian
immunodeficiency virus (SIV), murine leukemia virus (MLV) and human T-cell
leukemia virus type 1 (HTLV-1) (Cullen et al., 2006). A3G, A3F, and A3B are
upregulated in hepatocytes by IFN-α and inhibit the replication of the hepatitis B
virus (HBV) (Tanaka et al., 2006; Bonvin et al., 2006). Similar to retroviruses, HBV
virions package an RNA that must be reverse transcribed to generate a double-
stranded DNA genome. This allows the APOBEC3 proteins to inhibit HBV by
similar mechanisms described above.
Additionally, the APOBEC3 enzymes protect the human genome against
endogenous retrotransposons. A3G and A3F block retrotransposition of Ty1
retrotransposons in yeast (Dutko et al., 2005; Schumacher et al., 2005). Recently,
A3A, A3B, A3C and A3F inhibit LINE 1 and Alu retrotransposition (Muckenfuss et
al., 2006; Stenglein et al, 2006; Bogerd et al., 2006).

8
3. APOBEC-2
Apo2, also known as ARCD-1, is ubiquitously expressed at low levels in
both the human and mouse and is predominantly expressed in cardiac and skeletal
muscle (Anant et al., 2001). Apo2, along with AID, is phylogenetically the most
ancient member of the APOBEC family (Conticello et al., 2004). Pro-inflammatory
cytokines, TNF- α and IL-1 β, induce Apo2 expression in hepatocytes via the NF-kB
response elements identified in the Apo2 promoter region (Matsumoto et al., 2006).
Apo2 can form heterodimers with Apo1 and inhibit the apoB RNA deamination
produced by Apo1 (Anant et al., 2001).
Similar to other APOBEC family members, Apo2 can be encapsidated into
HIV-1 virions when co-expressed with ∆vif HIV-1 DNA in 293T cells (Navarro et
al., 2005). It is therefore likely that Apo2 binds to the viral RNA since that is a
requirement for A3G virion incorporation (Khan et al., 2005). With the exception of
Apo3C, Apo2 and all of the other single domain APOBEC enzymes do not inhibit
HIV-1 viral replication and deamination is not observed on the reverse transcripts
(Navarro et al., 2005; Franca et al, 2006).
While the physiological roles have been characterized for many of the
APOBEC enzymes, the biological function of Apo2 remains to be determined.
Deamination of cytosine in ssDNA or RNA has been reported for all of the
APOBEC enzymes that are expressed in the human body with the exception of A3A,
Apo4 and Apo2 (Franca et al., 2006; OhAinle et al., 2006). Although Apo2 is able
to bind to the apoB RNA substrate, deamination is not observed (Anant et al., 2001).
9
Information gained from the Apo2 tetramer crystal structure presented in this body
of work offers some structural insight as to how Apo2 might be tightly regulated.

4. APOBEC-1
Apo1, the first APOBEC member to be characterized, is expressed in the
small intestine and edits the apoB mRNA in-vivo (Navaratnam et al., 1993). This
enzyme is the first and, thus far, only human APOBEC shown to deaminate RNA in-
vivo and/or in-vitro. Apo1 deaminates a particular cytosine of the apoB RNA,
specifically, cytidine 6666 to uridine, thereby creating a premature stop codon which
results in a truncated protein with a different function. ApoB plays a central role in
lipid metabolism and its two protein forms, apoB100 (full length) and apoB48
(truncated), are used to transport cholesterol and triglyceride, respectively, in the
blood.
In order for the precise mechanism of apoB mRNA editing to occur, Apo1
minimally requires the apoB RNA and an auxiliary protein, APOBEC-1
complementation factor (ACF). ACF is a novel RNA-binding protein that

serves as
the RNA recognition component of the editing enzyme (Lellek et al., 2000).
Additionally, a sequence of approximately ~26 nucleotides flanking the edited C on
the apoB RNA has been shown to be important for RNA recognition. Under
physiological conditions, the apoB RNA editing process actually combines the
efforts of a multi-component holoenzyme called the editosome.
10
In addition to editing mRNA, recombinant Apo1 can deaminate ssDNA in-
vitro (Petersen-Mahrt and Neuberger, 2003) and has previously been shown to
inhibit retroviruses in-vitro (Bishop et al., 2004).

5. APOBEC-4
Apo4 is a recently identified APOBEC subfamily that was discovered
through computational analysis by performing iterative sequence similarity searches
with (Rogozin et al., 2005). Although the protein’s physiological and biochemical
function still remains unknown, Apo4 contains conserved residues found in other
APOBEC proteins suggesting it may deaminate an RNA or DNA substrate.
Interestingly, despite the conserved sequence homology within residues important
for APOBEC activity, outside of these regions Apo4 seems to be the most sequence
divergent of the APOBECs. Expression arrays show that Apo4 is preferentially
expressed in the testes where it is suspected to be involved in spermatogenesis
(Rogozin et al., 2005).

6. Cytidine Deamination
Deamination is the hydrolytic conversion of cytidine to uracil and can occur
on free base cytosine, cytidine/deoxycytidine, dCMP or in the context of a
polynucleotide (DNA or RNA) substrate. Free nucleotide cytidine deaminases
(fntCDA) are found in both prokaryotes and eukaryotes and are essential for
nucleotide metabolism in the pyrimidine salvage pathway. Additionally, fntCDAs
function to degrade cytidine-based anti-tumor agents converting relatively non-toxic
5-fluorocytosine into a toxic 5-fluorouracil that can attack malignant cells (Greco
and Dachs, 2001).

Figure 1-4. Cytidine deamination. A view of the active site of cytidine deaminases
(Apo2). The active center zinc (red sphere) is coordinated by a histidine and two cysteines.
In fnt deaminases, the zinc is coordinated by three cysteines. An activated water molecule
serves as a donor. The glutamate acts as a proton shuttle during catalysis.

The fntCDA mechanism of cytidine deamination has been well characterized
and is presumed to be similar to that of APOBEC proteins. The active center is
distinguished by a zinc ion that is coordinated by three residues, either three
cysteines (fntCDAs), or two cysteines and one histidine (APOBECs), and a water
molecule (Fig. 1-4). The water acts as proton donor and a nearby glutamate
functions to shuttle the proton between the water and the target cytidine. The
activated water molecule produces a hydroxide ion for nucleophilic attack on the C4
of the cytidine (Fig. 1-4). The end product is the release of ammonia and uracil.
APOBEC enzymes are the first enzymes discovered that can deaminate cytosine on
DNA or RNA (Conticello et al., 2004). An understanding of their mechanisms and
11
substrate specificities has been limited by the lack of an APOBEC three-dimensional
crystal structure.

7. Free-Nucleotide Cytidine Deaminase Structures
Previous attempts to understand APOBEC proteins from a structural point of
view have been through comparative modeling using crystal structures of fntCDAs
owing to their close sequence homology with APOBEC within active site residues.
The fntCDAs oligomerize in two manners according to whether they deaminate free
cytidine or cytosine. The cytidine deaminases form homotetramers, whereas, the
cytosine deaminases form homodimers (Fig. 1-5a, 1-5b) (Johannason et al., 2002; Ko
et al., 2003). The dimeric cytosine deaminases have a long C-terminal tail and are
thought to have arisen from the tetrameric deaminases through a gene duplication
event (Johannason et al., 2002). Both deaminases share similar active-site

a b

12
Figure 1-5. Structure of a free nucleotide cytidine and cytosine deaminase. Both
deaminases have a square-shaped overall architecture a. The tetramer structure of human
cytidine deaminase (PDB accession number: 1MQ0). b. The dimer structure of yeast
cytosine deaminase (PDB accession number: 1UAQ).
13
architectures and an overall square-shaped organization of subunits (Fig. 1-5a, 1-5b).
In addition, the active sites within the two structures are covered by flaps either from
one monomer (cytosine deaminases) or from neighboring monomers (cytidine
deaminases) which may play a role in controlling substrate binding and subsequent
release (Ko et al., 2003). Modeling APOBEC enzymes off of these structures, it is
difficult to imagine how a DNA or RNA molecule could be accommodated in an
active site. To understand the mechanisms for how these APOBEC enzymes are
able to achieve such diverse functions, an APOBEC crystal structure is necessary.
Therefore, the focus of my research was to solve the crystal structure of any
APOBEC protein in order to gain valuable structural information about this family of
enzymes that previously did not exist. My research has resulted in the crystal
structure of Apo2 and modeling of other important APOBEC family members.
14

Chapter 2
Purification of APOBECs
Difficulties with solubility and purity of APOBEC enzymes have prevented
researchers from obtaining large quantities of protein sufficient for structural studies.
Previous modeling of APOBECs has been based off of the crystal structures of
fntCDAs which do not deaminate nucleic acid. In order to better understand the
APOBEC family members from a structural perspective, we purified APOBEC
proteins in hopes of obtaining pure protein for x-ray crystallography studies.

1. AID Purification
The first APOBEC enzyme that I attempted to purify, in collaboration with Dr.
Ronda Bransteitter, a former graduate student in Myron Goodman’s lab (hereafter
this collaboration will be referred to as “we”), was AID. Purification of this enzyme
proved difficult because AID purified in combination with numerous other proteins
and nucleic acid. Crystallization of a protein requires large quantities of very pure
protein; therefore, developing a system to purify AID away from these dreadful
contaminants was necessary.
The initial problem with AID purification was due to AID insolubility.
Expression was not an obstacle as AID expresses well in E. coli. Unfortunately,
most of the AID expressed sedimented to the insoluble fraction (cell pellet) after cell
lysis followed by centrifugation. AID’s insolubility could potentially be a result of
15
the protein binding nucleic acid or other nucleic acid binding proteins which
sediment to the insoluble form in the cell pellet.
Typically in an x-ray crystallography lab, there are some initial approaches to
the challenging problems of protein insolubility. In purifying AID, we tried some of
these approaches by altering: the expression temperature (16°C, 25°C, 37°C), the
shaking speed (150RPM, 200RPM, 250RPM), the OD
600
induction (0.2, 0.4, 0.6, 0.8,
1.0), the pH of the lysis buffer (6.0, 7.0, 8.0, 9.0, 10.0), and the salt concentration of
the lysis buffer (100 mM, 250 mM, 500 mM, and 1 M NaCl). We found the best
conditions for purifying AID included the following: expression temperature 16°C,
shaking speed 200RPM , OD
600
induction (0.4), lysis buffer (Hepes pH 7.5, 250mM
NaCl, 1mM DTT). Although these conditions did not actually increase AID
solubility, they were the conditions in which AID seemed to be the most stable and
enzymatically active.
AID contains 7 cysteines in its amino acid sequence. We considered the
possibility that these cysteines, depending on their location in the protein, may pose a
problem during purification by forming disulfide bonds with other neighboring
monomers. Potentially, these aberrant disulfide bonds could result in aggregation of
the protein. On the other hand, the cysteines could play an important role in protein
folding by helping to stabilize the protein structure by forming internal disulfide
bonds. Therefore, we tried altering the amount of the reducing agent, dithiothreitol
(DTT), used during purification. We observed no difference between 1 mM, 5 mM
and 10 mM DTT; therefore, we used 1mM DTT for AID purification.

16
In an attempt to increase the solubility of AID, several methods were used to
disrupt the protein interactions of AID with its contaminants. AID contains an N-
terminal region with an exceptionally large positive charge (+11). Previously, it was
proposed that this positively charged region facilitates binding of AID to its
negatively charged substrate, ssDNA (Bransteitter et al., 2004). For this reason, we
made several truncations of the N-terminal region of AID based on secondary
structure predictions in hopes to prevent binding of AID to nucleic acid. The
constructs created maintained amino acid residues 3-198, 26-198, and 39-198 (full
length AID contains residues 1-198). Following expression and further purification,
these N-terminal deletions appeared to make no difference in the solubility of AID
compared with wild type AID. Five additional C-terminal deletions of AID were
also created containing amino acid residues 1-188, 1-181, 1-160, 1-133, and 1-120.
Similar to the N-terminal deletion constructs, these constructs displayed no increase
in solubility.
As discussed previously, AID expresses at moderate levels in E. coli; therefore,
we attempted to enhance solubility of the protein from the cell pellet. One strategy
was to use reagents to free the cell lysate of nucleic acids which may be binding
AID, consequently causing it to pellet to the insoluble fraction. RNase and DNase
were both added at different stages during the purification process (cell lysis and
incubation with resin). Unfortunately, this resulted in no change in AID solubility.
Different pHs (6-10), different salt concentrations (50 mM-1 M NaCl), and addition
of Triton-X-100 also had no positive effect on solubility for AID. Also, addition of
17
ssDNA (random 80mer), and another known AID binding factor, replication protein
A (RPA), to the AID cell lysate did not increase solubility of AID.
Misfolding of AID could be another potential explanation for AID pelleting to
the insoluble fraction. Interestingly, another protein which migrates on SDS PAGE
with an apparent molecular mass of about 60 kD, purifies in approximately a 1:1
ratio with GST-AID in the soluble and insoluble fraction. Perhaps this is the
essential E. coli chaperone protein GroEL, otherwise known as heat shock protein
60, which, together with GroES, assists in folding of proteins. In an attempt to try to
refold and solubilize AID and denature GroEL, we treated the cell lysate with a
denaturant followed by dialysis. Several concentrations of UREA (3.2 M, 3.5 M, 3.8
M and 4.0 M) were added to GST-AID lysate samples followed by overnight dialysis
with lysis buffer. Unfortunately there was no change in the amount of GST-AID in
the cell pellet and soluble GST-AID still was found in a 1:1 ratio with the 60kD
protein. Higher concentrations of UREA should be tested in the future.
Despite the fact that we were unable to obtain large amounts of pure AID, we
succeeded in purifying small amounts of GST-tagged full-length AID from
glutathione resin (Fig. 2-1). GST-AID did not cleave using thrombin and the full-
length fusion protein still remained bound to the resin. A 3-glycine linker was added
between the GST protein and AID to provide for more flexibility to allow thrombin
to access the cleavage site between the proteins. With the glycine linker inserted, the
GST-AID fusion protein cleaved, but AID did not elute and instead remained bound
to the resin.

Figure 2-1. Purified recombinant GST-AID from E.coli. GST-AID was purified by
glutathione-Sepharose affinity chromatography and run on a 15% SDS PAGE gel. Lane 1,
protein molecular mass standards. Lane 2, 15ul of a reduced glutathione elution from GST
resin. The GST-AID band is indicated by the red asterisk (~50kD).

Due to the insolubility of the cleaved AID on the resin following cleavage, we
used reduced glutathione to elute the fusion protein from the glutathione resin (Fig.
2-1). The exact reduced glutathione buffer conditions were crucial as we discovered
that a very high concentration of reduced glutathione (50mM), along with a high
concentration of Tris (100mM) was necessary to elute the AID fusion protein.
Nonetheless, the AID fusion protein never completely eluted from the resin. The
eluted fusion protein was cleaved with thrombin in solution and AID was resolved
by gel filtration (Superose 6). The remaining cleaved AID eluted in the void volume
fractions, presumably because it was bound to nucleic acids or formed aggregations
with its self or contaminants. Also, the fusion protein was purified by gel filtration
(Superose 6) and similarly eluted in the void volume of the column. Unfortunately,
none of the purifications schemes tested resulted in pure AID.
18
19
2. AID Activity
Despite incomplete purification, GST-AID deamination activity was tested
using known AID substrates. The specific activity of AID expressed in E. coli was
at least 50-fold less when compared to insect cell purified AID (Bransteitter, 2003).
Specific phosphorylation of AID has previously been shown to affect AID function;
therefore, differences in AID activity from protein purified from these two different
model systems could be a result of the post-translational modifications occurring in
insect cells. Additionally, as previously discussed, AID expressed in E. coli may be
partially misfolded which may have a negative effect on the protein’s activity. We
also found that AID expressed at 16°C was more active then AID expressed at 25°C
or 37°C. In spite of the large difference between activity in AID expressed in E. coli
and insect cells, we did observe sufficient deamination activity to use for
biochemical activity tests (see Chapter 5).

3. APOBEC-3B, 3F and 3G Purification
Following our attempts (i.e. long days and exhaustive nights) at purifying AID,
we decided to try to purify other APOBEC family members for crystallization
purposes. At the time (2004-2005), only a few of the APOBEC proteins were
characterized with known activities. Moreover, the only structural information that
was available for APOBECs was from the solved crystal structures of the free base
cytidine deaminases. We knew if we could crystallize and solve the structure of any
of the APOBEC proteins, we would gain valuable structural information that would

Figure 2-2. Purified recombinant GST-APOBEC proteins from E.coli. GST fusion
proteins were purified and run on 12% SDS PAGE gels. Respective fusion proteins are
indicated by a red asterisk. Lane 1, 15ul GST-AID. Lane 2, 15ul GST-A3G. Lane 3, 15ul
GST-A3B. Lane 4, 15ul GST-A3F. Lane 5, 2ul GST-Apo2.

be relevant to all of the APOBEC proteins. Therefore, we attempted to express and
purify as many APOBEC proteins that we had clones for: A3B, A3F, A3G and
Apo2.
A3B, A3F and A3G were expressed in E. coli using the same purification
scheme as AID. All of these proteins had extreme purification problems similar to
AID and did not purify at concentrations necessary for crystallization purposes.
Before we had the chance to really characterize these proteins or optimize their
purification protocols, Apo2 surprised us!

4. APOBEC-2 Purification
Apo2, in contrast to the previously purified APOBEC proteins, displayed high
levels of expression and solubility. In fact, Apo2 was so soluble that the first Apo2
20
21
expression test samples I ran on a protein gel were so overloaded with protein I
falsely accused my undergraduate student of making the protein gels wrong causing
them to run funny (Fig. 2-3a). After re-running the gel and examining the results
further, I realized Apo2 was the first APOBEC protein I was able to express in large
soluble quantities (Fig. 2-3b) This was just the beginning of many thrills and reasons
for screams of delight and celebration (i.e. Pancho’s margaritas and caipirinhas) that
Apo2 presented for us.
The protein expression and purification of Apo2 are described as follows:
a 2ml overnight cell culture was used to inoculate 1 liter flask of 2xYT at 37°C
(Ampicillin 100ug/ml); cells were shaking at 225RPM and were grown to an
A
600
=0.4 then transferred to 25°C for 30 minutes; IPTG was added to a final
concentration of 0.2mM; the cells grew overnight for approximately 16-20 hours.
After pelleting the cell, they were resuspended in 60ml of lysis buffer (50 mM Hepes
(pH 7.0), .25 M NaCl, 1 mM DTT) and lysed by sonication. Cell lysate was clarified
by centrifugation at 10,000 g for 1 hour. The supernatant containing the GST-Apo2
fusion protein was incubated on the nutator for 2 hours with 10 ml glutathione resin.
Gravity glutathione affinity columns were used to further purify the bound fusion
protein. Apo2 was cleaved from the GST using thrombin with an approximate ratio
of 20 units thrombin to 1 ml of glutathione resin. The digestion was carried out for
45 minutes at room temperature and then resin and protein was transferred to 4°C for
45 minutes. Typically, four subsequent thrombin digestions were necessary for full
Apo2 cleavage. Apo2 was further purified by Superdex-75 gel filtration in a buffer
containing 25 mM Hepes (pH 7.0), 50 mM NaCl, 1 mM DTT. Peak fractions were

a

b

Figure 2-3. Expression test of purified APOBEC-2. a GST-Apo2 was purified and run
on a 15% gel. 5ul protein was run in each lane and gel was overloaded with protein causing
it to run funny. b GST-Apo2 was lysed by sonication and supernatant was incubated with
GST resin. Lane 1, protein molecular mass standards. Lane 2, 2ul of cells. Lane 3, 2ul of
cells after sonication. Lane 4, 5ul of supernatant. Lane 5, tip touch of cell pellet. Lane 6, 5ul
of GST-resin flow though (wash). Lane 7, 10ul GST-Apo2 bound glutathione resin. The
GST-Apo2 band is indicated by red asterisk (~52kD).

collected and concentrated to a final concentration of 25mg/ml. Total Apo2 protein
yield from one liter of cells on average was 100mg.
22
23
Proteins frequently have difficulty crystallizing for many reasons. Flexible
regions within a protein often hinder successful protein crystallization. Predictions
of disordered regions for the Apo2 amino acid sequence show the N terminus to be
highly flexible to approximately residue 40. Based on disorder and secondary
structure predictions, deletion contructs of Apo2 were created to try to circumvent
crystallization problems. The Apo2 truncation constructs created were of amino acid
residues: (Tr1)1-216, (Tr2)1-139, (Tr3) 5-224, (Tr4) 41-224, (Tr5) 5-216, (Tr6) 5-
139, (Tr7) 41-216, (Tr8) 41-139 (full length Apo2 contains residues 1-224).
Expression profiles of the truncation constructs showed Tr3 and Tr4 expressed
similar to full length Apo2. Tr1, Tr5, and Tr7 expressed at moderate levels and Tr2,
Tr6 and Tr8 expressed at low levels when compared to full length Apo2.

5. APOBEC-2 Crystallization
Since Tr3 and Tr4 expression levels were equivalent to full length Apo2, all
three proteins (Tr3, Tr4, and full length Apo2) were purified for crystallization
purposes as described above. Using an automated high-throughput robot system,
concentrated pure protein was set-up at 18°C in 96-well trays with an assortment of
Hampden and NeXtal crystal screens. Several protein concentrations and dilutions
of crystal screens were tried; however, full length and Tr3 protein never crystallized.
Perhaps this is due to the predicted disordered N-terminal region still present in both
of these constructs. Tr4 thankfully crystallized in several similar buffer conditions,
yet different crystal forms, yielding another exhilarating Apo2 moment (Fig. 2-4a, 2-
4c)!
Tr4 crystals grew quickly (again- thank you Apo2-we love you). Initial Tr4
crystal hits grew plentifully as small (10mµ by 10mµ), needle clusters. Manual
optimization of pH, PEG concentration, additives and salt concentration eventually
produced more manageable but thin plate crystals (150mµ by 50 mµ). Some crystals

Figure 2-4. Purification and Crystallization of tAPOBEC-2. a GST-tApo2 was
purified and run on a 15% SDS PAGE gel. Lane 1, protein molecular mass standards. Lane
2, .5ul tApo2. Lane 3, 1ul tApo2. Lane 4, 2ul tApo2 b, Diffraction pattern from tApo2
multiple plate crystals showing multiple, smeared diffraction c Examples of tApo2 crystal
forms. Upper left panel represents multiple plate crystals that produced a smeared multiple
diffraction pattern. Upper right panel represents optimized plate crystals that still produced
multiple diffraction patterns. Lower panel shows an example of a well of tApo2 crystals
with multiple plate forms.

24
25
appeared to be large, multiple plate crystals, whereas others were small and very thin
crystals (Fig. 2-4c). The first Tr4 plate crystal diffracted to 2.1 Ǻ on the home
detector (I am positive everyone in the building heard my screams of excitement).
One angle of the plate crystal produced a clean diffraction pattern; unfortunately, the
next 90° test shot produced a smeared, multiple diffraction pattern (Fig. 2-4b). This
type of diffraction pattern is indicative of multiple thin plates growing one on top of
each other.
Tireless efforts to optimize the plate crystals resulted in single, thicker plate
crystals yet no significant differences in diffraction patterns (Fig. 2-4c upper-right
panel). Although it appeared that it was a single crystal in the microscope, in
actuality, the plate crystals had several layers. All but a few of the countless crystals
tested on the home detector yielded multiple diffraction patterns from at least one
angle. Therefore, before my first trip to the synchrotron in January 2006, it was
decided that the only way to ensure we would obtain a clean diffraction pattern from
a crystal is through numbers. At least 90 crystals were looped and frozen in hopes to
guarantee that one of the crystals might produce a clean, single diffraction pattern.
Fortunately, the numbers scheme worked and one of the crystals produced a
clean diffraction pattern (it happened to be a test crystal that I thought would not be a
single crystal). The best crystal ended up being a small shard broken off of a larger
plate crystal.
After collecting good native data at the synchrotron in January, selenium
methionine protein was grown and crystals were taken back to the synchrotron in
February. The selenium tApo2 protein expressed at the same high levels as the

Figure 2-5. The Synchrotron and Bunnies. Collage of the Advanced Light source (ALS)
in Berkeley, CA. Upper left panel, the dome of the synchrotron storage ring, a tubular
vacuum chamber, which allows an electron beam to travel at nearly the speed of light.
Upper right panel, Dr. G.G., Aaron, Dahai, and X-Man hard at work at the ALS (Jan. 2006).
Lower left panel, “Scooter” and “Alpha” the Culver City late night bunnies who greeted us
on my street at 3am after working late in the lab. RIP Scooter. Lower right panel, the tApo2
“BINGO” crystal that produced a single diffraction pattern, leading to the Apo2 structure.

native protein and also grew in similar crystal conditions. Again, a large amount of
crystals were looped and frozen (80 crystals) for the second synchrotron trip. After
screening quite a few, we eventually obtained a clean, single diffraction pattern that
lead us to solving the Apo2 crystal structure (Fig. 2-5).

26
27
Chapter 3
The Structure of Human APOBEC-2
APOBEC deaminases constitute an enzyme family with remarkably diverse
functions comprising antibody maturation, inactivation of viral genomes including
HIV, inhibition of retrotransposition, and RNA editing. Here we report the first
crystal structure of an APOBEC family member, APOBEC2 (Apo2). Apo2 forms a
rod-shaped tetramer that differs markedly from the square-shaped tetramer of the
free nucleotide cytidine deaminase (fntCDA). The Apo2 active sites can
accommodate large RNA/DNA substrates. Two long α-helices prevent the formation
of a square-shaped tetramer and facilitate the elongated tetramer formation via head
to-head interactions of two Apo2 dimers. Within an Apo2 dimer, two monomer β-
sheets are augmented to form one broad β-sheet.

1. General Structural Features
Apo2, containing residues 41-224, was crystallized in the space group
p212121 (Table 3-1), and the structure was solved using the selenium multi-
wavelength anomalous dispersion (Se-MAD) phasing method. There are four
monomers within each asymmetric unit that form a tetramer with an atypical
elongated shape (Fig. 3-1a). The formation of the tetramer occurs through two
different monomer-monomer interfaces. This tetrameric architecture is unusual and
differs from the canonical square-shape or ring-shape of the fntCDA tetramer (Fig.

28

Native
(λ = 0.9774 Å)
Peak Se
(λ = 0.9796 Å)
Inflection Se
(λ = 0.9798 Å)

Data collection

Space group P2
1
2
1
2
1
P2
1
2
1
2
1
P2
1
2
1
2
1
Cell dimensions
a, b, c (Å) 37.841, 89.41,
245.77
37.862, 88.952,
244.533
37.888, 89.016,
244.644
α, β, γ (°) 90.0, 90.0, 90.0 90.0, 90.0, 90.0 90.0, 90.0, 90.0
Resolution (Å) 50-2.5 (2.59-
2.50)*
50-2.8 (2.90-
2.80)
50-2.8 (2.90-
2.80)
Observations 383,800 547,891 540,462
R
merge
15.1(33.3) 10.7(32.2) 10.2(35.5)
I / σI 17.7(2.5) 21.9(4.3) 21.1(3.8)
Completeness (%) 96.3(85.5) 96.4(83.0) 96.9(85.2)

Refinement
Resolution (Å) 50-2.5
No. reflections 24,699
R
work
/ R
free
24.56/29.54
No. atoms
Protein 5,994
Zinc 4
Water 88
B-factors (Averaged)
Protein 29.97
Zinc 34.35
Water 26.97
R.m.s deviations
Bond lengths (Å) .009299
Bond angles (°) 1.01753

Table contains data collection statistics obtained from a total of two crystals.
*Highest-resolution shell is shown in parentheses.

Table 3-1 Data collection, phasing and refinement statistics (MIR)

3-1b) with all four monomers interacting with each other (Johansson et al., 2002).
The elongated Apo2 tetramer has the appearance of a butterfly (Fig. 3-1a) and an
end-end span of approximately 126.9Å.
a

b

Figure 3-1. The structure of APOBEC-2. a. The APO2 tetramer structure. It has an end-
to-end span of 126.9 Ǻ. Zinc atoms in the active centers are shown as red spheres. b. The
square-shaped structure of human cytidine deaminase (PDB accession number: 1MQ0), a
fntCDA.

2. Monomer Fold
29
Despite the unusual tertiary structure of the Apo2 tetramer, the monomer
appears to adopt the typical core fold of the fntCDAs with a five-stranded β-

sheet flanked by helices on both sides (Fig. 3-2a, 3-2b). However, unique features of
the Apo2 monomer structure determine the formation of a new type of oligomer
occurring at the dimer and tetramer interfaces. One new attribute is the additional α-
helices surrounding the core β-sheet (Fig. 3-2a, 3-2b); six long helices are present in
the Apo2 monomer while only three or four are observed in the fntCDA monomer
a b

Figure 3-2. The APOBEC-2 monomer structure. a. One view of the monomer. b.
Another view of the monomer rotated by 90 degrees, showing the unique features of Apo2:
the short β ’ strand and helices h4 and h6. H4 and h6 dictate how Apo2 oligomerizes.

(exclusive of the shorter 3 α helices) (Johansson et al., 2003; Xie et al., 2004; Teh et
al., 2006; Chung et al., 2005). Based on the close sequence homology of Apo2 with
other APOBEC family proteins, the long helix (h4) inserted between strands β4 and
β5 is likely to serve as a structural signature (Fig. 3-2a, 3-2b). Helix 4 (h4) is located
on the same side as h2 and h3 (Fig. 3-2b) and on the outer edge of the molecule
30
31
packed between h3 and h6 (Fig. 3-2b). Helix 6 (h6) is located at the C-terminus and
is oriented approximately vertical to the β−sheet and positioned on both sides of the
dimeric β-sheet at the edge of the β5 strands. In this configuration, h3 and h6 make
extensive bonding contacts with h4. These contacts have a central role in stabilizing
the helices within the Apo2 monomer subunit. The placement of h4 and h6 around
one edge of an Apo2 core fold is the vital feature that prevents the formation of a
square-shaped fntCDA tetramer. The essential point is that h4 and h6 occupy the
space where another neighboring molecule would need to sit to form a square
tetramer.

3. Dimer Formation
The Apo2 dimer is formed by the pairing of two long β-strands (β2) (Fig. 3-
3a), which expands the β-sheet sideways forming one wide and slightly bent β-sheet
that resembles a ribcage (Fig. 3-3a, Fig. 3-1a). This enlarged core β-sheet of the
dimer appears to have folded from a single molecule.
Twelve residues (residues 82-93) on each β2 strand form 12 hydrogen bonds
through main chain carbonyl oxygen and nitrogen atoms and provide the principal
bonding force between the two monomers. The dimer interface is reinforced by the
side-chain interactions occurring through the loops and helices located on both sides
of the β-sheet. Interactions with several water molecules also help to stabilize this
interface.
The dimer is nearly symmetrical (Fig. 3-3a, 3-3b, 3-3c) with 6 helices (h2, 3,
4 of both molecules) located on one side of the augmented β-sheet and 4 helices (h1
and 5 of both monomers) located on the other side. Capped on both edges of the β-
sheet are h4 and h6. However, one part of the dimer shows obvious asymmetry at
the turn between helix 1 and strand β1 (h1/β1-turn). This h1/β1-turn (residues 57-68)
a

b c

Figure 3-3. The APOBEC-2 dimer structure. a. The Apo2 dimer is formed by two
monomers (in yellow and red) that are paired along their β2 strands. The paired β2 strands
augment each individual β-sheet sideways to form one central β- sheet that is encased by α-
helices. The fold of the Apo2 dimer appears to be derived from one molecule instead of the
actual two monomers. b,c Two views of an Apo2 dimer are rotated by 180° degree with
respect to each other. Notice the h1/β1-turn (in red) has two different conformations; one has
a β-hairpin (b) and the other has a loop (c). Helices 4 and 6 are positioned on both edges of
the dimeric β-sheet.structure is composed of a small β-strand (residues 57-63) designated
β1’ and a short strand (residues 65-69) that becomes an extension of the N-terminal portion
of the β1 strand.

32
33
assumes a hairpin structure (β1’-hairpin) in one monomer (Fig. 3-3b), whereas the
same turn assumes a loop conformation (L1) in the other monomer (Fig. 3-3c). The
β1’-hairpin structure is composed of a small β-strand (residues 57-63) designated
β1’ and a short strand (residues 65-69) that becomes an extension of the N-terminal
portion of the β1 strand. The Apo2 h1/β1-turn may have a conserved functional role.
Near the Zn-active center, the h1/β1-turn, either in the loop or β-hairpin
conformation, is in the approximate position of a fntCDA loop. The fntCDA loop is
supplied in trans by a different monomer, whereas in Apo2, the h1/β1-turn is part of
the single monomer fold. Therefore, this h1/β1-turn near the Zn-active center in cis
may assume a similar functional role as the loop supplied in trans in the canonical
square-shaped fntCDA tetramer. In the following sections, we will discuss how and
why the structural transformation between a hairpin and a loop of the h1/β1-turn is
likely to have significant functional consequences.

4. Tetramer Formation
The Apo2 tetramer is formed by two dimers joining through head-to-head
interactions. The two dimers make contacts through the monomer that has the loop
structure (L1) at the h1/β1-turn, leaving the monomer with the β1’-hairpin structure
on the outer ends of the tetramer (Fig. 3-4a). The total buried area is 1745 Å at the
tetramer interface, where h4 and h6 play an essential role in bonding interactions
(Fig. 3-4a). These same two helices sterically hinder the formation of the square-
shaped fntCDA-type tetramer. Consequently, h4 and h6 appear to be a major
determinant for the elongated tetramer formation in Apo2.
a

b

Figure 3-4. The APOBEC-2 tetramer interface. a. The tetrameric interface region shows
the locations of h4, h6, and L1, which provide extensive interactions between the monomers
in the interface. b. A view of the residues in h4, h6 and L1 that are involved in the
interactions at the tetramer interface. Hydrophobic, polar and charged amino acid side chains
are all involved in the interactions. Some of the charged amino acids also use the aliphatic
side chains to form van der Waals interactions with the hydrophobic residues.

Residues from h4 and h6, as well as the loop L1, interact extensively at the tetramer
interface (Fig. 3-4b). Hydrophobic, polar and charged amino acid side chains are
34
35
involved in these interactions. Residues Y61, F155, M156, W157, P160, Y214, and
Y215 from each side of the interface form extensive hydrophobic packing
interactions. Residues R57, S62, S63, R153, E158, E159 and E161establish salt
bridges and hydrogen bonds (Fig 3-4b). Some of the charged amino acids even use
their aliphatic side chains to form van der Waals interactions with hydrophobic
residues. Water molecules also play a role in the tetramer interface interactions.

5. Distinctive Properties of the APOBEC-2 Tetramer
A prominent feature of the tetramer structure of Apo2 that distinguishes it
from the fntCDA tetramer is that the active sites are accessible for large RNA or
DNA substrates (Fig.3-5a). In the square-shaped fntCDA tetramer, there are loops
from two neighboring monomers that surround the active sites and block
accessibility to larger nucleic acid molecules (Fig. 3-5b, 3-5c). Therefore, these
fntCDA active sites can only accommodate nucleotide substrates. Although the yeast
free fntCDA, CDD1, has been reported to deaminate the APOB mRNA in vitro, its
known biological substrate in vivo is a free nucleotide, and the CDD1 structure is a
canonical square-shaped tetramer (Xie et al., 2004).
In fntCDAs and APOBEC proteins, the active center Zn atom is coordinated
by three residues (either three Cys, or two Cys and one His), with a water molecule
located at a hydrogen bond distance of approximately 3.0 Å from the Zn. This
coordination bond distance is reduced to 2.0 - 2.2 Å after the water is activated and
becomes a Zn-hydroxide for nucleophilic attack. This type of Zn coordination is also
present in Apo2 (Fig. 3-6a), but only in two of the monomers located on both ends of
the tetramer. Surprisingly, the active sites for the other two monomers in the middle
of a tetramer contain the E60 residue, which makes a fourth coordination bond with
a

b c

Figure 3-5. The unique features of the APOBEC-2 active site. a. The Apo2 tetramer
consists of four independent active sites that are accessible to large nucleic acid substrates,
such as DNA or RNA. The residues and the Zn atom (sphere in purple) at the active sites are
drawn and labeled. b. A slabbed view shows the active site residues of the fntCDA monomer
(Chung et al., 2005). c. A full view of the fntCDA active site shows the limited substrate
accessibility. Loops from two neighboring monomers (colored in grey and in wheat) in the
fntCDA tetramer fold over the active site making it accessible only to small free nucleotides
and not to large DNA or RNA molecules.

the Zn atom at a bond distance of 2.2 Å (Fig. 3-6b). This coordination of Zn by four
amino acid residues is clearly unexpected considering that all known cytidine
36
deaminase structures have only three amino acid residues participating in Zn-
coordination (Johansson et al., 2003; Xie et al., 2004; Teh et al., 2006; Chung et al.,
2005; Betts et al., 1994; Smith et al., 1994).

a b

Figure 3-6. APOBEC-2 zinc coordination. a. The active sites of the outer molecules of the
Apo2 tetramer show Zn-coordination by three residues (H98, C128 and C131) at a bond
distance of 2.0 – 2.2 Å and a water molecule located at a hydrogen bond distance of 3.4 Å.
b. The active sites of the middle two molecules of the Apo2 tetramer show Zn-coordination
by four residues. The fourth coordination bond (2.2Å) is formed by E60, which results in the
exclusion of the water molecule necessary for deamination.

6. Hairpin-Loop Switch-Mediated Zinc Coordination and Enzyme Activation
A closer examination of the structure reveals a “built-in” mechanism for a
conformational switch between the two types of Zn coordination. The switch is
mediated by sequences contained in the h1/ β1-turn (residues 57-68), in which E60 is
located. The h1/ β1-turn can adopt either a hairpin ( β1’-hairpin) or a loop (L1)
conformation (Fig. 3-3b, 3-3c), which controls whether or not E60 coordinates with

37
a b

c

Figure 3-7. The hairpin-loop switch and zinc coordination. a. The h1/ β1-turn folds into
the β1’-hairpin structure in the outer two Apo2 monomers. The hydrophobic ring of Y61
interacts with the guanidine group of R65 stabilizing the β1’-hairpin conformation. In this
conformation, the Zn atom (sphere in red) is coordinated by three residues. The water
molecule is drawn as a sphere in blue. Yellow dashed lines represent coordination bonds and
gray dashed lines indicate a hydrogen bond between the water and the Zn. b. The h1/ β1-turn
folds into the loop structure in the inner two Apo2 monomers. In this loop conformation, the
E60 is only 2.2Å from the Zn atom and forms a coordination bond. Y61 now rotates away
from R65 and interacts with the guanidine group of R57, which facilitates the disruption of
the β1’-hairpin and stabilizes the loop conformation. The water molecule is excluded from
this active site. c. Superposition of two monomers emphasizes the difference between the
two conformations of the h1/ β1-turn. The h1/ β1 loop (purple) is pulled down approximately
8.5Å toward the active site. In this position, the E60 coordinates with Zn and substrate
accessibility appears to be restricted.
38
39
Zn. The E60 carboxyl group is located 6 Å from the Zn when the h1/ β1-turn is in the
β1’-hairpin conformation (Fig. 3-7a). The β1’-hairpin is stabilized by main-chain
hydrogen bonds within the β1’-hairpin and reinforced by interactions between the
hydrophobic ring of Y61 and the guanidine group of R65.
In the center monomers, the h1/ β1-turn folds into the loop conformation (L1),
which may be stabilized by its interactions within the tetramer interface. In this
conformation, the Y61 on the h1/ β1-turn rotates its side chain to interact with R57
instead of the R65 (Fig. 3-7b). The new pairing of Y61 with R57 destabilizes the
β1’-hairpin while stabilizing the loop conformation. In the loop conformation, the
E60 has a coordination distance of 2.2 Å from the Zn atom (Fig. 3-7b).
The hairpin-loop switch may have two significant consequences. First,
switching to the loop and forming the fourth Zn-coordination by E60 prevents
coordination by water and subsequent Zn-hydroxylation necessary for deamination
(Fig. 3-6b, 3-7b). Second, Zn-coordination by E60 pulls the h1/ β1-turn toward the
active center by approximately 8.5 Å (Fig. 3-7c), which could restrict substrate
access to the active center. On the other hand, breaking of the fourth coordination of
E60 may allow the loop to become displaced away from the active center to form the
β1’-hairpin as observed in the outer monomers. The E60 would no longer prevent
the Zn hydroxylation and nucleic acid substrates would have full access to those
active sites. The hairpin-loop switch can be a molecular mechanism for regulating
substrate access and enzyme activity mediated by through Zn-coordination.

7. Oligomerization
We performed additional experiments in order to understand more about the
oligomerization behavior of Apo2 in solution. In gel filtration assays the monomer
was the major species while the tetramer was a minor species when compared with
the dimer, suggesting a stronger dimeric interaction. By using size exclusion and
dynamic light scattering, as well as cross linking experiments, we have demonstrated
that both Apo2 forms dimer and tetramer in solution (Fig. 3-8). Cross-linking
experiments revealed the presence of an Apo2 monomer, dimer, trimer and tetramer
in solution.

Figure 3-8. Time course of glutaraldehyde crosslinking with truncated APOBEC-2
protein (22kD) Reactions with Apo-2 (2ug) were performed for the indicated time with
0.25% glutaraldehyde at room temperature and quenched with 1M Tris, pH 8.5, 2X SDS
loading buffer and run on a 12% SDS PAGE gel for 70 min at 200V. Lanes marked M and 0
were analyzed by Coomassie staining and lanes marked 5, 15, 30 and 60 were analyzed by
silver staining. Crosslinking reveals a monomer, dimer, trimer and tetramer band.

40
41

8. Discussion

The Apo2 structure revealed striking differences from the previously solved
fntCDAs including how it oligomerizes and potentially binds nucleic acid substrate.
Additionally, the structure shows how the active site of Apo2 could be tightly
regulated by a loop-hairpin switch conformation and an inhibitory glutamic acid
providing explanations for why the activity and substrate of Apo2 has not been
discovered.

42
Chapter 4
The Quest for APOBEC-2 Activity
The Apo2 physiological and in-vitro activity has, to date, remained an
“enigmatic conundrum”. A slew of possible explanations exist for how Apo2
continues to elude biochemists. Other APOBEC family members have been shown
to be regulated by several factors such as phosphorylation, co-factors, and inhibitors.
In addition, all APOBECs, whose activity has been revealed, have been shown to be
substrate specific and have tissue-specific expression. Therefore, we attempted to
elucidate the biochemical activity of Apo2 by investigating these potential regulatory
factors.
Apo2 and AID are the most evolutionary ancient enzymes among the
APOBEC family and have been shown to have originated at a similar time about 500
million years ago. Apo1 and the Apo3s arose from a gene duplication of AID
(Conticello et al., 2005). It seems plausible that Apo2, being one of the oldest family
members, may also be one of the most tightly regulated of the family allowing it to
have a very specific function; whereas, the more recently evolved APOBECs are
more “promiscuous”, allowing them to adapt for more diverse, alternative functions.
Owing to the fact that other APOBEC proteins are specifically regulated through
several mechanisms, we attempted to decipher the regulations of Apo2 activity.

43
1. APOBEC-2 Substrates
Because APOBECs display such a variety of deamination specificities, we
reasoned that Apo2 may have a preference for a specific cytidine. For this reason,
we attempted to tease out Apo2 activity using three different 80 nucleotide DNA
substrates, each scattered with abundant cytidines in different nucleotide sequence
contexts (see Chapter 7). As a control, all three of these substrates also contain
cytidines within AID and A3G hotspots. Several times we were convinced we had
detected Apo2 activity because of the presence of what we considered to be “ghost”
bands (see asterisk Fig 4-2a, Fig. 4-2b). However, these misleading bands turned out
to be false positives because the control lanes, without NaOH (Fig. 4-2a) or UDG
(data not shown), still contain the band. If the band was genuinely formed as a result
of deamination by Apo2, the band would disappear in the lane without addition of
UDG or NaOH. UDG is necessary to remove the uracil base and NaOH is necessary
to cleave the backbone of the DNA at the abasic site (Fig. 5-2a). Therefore,
unfortunately, Apo2 displayed no deamination activity on any of the substrates.
Even though we tested several substrates with numerous potential deamination sites,
it is possible that Apo2 deaminates a cytidine in a particular sequence context that
we did not test.
Full length Apo2 has a theoretical isoelectic point (pI) of 4.81 and truncated
Apo2 has a pI of 5.57; whereas, A3G and AID have pIs of 9.01 and 10.03,
respectively. All of the other APOBECs have theoretical pI’s of approximately 9-10.

a b

Figure 4-1. I saw a ghost! a. Deamination assay with 80-nt ss DNA (T2) and Apo2. A3G
and AID were run alongside as control. Red asterisk marks the “ghost” bands seen in Apo2
deamination assays. The misleading band is still seen with no addition of NaOH; therefore,
this band is considered a false positive. b. Collage of Dr. Ronda Bransteitter and soon-to-be
Dr. Courtney Prochnow celebrating Apo2 “activity”.

Because Apo2 has such a low pI compared with other APOBECs, we considered that
our buffer conditions for the deamination assays may not be ideal for the overall
negatively charged Apo2 to bind nucleic acid. Standard buffer conditions for AID
and A3G are at a pH of 8.0 which would give both proteins and overall positive
charge. For that reason, we tested for Apo2 deamination using different buffer
conditions ranging in pH. We reasoned that at a pH of 4, truncated Apo2, with a
theoretical pI of 5.57, may carry a net positive charge which may facilitate binding to
44
45
negatively charged nucleic acid. However, even under different buffer conditions,
Apo2 did not show any evidence of deamination activity.

2. Going “Fishing” for a Co-factor
Apo1 is the first and, so far, only APOBEC enzyme shown to require a
cofactor for deamination of its substrate (Kinter et al., 2000). ACF is the protein
recently discovered to increase Apo1 editing activity by helping to recognize the
mooring sequence flanking the edited cytidine. In addition, phosphorylation and
expression of ACF has been shown to regulate Apo1 editing. Since Apo1 requires a
co-factor for deaminase acitivty on it RNA substrate, we decided to try to “fish” out
an Apo2 co-factor in mammalian cells which may be necessary for Apo2
deamination activity.
FLAG-Apo2 was transiently transfected into 293T mammalian cells and
immunoprecipitated with anti-FLAG antibody. The protein was purified by gel
filtration (Superose 6) and the FLAG-Apo2 eluted in a LMM (low molecular mass)
fraction corresponding to a molecular weight of approximately that of an A3G
monomer (~45kD). Deamination assays were performed using Apo2 from whole
cell lysate and from immunoprecipitated fractions. No deamination was observed on
the T2 ssDNA substrate (see Chapter 7) used in the assay (data not shown). Future
experiments in skeletal and cardiac cells, where Apo2 RNA has been previously
shown to be expressed, are being designed. Perhaps these cells contain the necessary
co-factor that may be essential for Apo2 activity.

46
3. Does APOBEC-2 have an RNA Inhibitor?
Detecting AID activity proved to be difficult because it co-purified with a
bound RNA molecule that inhibited its deamination activity on ssDNA. Dr.
Bransteitter was the first to show directly that AID deaminates ssDNA with the
addition of RNase A (Bransteitter et al., 2003). Other APOBEC proteins also
associate with RNA molecules that inhibit their deamination activity (Chui et al.,
2005). A3G from 293T cells elutes in the void volume of the Superose 6 column in
a high molecular mass (HMM) complex. On addition of RNase A, the HMM A3G
complex dissociates and A3G elutes in later fractions corresponding to a low
molecular mass of an A3G dimer to monomer (Chui et al., 2005).
Based on these reports, we decided to express Apo2 in 293T cells and see if it
would associate with an RNA molecule. Although Apo2 did not elute in a HMM
complex, the protein did elute in fractions that correspond to a molecular mass that is
higher than a monomer. To test for RNA binding to Apo2 purified from 293T cells,
fractions that tested positive by immuno-blotting were collected, treated with RNase
A and re-purified by gel filtration. A slight shift to the right (decrease in molecular
mass) by one fraction was observed, possibly due to small fragments of RNA (data
not shown). Further tests are necessary to confirm this.

4. Is APOBEC-2 Regulated by Post Translational Modifications?
Another enticing consideration is that post translational modifications may
play important roles in regulating Apo2 activity. Phosphorylation has been shown to
be an important mechanism regulating APOBEC deamination activity. Previous
47
reports show that both AID and Apo1 activities are influenced by phosphorylation of
certain residues (Basu et al., 2005, Chen et al., 2001). AID has specific
phosphorylation sites that are important for regulating AID deamination activity
(Basu et al., 2005). Additionally, the editing activity of Apo1 is affected by altering
and mimicking possible phosphorylation sites (Chen et al., 2001). Earlier studies
have shown that mutations of serine to aspartate or glutamate and serine to alanine
mimic phosphorylation and dephosphorylation, respectively (Lu et al., 2002). In an
attempt to mimic phosphorylation states of Apo2, we designed a series of serine-to-
aspartate mutations at serine residues located in regions we predict to be potential
phosphorylation sites.
Apo2 has six residues in the vicinity of the active site that could potentially
be phosphorylated and affect deamination activity of the protein (Fig. 4-2). Residues
S124, S125, and S126 are located in the β3-h3 loop which is positioned near the
active center (Fig. 4-2). In all other APOBECs, these three residues form an SWS
motif where Apo2 residue S125 corresponds to a tryptophan. Previously it has been
suggested that this tryptophan is involved in contacting nucleic acid and helping to
position the substrate into the active center. Therefore, phosphorylation of S125 or
its neighboring residues, could affect Apo2 activity. We tested the single
phosphorylation mimic mutant S125D and the double mutant S125D/S126D. We
observed no deamination activity for both mutants tested (data not shown).
Additional Apo2 phosphorylation mutants and combination of mutants should be
tested for deamination.

Figure 4-2. APOBEC-2 potential phosphorylation sites near the active center. Model
of the structure of an Apo2 monomer with the h1/ β1 turn in the open hairpin conformation.
Potential phosphorylation sites are represented as sticks and include S124, S125, and S126
located on the β3-h3 loop proximal to the active center and Y61, S62 S63 located on the
regulatory h1/ β1 turn. A red sphere represents the active site zinc ion.

The h1/ β1 turn, which may act as a conformational switch as previously
described (Chapter 3), contains residues Y61, S62 and S63 which neighbor the
regulatory E60 (Fig. 4-2). In one conformation the turn forms a hairpin resulting in
an “open” conformation in which the active site appears to be open and accessible
for nucleic acid substrate (Fig. 3-7a). In the other conformation, the turn forms a
loop which collapses over the active site rendering it inactive and inaccessible to
nucleic acid substrate (Fig. 3-7b). These 3 potential phosphorylation sites may
potentially play a role in regulating the different conformations depending on
48
49
phosphorylation. Therefore, we designed two mutants, S63D and S62D/S63D, to try
to mimic phosphorylation at these sites. Neither of these Apo2 mutants displayed
deamination activity (data not shown). It is possible that we did not create the
correct phosphorylation mimic mutation or combination of mutations in order to
affect Apo2 activity to observe deamination. More tests and mutations should be
designed in order to determine clearly whether phosphorylation plays a role in Apo2
activity.
Higher eukaryotes, unlike E.coli, have the ability to perform post
translational modifications (phosphorylation, glycosylation, acetylation,
myristylation) that may regulate protein activity. For this reason, we expressed
GST-tagged full length and trApo2 in SF9 insect cells and FLAG-tagged trApo2 in
293T cells to determine if post translational modifications play a role in regulating
Apo2 activity. Similar to expression levels in E. coli, Apo2 expressed at high levels
in insect cells and was purified using the same glutathione purification protocol used
for trApo2 expressed in E. coli. We performed deamination assays with insect cell
expressed Apo2 using the three DNA substrates described previously (T1, T2, T5)
and still observed no deamination (data not shown). The FLAG-tagged trApo2 was
immunoprecipitated with an anti-FLAG antibody as previously described and
similarly tested for deamination activity. It was also inactive on the T1, T2 and T5
substrates.

50
5. The Conserved SWS Motif
As discussed above, the Apo2 S125 residue corresponds to a tryptophan that
is located in the SWS motif conserved in all other APOBECs. Because of its
location on a loop proximal to the active center, it has been suggested that this
tryptophan helps to position nucleic acid into the active site. Previous evidence
suggests that this tryptophan is necessary for APOBEC activity. In Apo1, mutation
of this tryptophan to a serine renders the protein inactive (Navaratnam, 1998). Also,
this is a hyper-IgM mutation in AID (Durandy, 2006).
Because other APOBECs require the SWS motif and Apo2 is the only
APOBEC with a serine at this position, we mutated the Apo2 serine to a tryptophan
to investigate whether this change would restore Apo2 activity (Fig. 4-2). We tested
this mutant for deamination activity using the T1, T2 and T5 substrates, however, the
Apo2 S125W mutant was still inactive. Perhaps a specific pattern of
phosphorylation of the three serines is required to stimulate Apo2 activity.

6. Inhibition by the Regulatory Loop or Glutamic Acid Residue?
The previously described “built in” conformational switch may regulate
Apo2 activity by coordinating zinc in two different manners (see Chapter 3). In the
Apo2 tetramer structure, the two inner monomers contain the L1 loop conformation
which appears to completely inactivate the protein by excluding a water molecule
(Fig. 3-7b). Although the outer monomers in the tetramer appear to be open and
accessible to substrate, this structure is a static “picture” of the protein and we do not
actually know if these loops always remain open. It is conceivable that Apo2

a b

Figure 4-3. The L1 loop on APOBEC-2 contains the regulatory E60. The structure of an
Apo2 monomer with a “closed” L1 loop. a. The tApo2 monomer contains the regulatory
switch loop (shown in blue). The glutamate 60 residue (shown in green) forms a forth ligand
with the zinc (represented by a red sphere) inactivating this active site. b. tApo2 model
showing the regulatory loop removed.

regulates protein activity by keeping these loops closed until it becomes “activated”
perhaps through post translational modifications, co-factor, etc. To determine if the
L1 loop is negatively regulating Apo2 activity, we designed a truncation that
completely removes the N-terminal region of the Apo2 protein (amino acids 127-
224), including the h1/ β1 turn containing the L1 loop (Fig. 4-3). This construct
displayed no deaminase activity (data not shown).
The regulatory glutamate residue (E60) located on the h1/ β1 turn is unique to
Apo2, whereas all other APOBEC enzymes have a conserved arginine residue at or
near this position. In the inner monomers of the Apo2 tetramer crystal structure, the
51
52
E60 residue forms a fourth ligand with zinc which consequently, results in exclusion
of the water molecule, thereby rendering those active sites completely inactive
(Chapter 3, Fig. 3-6b) (Fig. 4-3). Perhaps this negative regulation is exclusive to
Apo2. In an attempt to reverse this possible negative regulation and detect Apo2
activity, we mutated the glutamate 60 residue to an arginine and tested the mutant for
deamination activity. We observed no deamination activity for this mutant.

7. APOBEC-2 DNA Binding
Apo2 has previously been shown to bind apoB mRNA, however, there is no
evidence that it binds DNA. Clearly, for deamination to occur a deaminase must
actually bind the target substrate to be deaminated. We performed several DNA
binding tests to determine whether Apo2 binds ssDNA (see Chapter 7). Native gel
experiments with high concentrations of full length Apo2 and tApo2 display one
prominent band suggesting the presence of an Apo2 monomer. We performed native
gel experiments with Apo2 with and without fluorescein tagged DNA to see if we
could detect a shift in protein mobility which would suggest DNA binding (see
Chapter 7). Upon addition of DNA, we did not see a protein shift (Coomassie stain)
nor did we observe the fluorescein tagged DNA in the native gel (data not shown).
Because of the low pI of Apo2 (see Chapter 2), in addition to standard APOBEC
DNA binding conditions (pH 7-8) , we tried several different buffer conditions to try
to ensure that the protein would have an overall positive charged to bind to
negatively charged nucleic acid (pH 4,5,6). We also tried different lengths of
53
ssDNA (50mer, 80mer). Nevertheless, we did not observe any DNA binding for full
length Apo2 or tApo2 (data not shown).

8. Discussion
Previous attempts to detect Apo2 activity have shown that Apo2 deaminates
free cytidine and deoxycytidine in-vitro (Liao et al., 1999; Anant et al., 2001);
however, follow-up studies showed the presumed Apo2 deamination activity was
actually the result of a contaminant E. coli cytidine deaminase (Mikl et al., 2005).
Because APOBECs are regulated by several mechanisms and the biological function
of Apo2 is still unknown, we attempted to identify Apo2 substrates and deamination
activity using various methods. Our exhaustive efforts did not produce any evidence
that Apo2 deaminates nucleic acids substrates.
Despite the fact that we did not show deamination activity, Apo2 may still be
able to deaminate a still unidentified and highly specific nucleic acid substrate.
Another plausible theory could be that Apo2 may act on microRNA substrates that
regulate the expression of genes.

54
Chapter 5
Structure Modeling of AID and Implications for AID
Biochemical Activity
Based on the close sequence homology between APOBEC proteins, we tested
Apo2 structure-based predictions using activation-induced cytidine deaminase
(AID). We show that AID deamination activity is impaired by mutations predicted
to interfere with oligomerization and substrate access. Moreover, the structure
suggests how mutations in HIGM-2 patients inactivate AID resulting in defective
antibody maturation.

1. APOBEC-2 and AID Conservation of Sequence and Structure
Apo2 shares a 30 % amino acid sequence identity with AID. Additionally,
the buried residues in Apo2 are highly conserved sharing a 75% sequence identity
(96% homology) with AID (Fig. 5-1a). A large majority of the Apo2 residues
important for dimerization and tetramerization align with AID residues. These
residues share sequence identity or have similar hydrophobic, polar and charged side
chains (Fig. 5-1a). The highly conserved residues buried inside the structure and
located at the observed dimeric/tetrameric interfaces strongly suggest a structural
conservation of AID with Apo2.
Previous studies also support significant structural similarity between the two
enzymes. Residues G47-G54 appear necessary for AID dimerization (Wang et al.,
2006). These residues align with the dimerization sequences of Apo2 (Figs. 3-3a and
a

b c

Figure 5-1. Structural guided mutagenesis of AID impairs deamination activity. a.
Structure guided sequence alignment of Apo2 and AID show that the two proteins are highly
conserved. Red letters indicate identical residues between the two proteins; grey shadows
indicate the buried residues; red squares represent the active center residues; green dots are
above the residues in the tetramer interface; blue diamonds are above the residues at the
dimer interface; purple stars are below the residues that are mutated in AID from HIGM
patients; black triangles are below the residues of AID that were mutated based on the Apo2
structure. b. Mutated AID residues are modeled onto the Apo2 structure (drawn in green) at
the dimer interface. Mutations of these residues (drawn in green) disrupted the deaminase
activity of AID. c. Mutated AID residues at the tetramer interface are modeled onto the
Apo2 structure (drawn in green). Mutations of these residues significantly impair AID
deamination activity.

5-1a). Point mutation of R50, N51, K52, and G54 to alanine in AID results in a
significant reduction in dimer formation (Wang et al., 2006), which is consistent with
their alignment with Apo2 residues near the C terminal region of the β2 strand (Apo2
residues E91, D92, E93 and A95 respectively) (Fig. 5-1a). Based on the Apo2
55
56
structure, the corresponding point mutations to alanine likely reduce van der Waals
interactions and hydrogen bonding with water molecules at the dimer interface (Fig.
5-1a). These interactions are needed for dimer stabilization. The Apo2 crystal
structure also predicts that residues involved solely in main-chain hydrogen bonds at
the dimer interface, with no contributions from their side-chain interactions, should
not disrupt AID dimerization. This is consistent with data showing that AID
mutations of this type (S43, L44, D45), which align with Apo-2 β-2 strand residues
(Q84, A85, S86), do not disrupt AID dimerization (Wang et al., 2006).

2. AID Mutants as a Functional Surrogate
The APOBEC proteins AID and A3G processively deaminate C → U on
ssDNA favoring specific sequence motifs (Pham et al., 2003; Chelico et al., 2006).
Although extensive sequence conservation among APOBEC proteins strongly
suggests that Apo2 should be active on either ssDNA or RNA, deamination activity
has not been observed (Mikl et al., 2005). Specific modifications such as
phosphorylation or a yet-to-be identified co-factor or nucleic acid target motif may
be required for Apo2 deamination activity. Nevertheless, there are valuable
functional insights that can be derived by comparing Apo2 with
other APOBEC proteins. Here we use AID as a surrogate to test how Apo2-
structurally guided mutations affect AID deamination activity.
To investigate how oligomerization affects C deamination activity, AID
amino acids that align within the Apo2 residues at the tetramerization and
dimerization interface were targeted for mutation (Figs. 5-1a, 5-2b). GST fusions of
57
wild type and mutant AID proteins were generated in E. coli. The wild type GST-
AID deaminates the cytosine within the AID hotspot sequence motif (5’ WRC 3’) on
ssDNA (Pham et al., 2003) with a specific activity of 2.3 pmol/µg/hr (Figs. 5-2a, 5-
2b). However, the AID mutants with amino acid substitutions located at the
tetrameric interface showed either no detectable deaminase activity or significantly
reduced activity (Fig. 5-2b). Mutants R112C and Y114A/F115A were inactive, while
mutants K16A and C116A had a 3.3-fold reduction in activity (Fig. 5-2b). These
AID mutational results suggest that AID tetramerization is important for deamination
activity. Since all the residues located at the tetramer interface are also present on the
exposed surface on the two outer monomers of a tetramer, it is also possible that
these residues on the exposed side could play a role in AID activity.
The AID mutations within the predicted dimerization domain also indicate
the importance of dimerization for AID activity. The AID double mutant,
F46A/Y48A, displayed a 4-fold decrease in deamination activity (Fig. 5-2b). The
alanine substitutions do not affect the main chain interactions between the β2 strands
but will disrupt hydrophobic side chain interactions with h5 and the loop N-terminal
to h1 (Fig. 5-2b). The dimer interface is extensive; therefore, two mutations should
not completely disrupt dimeric AID, which explains why weak deamination activity
was observed with this double AID mutant. This mutational result, as well as that
reported previously (Wang et al., 2006), suggests the relevance of the dimeric
interface in Apo2 structure.
When Apo2 E60 forms the fourth coordination bond with Zn, Zn
hydroxylation does not occur and the h1/ β2-turn is folded into the loop
a

b

Figure 5-2. Mutant AID deamination activity. a. A sketch describing the cytidine
deamination assay. F, fluorescein; UDG is uracil DNA glycosylase. b. Bar representation of
the specific activities for wt and mutant AID proteins. inset. Denaturing PAGE analysis of
the deamination activity for wt and mutant AID proteins. The 30-nucleotide (nt) band
indicates deamination activity.

conformation, which can block the active center. We reason that this E60 should
have a negative regulatory role for Apo2 activity. AID has an Arg (R19) at the
equivalent position (Fig. 5-1a). To investigate the functional role of this residue
in AID, we generated an AID R19E mutant. As predicted from the structure, this
mutant showed a significantly decreased deamination activity (about 4.6-fold less
than the wild type, Fig. 5-2b).
58
59
To investigate the necessity of the β1’-hairpin conformation that facilitates
the open access of the active site to nucleic acids substrates, we generated an AID
mutant with an R24E mutation. The R24 residue is equivalent to the R65 in Apo2,
which interacts with Y61 to stabilize the β1’-hairpin conformation. We predicted
that the disruption of the Y61/R65 interaction would collapse the β1’-hairpin, block
substrate access to the active sites and impair deamination activity. Indeed, the AID
R24E mutant was completely inactive on ssDNA (Fig. 5-2b).

3. APOBEC-2 Structural Insights for Hyper-IGM-2 Immunodeficiency
Mutations in the gene encoding human AID are known to cause Hyper-IgM-2
(HIGM-2) syndrome characterized by defective isotype-switching and somatic
hypermutation, which are required for the production of high affinity antibodies
(Durandy et al., 2006; Minegishi et al., 2000). The mutated AID residues of HIGM-
2 patients are highly conserved in Apo2 (Fig. 5-3a). A likely explanation for why
and how HIGM-2 mutations disrupt AID function is given by the structure of Apo2
(Figs. 5-3b-d). Based on the crystal structure, HIGM-2 AID mutations can be
divided into four classes. The first class occurs at the tetramerization interface (Fig.
5-3b), which contains A111, R112, L113, and N168. All are expected to affect AID
tetramerization. The second mutant class includes residues in and near the active
center (Fig. 5-3c), H56, E58, S83, S85 and C87, which are conserved among all
APOBEC enzymes. The AID R24 residue is also mutated in HIGM-2 patients. As
previously discussed, the role for R24 is to stabilize the β1’-hairpin, which keeps the
active site open for DNA or RNA access (Fig. 3-7). The third class consists of
a

b c

d

Figure 5-3. AID HIGM-2 mutations. a. Alignment of mutated residues of AID from
HIGM-2 patients with the corresponding residues in APO2, showing high sequence
conservation. b. Mapping the residues in AID HIGM-2 mutations, R112, L113, N168, to the
tetramer interface as modeled from the APO2 structure. c. Mapping the AID HIGM-2
mutations, E58, H56, S83, S85, C87, R24, near the active site. d. Mapping of AID HIGM-2
mutations, W80, L106, M139 and F151, to the interior core structure of an AID dimer.

60
61
residues located on the enzyme surface (Fig. 5-3c). The residues located at the
tetramer interface (A111, R112, L113, N168) are also exposed on the surface of the
outer ends of the tetramer (Fig. 5-3c). The presence of these amino acids on the
exposed surface of the enzyme might presage a role beyond that required for
tetramerization. A fourth class of HIGM-2 AID mutations are those with large
hydrophobic side chains buried deeply within the core structure (Fig. 5-3d),
including W80, L106, M139, and F151. Three of these residues (W80, L106, and
M139) are located near the active center. Mutating these residues should disrupt the
folding and stability of AID.

4. Discussion
Since many of the APOBEC enzymes are reported to form dimers and
multimers, the APO2 structure may shed light on how these enzymes oligomerize.
The X-ray structure of Apo2 provides a clear direction to pursue future functional
studies of APOBEC proteins with a eye toward developing therapeutic strategies to
deal with problems arising from either the failure to deaminate C, or from the
potentially more serious problem of deaminations occurring at inappropriate times
and places leading to deleterious mutations causing disease (Pham et al., 2005).

62
Chapter 6
Comparative Modeling of APOBEC3G and Implications
for other APOBECs

Earlier attempts at predicting APOBEC protein structure was performed through
comparative modeling of Apo1, AID and A3G with, at the time, the only available
fntCDA structures as a guide (Xie et al., 2004, Navaratnam et al., 1998). Now a
high-resolution crystal structure of an APOBEC protein, Apo2, is available to use for
a more accurate model of other APOBEC proteins (Prochnow et al., 2007).
Although the cytidine deaminase active sites appear similar to the fntCDA structures,
the differences include oligomerization and substrate binding sites (see Chapter 3).
After modeling AID, we chose to model an APOBEC protein with a double
deaminase domain, A3G. Using the more appropriate and homologous Apo2
tetramer crystal structure as a template, we modeled an A3G monomer and dimer.
To better understand A3G both from a structural and functional perspective, we
mapped residues onto the model that have been previously shown to be important for
A3G function. Furthermore, this structural model of A3G can also be beneficial for
use in rational structure-based drug design for developing therapeutics to inhibit VIF
mediated degradation. In addition to A3G, because of their close sequence
homology to Apo2, all APOBEC family members can potentially be modeled in a
similar manner by using the Apo2 structure as a template.

1. A3G Modeling
Truncated Apo2 (trApo2) crystallized as four monomers in an asymmetric unit
forming what appears to be a tetramer (Fig. 3-1a). Previously, through structure
guided mutagenesis studies using AID, we identified residues in both the dimer and
tetramer interface that could potentially affect oligomerization (see Chapter 5).
Apo2 and AID are single domain deaminases, whereas A3G is a double domain
deaminase. Therefore, the trApo2 (residues 41-224) sequence was aligned with both
the N-terminal catalytic domain (CD1) and C-terminal catalytic domain (CD2) of

Figure 6-1. Sequence alignment of truncated Apo2 and A3G cytidine deaminase
domains (CD1, CD2). Structure guided sequence alignment of Apo2 and A3G domains
showing significant sequence homology. The secondary structure of Apo2 is represented
with black cylinders (helices) and black arrows (β−strands). Red letters indicate identical
residues between proteins; red squares indicate the active center residues; green dots
indicate the Apo2 tetrameric interface residues; magenta diamonds indicate Apo2
dimeric interface residues.
63
64
A3G (Fig. 6-1). TrApo2 shares a 31% identity (43% homology) and 33% identity
(46% homology) with the CD1 and CD2 of A3G, respectively. Additionally, the
buried residues in Apo2 are highly conserved sharing a 59% identity (83%
homology) and 52% identity (83% homology) with the CD1 and CD2 of A3G,
respectively. The highly conserved buried residues strongly suggest a structural
conservation of both domains of A3G with Apo2. Therefore, we predict that the two
catalytic domains of an A3G monomer would fold in a similar manner as does an
Apo2 dimer, through the long β2 strands (Fig. 3-3a, Fig. 6-2a). In order for the two
domains to interact, helix 6 of the CD1 domain would need to connect to helix 1 of
the CD2 domain. There are 14 residues available between the domains that could
form the predicted flexible loop necessary for this connection.
It is possible for A3G to dimerize in a manner that would resemble the
tetramer of Apo2. The elongated organization of the subunits seen in our A3G
model is similar to previously published small x-ray scattering data (SAXS) of an
A3G dimer (Wedekind et al., 2006). Since residues located in the tetramer interface
of Apo2 would potentially correspond to the dimer interface of A3G, we studied the
alignment to search for similarities and differences within these potential interface
residues (Fig. 6-1). Interestingly, overall, the enzymatically inactive CD1 domain of
A3G aligns slightly better with trApo2 than does the catalytically active CD2 domain
of A3G (Fig. 6-1). Apo2 has an extensive tetrameric interface consisting of
hydrophobic, charged and polar residues which are all involved in the interface
interactions. The Apo2 tetramer interface residues are more homologous to the
residues located in the A3G CD1 domain than the A3G CD2 domain. The A3G CD2
domain lacks the major hydrophobic residues which appear to be crucial for Apo2 to
form a tetramer, whereas the A3G CD1 domain has conserved these residues (Fig. 6-
a

b

Figure 6-2. Model structure of an A3G monomer and dimer based on the Apo2
tetramer crystal structure. a. Model of an A3G monomer with the non-catalytic CD1 and
catalytic CD2 labeled. The CD1 subunit was modeled using an Apo2 monomer with a
“closed” L1 loop whereas the CD2 subunit was modeled using an Apo2 monomer with an
“open” β1’-hairpin conformation. The red sphere represents the zinc ion present in the
active sites of both subunits. Residues involved in the A3G dimer interface interactions are
displayed on CD1 as sticks in the model. b. A3G is modeled as a dimer though head-head
CD1 interactions. The red sphere represents the zinc ion in all four A3G cytidine deaminase
domain active sites.

1). These A3G residues include: R122, Y124, Y125, F126, W127 and D128 (Fig. 6-
3). Note that these residues create a very hydrophobic area on the surface of the
65
66
protein. For this reason, using the Apo2 structure as a guide, we modeled an A3G
dimer with the dimeric interface joining though monomer head-to-head CD1
interactions (Fig. 6-2). In this conformation, the hydrophobic residues residing near
the surface of the molecule would pack together with the same molecules on the
other CD1 monomer.
As discussed previously (see Chapter 3), the Apo2 h1/β1 hairpin-loop switch
may act as a molecular mechanism for regulating substrate access and enzyme
activity mediated through zinc coordination. In the crystal structure of Apo2, the
inner two monomers, positioned in the tetramer interface, appear to have active sites
which are “turned off” by the collapsed L1 loop. In contrast, the two monomers
positioned on the outside of the tetramer appear to be “turned on” with the B1’
hairpin rotated in an open conformation (Fig. 3-4a). Interestingly, this active site
organization in Apo2 parallels that shown in our A3G dimer model (Fig 3-4a). In
this model of the A3G dimer, the two A3G CD1 active sites, which have previously
been reported to be catalytically inactive (Navarro et al., 2005), are located at the
interface of the dimers. The functional, catalytically active, CD2 active sites are
situated on the outer edges of the dimer.
Although A3G does not contain a glutamate in its sequence corresponding to the
regulatory E60 residue of Apo2, we modeled the A3G h1/β1 turn as a closed L1 loop
in the A3G dimer. Our reasoning is that the monomer contacts within the dimer
interface may force the h1/β1 loop to remain closed, subsequently inhibiting
substrate access thereby deactivating the active site. Why the A3G CD1 domain
remains catalytically inactive is still unknown, hence, this “closed loop” model may
67
provide a possible explanation. On the other hand, because A3G CD1 has an
arginine instead of a glutamate near the corresponding Apo2 E60 position, it is also
reasonable to model the h1/β1 turn in the open β1’-hairpin conformation (Fig. 6-2b).
In this representation, the CD1 active site would be “open” and accessible to nucleic
acid. In this case, an explanation for CD1 inactivity other than restricted substrate
access would have to be pursued.

2. Mapping of A3G functional residues to the dimer interface
In order to gain more insight into A3G function from the model, we mapped
residues that have previously been shown to be important for A3G function onto the
A3G dimer model. Residues located in the modeled A3G dimer interface have
previously been shown to be important for different A3G activities such as
dimerization, virion incorporation, and HIV-1 inhibition.
Unpublished data presented at the 2007 Cold Springs Harbor Retrovirus
Conference from Dr. Pathak’s lab showed that the W127 residue of A3G is
important for dimerization. Co-immunoprecipitation assays using wild-type A3G
and mutant A3G (W127A) demonstrated that the mutant oligomerization is impaired.
The W127A mutation does not co-immunoprecipitate compared with wildtype A3G
(data not published). This evidence supports our head-head modeling of an A3G
dimer through the CD1 region because the W127 residue lies specifically in the
proposed dimeric interface (Fig. 6-3). Based on the Apo2 tetramer structure, this
tryptophan appears to play a key role in tetramerization by providing a bulky
aromatic side chain for hydrophobic interactions within the interface. Additionally,
this is one of the hydrophobic residues A3G CD1 domain shares with Apo2 that the

Figure 6-3. Model structure of an A3G dimer interface joining between two CD1
subunits. A3G residues in the CD1 proposed dimer interface that may be important for
dimerization. The A3G CD1 contains the major hydrophobic residues (Y125, Y126, W127)
which appear to be crucial for Apo2 to form a tetramer. These residues, including R122
have previously been shown to be important for HIV-1 virion incorporation, HIV-1
inhibition and dimerization. D128 is crucial for species specific VIF recognition of A3G.

CD2 domain has not conserved. Therefore, this data corroborates our model of an
A3G dimer formed though head-head interactions.
68
A recent biochemical study specifically examined the residues located on the β4-
h4 loop in A3G which, in the Apo2 structure, is the main region involved in
tetramerization (Huthoff and Malim, 2007) (Fig. 6-3). Interestingly, several of these
residues were shown to be important for A3G biological function such as virion
69
incorporation and HIV-1 inhibition. Specifically, constructs with mutations at
residues Y124A and W127A were most affected showing severe defects in virion
incorporation and loss of antiviral activity. Mutants Y125A and F126A showed
intermediate levels of disruption of virion incorporation compared to wildtype A3G
(Huthoff and Malim, 2007). It is interesting that antiviral activity is affected in these
mutants because the mutations are in the CD1 of A3G which is catalytically inactive
(Fig. 6-3). Therefore, it is possible that the mutations are affecting dimerization
which is necessary for antiviral activity. More tests are necessary in order to confirm
this hypothesis.
A3G restricts HIV-1 replication in the absence of HIV VIF (see Chapter 1).
However, in a wildtype HIV infection VIF targets A3G for ubiquitylation and
proteasome-mediated degradation. Several studies have shown that the A3G residue
D128 plays an essential role in VIF species specific recognition of A3G. When the
D128 residue is mutated to a lysine, A3G becomes resistant to VIF mediated
degradation. Interestingly, this residue is located within the predicted dimer
interface. This brings to light a very important question of whether VIF recognizes
an A3G dimer, multimer or a monomer.

5. Discussion
Besides A3G and AID, other APOBEC enzymes can also be modeled using the
Apo2 structure as a template. A comparison of the Apo2 structure could help reveal
why deamination activity has not been observed for Apo2 and possibly other
APOBEC enzymes such as APOBEC3A and APOBEC4. Structural models of the
70
APOBEC3 proteins can facilitate our understanding of how these enzymes can
restrict viruses and retroelements. Ideally, we hope to use the crystal structure and
APOBEC3 protein models to determine how to protect A3G and A3F from VIF
targeted proteasome-mediated degradation. Such insights could lead to cures and
preventive treatments for HIV infection.
Interestingly, during the process of writing this thesis, we solved the crystal
structure of CD2, the catalytically active domain of A3G. Notably, the core
structural fold of this domain is essentially identical to that of Apo2.

71
Chapter 7
Experimental Procedures
1. Cloning and Purification of APOBEC-2
HsApobec-2 containing residues 41-224 was cloned and expressed and
purified in Escherichia coli as a recombinant glutathione S-transferase (GST) fusion
protein. The protein was purified using a glutathione affinity chromatography
column and the GST tag was cleaved with thrombin. Three extra glycines and a
methionine were cloned to the N-terminus of the protein for more efficient thrombin
cleavage and selenium methionine (S-Met) incorporation, respectively. Further
purification of Apobec-2 was accomplished through Superdex-75 gel filtration
chromatography in a buffer containing 25mM Hepes, pH 7.0, 50mM NaCl, 10mM
dithiothreitol. The Se-Met labeled protein was prepared as described above except
cultures were grown using minimal media. The replacement of methionine by Se-
Met was verified by mass spectroscopy. Native and Se-Met labeled protein was
concentrated to 15mg ml -1 and crystallized at 18 ºC by the hanging drop vapor-
diffusion method from a reservoir solution of 85mM NaCitrate, pH 5.6, 160mM
LiSO4, 24% (wt/vol) polyethylene glycol monomethyl ether and 15% glycerol.

2. Structure Determination and Refinement of APOBEC-2
Native and Se-MAD data were collected at synchrotron and processed using
HKL2000 (Table 1). An initial solution was obtained at 3.5 Å using the program
72
Solve to locate eight selenium atoms using the peak wavelength data set. A search
with SHELXD found four additional seleniums (totaling twelve) and subsequently
the program SHARP identified an additional four weaker anomalous scattering
atoms, which later were recognized as Zinc atoms. Density modification and four-
fold NCS averaging were applied using the program RESOLVE. In addition, phase
extension was performed with the native data set in RESOLVE to 2.5Å resolution
using the two-wavelength MAD phases calculated in SHARP. The molecular model
was built based on this experimental map using the program “O” and was refined
with CNS. Two-fold non-crystallographic symmetry (NCS) constraint was applied
during the initial simulated annealing, but the final refinement was carried out
without NCS constraint. The protein geometry is excellent when examined using the
program PROCHECK (Table 3-1).

3. APOBEC-2 Glutaraldehyde Crosslinking
Reactions with Apo-2 (2ug) were performed for the indicated time with
0.25% glutaraldehyde at room temperature and quenched with 1M Tris, pH 8.5, 2X
SDS loading buffer and run on a 12% SDS PAGE gel for 70 min at 200V. Lanes
marked M and 0 were analyzed by Coomassie staining and lanes marked 5, 15, 30
and 60 were analyzed by silver staining.

4. Construction of AID Mutants
Mutant AID proteins (R112C, Y114A/Y115A, E117A/E118A, K16A,
C116A, R19E and F46A/Y48A) were constructed by site-directed mutagenesis using
73
the pGEX-KG-AID vector as the PCR template. The following primers and their
complementary strands were used: 5’-CTG AGG ATC TTC ACC GCG TGC CTC
TAC TTC TGT GAG GAC-3’ (R112C), 5’-ATC TTC ACC GCG CGC CTC GCC
GCC TGT GAG GAC CGC AAG GCT-3’ (Y114/Y115), 5’-GCG CGC CTC TAC
TTC TGT GCG GCC CGC AAG GCT GAG CCC GAG-3’ (E117/E118A), 5’-
AAG TTT CTT TAC CAA TTC GCA AAT GTC CGC TGG GCT AAG-3’
(K16A), 5’-ACC GCG CGC CTC TAC TTC GCT GAG GAC CGC AAG GCT
GAG-3’ (C116A), 5’-TAC CAA TTC AAA AAT GTC GAG TGG GCT AAG GGT
CGG CGT-3’ (R19E), and 5’-ACA TCC TTT TCA CTG GAC GCT GGT GCT
CTT CGC AAT AAG AAC GGC-3’ (F46A/Y48A). The entire coding region of
AID in the mutant constructs was verified by DNA sequencing.

5. Deamination Reactions
In 20-µl reactions, AID, Apo2 or mutant Apo2, was incubated with RNase A
(0.1 µg), Uracil DNA Glycosylase (UDG) (6 units), and an FdT incorporated ssDNA
substrate (50nM) in a reaction buffer (10 mM Tris/HCl, pH 8.0/ 1 mM EDTA/ 1 mM
DTT) at 37°C for 1 hour. Reactions were terminated by extracting twice with
phenol/chloroform/isoamyl alchohol (25:24:1). NaOH was added to a final
concentration of 0.2 N and incubated at 90°C for 7 minutes. An equal volume of
95% formamide/20 mM EDTA was added to the reaction. Reaction products were
resolved with 16% denaturing PAGE and visualized on a BioRad FX-scanner. The
ssDNA substrate was labeled by incorporation of an FdT base during synthesis on a
BioRad DNA synthesizer. For AID reactions, the sequence of the ssDNA substrate
74
is: 5’ taa agg FAMdTga aga gag gag aga gaa gta agC tga aga gag aga agg aag aga gtg
aag gag 3’. For Apo2 deamination reactions, several DNA substrates were tested.

6. APOBEC-2 DNA Deamination Substrates
DNA substrates used for Apo2 deamination were labeled by incorporation of a
FAMdT base 5 bases from the 5’ end. The DNA substrates are as followed:
T1: 5' GTG AFAMdTG AGA AAG CTG TTG CCC GTC TCA CTG AAA CCA
CCC TGG CGC CCA ATA CGC AAA CCG CCT CTC CCC GGT GAT GAG AA
3'
T2: 5' GTG AFAMdTG AGA AGA TTC ATT AAT GCA GCT GGC ACG ACA
GGT TTC CCG ACT GGA AAG CGG GCA GTG AGC GCA AGT GAT GAG
AA3'
T5: 5' GTG AFAMdTG AGA AAA CCC TGG CGT TAC CCA ACT TAA TCG
CCT TCG AGC ACA TCC CCC TTT CGC CAG CTG GTA GGT GAT GAG AA
3'

7. Comparative Modeling of A3G
Sequence alignments of truncated Apo2 with both A3G N and C-terminal
domains were performed using MultAlin (Corpet, R, 1998). The comparative model
of A3G was built using 3D-Jigsaw, an automated system used to build three-
dimensional models for proteins based on homologues of known structure (Bates,
1999). Sequences of both domains of A3G that aligned with Apo2 were inputted
into 3D-Jigsaw using the interactive mode. Apo2 monomers A and C (the inner
75
monomers) were selected as templates to model A3G N-terminal domains and Apo2
monomers B and D (outer monomers) were selected as templates to model A3G C-
terminal domains. All modeling was performed using the program Pymol (DeLano,
W., 2002).

76
References

Anant, S., N. Davidson, et al. (2003). “Hydrolytic nucleoside and nucleotide
deamination, and genetic instability: a possible link between RNA-editing enzymes
and cancer?” Trends in Molecular Medicine 9:147-152

Anant, S., D. Mukhopadhyay, et al. (2001). "ARCD-1, an apobec-1-related cytidine
deaminase, exerts a dominant negative effect on C to U RNA editing." American
Journal of Cell Physiology 281: C1904-16.

Bates, P.A. and M. Sternberg. (1999) "Model Building by Comparison at CASP3:
Using Expert Knowledge and Computer Automation." Proteins: Structure, Function
and Genetics, Suppl 3:47-54.

Betts, L., S. Xiang, et al. (1994). "Cytidine deaminase. The 2.3 A crystal structure of
an enzyme: transition-state analog complex." Current Biology 235: 635-656.

Bishop, K. N., R. Holmes, et al. (2004). "APOBEC mediated editing of viral RNA."
Science 305, 645.

Bödör, C., A. Bognár, et al. (2005). "Aberrant somatic hypermutation and expression
of activation-induced cytidine deaminase mRNA in mediastinal large B-cell
lymphoma." Journal of Haematology 3:373-6.

Bogerd, H., H. Wiegand, et al. (2006). "APOBEC3A and APOBEC3B are potent
inhibitors of LTR-retrotransposon function in human cells." Nucleic Acids Res.
1:89-95.

Bonvin, M., F. Achermann, et al. (2006). "Interferon-Inducible Expression of
APOBEC3 Editing Enzymes in Human Hepatocytes and Inhibition of Hepatitis B
Virus Replication." Hepatology 43: 1364-1374.

Bransteitter, R., P. Pham et al. (2004). "Biochemical analysis of hyper-mutational
targeting by wildtype and mutant AID." Journal of Biological Chemistry 279:
51612-51621.

77

Bransteitter, R., J. L. Sneeden, et al. (2006). "First AID (Activation-induced Cytidine
Deaminase) Is Needed to Produce High Affinity Isotype-switched Antibodies."
Journal of Biological Chemistry 281: 16833-6.

Cascalho, M. (2004). "Advantages and Disadvantages of Cytidine Deamination."
The Journal of Immunology 172: 6513–6518.

Chelico, L., P. Pham, et al. (2006). "APOBEC3G DNA deaminase acts processively
3' --> 5' on single-stranded DNA." Nat Struct Mol Biol. 5:392-9.

Chester, A., J. Scott, J. et al. (2000). "RNA editing: Cytidine to uridine
conversion in apolipoprotein B mRNA." Biochem. Biophys. Acta 1494: 1–13.

Chiu, Y. L., V. Soros, et al. (2005). "Cellular APOBEC3G restricts HIV-1 infection
in resting CD4+ T cells." Nature 7038:108-14.

Chiu, Y. L. and W. C. Greene (2006). "Multifaceted antiviral actions of APOBEC3
cytidine deaminases." Trends in Immunology 27: 291-7.

Chung, S. J., J. C. Fromme, et al. (2005). "Structure of human cytidine deaminase
bound to a potent inhibitor." Journal of Medicinal Chemistry 48: 658-660.

Corpet, R. (1998). "Multiple sequence alignment with hierarchical clustering".
Nucl. Acids Res. 22: 10881-10890.

Conticello, S. G., C. J. F. Thomas, et al. (2004). "Evolution of the AID/APOBEC
Family of Polynucleotide (Deoxy)cytidine Deaminases." Molecular Biology and
Evolution 22: 367-377.

Cullen, B. R. (2006). "Role and Mechanism of Action of the APOBEC3 Family of
Antiretroviral Resistance Factors." Journal of Virology 80: 1067-1076.

DeLano, W.L. (2002). "The PyMOL Molecular Graphics System." DeLano
Scientific
78
Dutko, J., A. Schäfer, et al.(2005). "Inhibition of a yeast LTR retrotransposon by
human APOBEC3 cytidine deaminases." Current Biology 7:661-6.

Durandy, A., S. Peron, et al. (2006). "Hyper-IgM syndromes." Current Opinion in
Rheumatology 18: 369-76.

Franca, R., S. Spadari, et al. (2006). "APOBEC deaminases as cellular antiviral
factors: A novel natural host defense mechanism." Med Sci Monit 12: RA92-RA98.

Greco O, G. Dachs. (2001). "Gene directed enzyme/prodrug therapy of cancer:
historical appraisal and future prospectives." Journal of Cellular Physiology 1:22-36.

Harris S. and M. Liddament. (2004). Retroviral Restriction by APOBEC Proteins.
Nature Reviews 4: 868.

Huthoff, H., M. Malim. (2007). "Identification of Amino Acid Residues in
APOBEC3G Required for Regulation by Human Immunodeficiency Virus Type 1
Vif and Virion Encapsidation." Journal of Virology 81: 3807-3815.

Jarmuz, A., A. Clhester, et al. (2002). "An anthropoid-specific locus of orphan C to
U RNA-editing enzymes on chromosome 22." Genomics 79: 285-296.

Johansson, E., N. Mejlhede, et al. (2002). "Crystal structure of the tetrameric
cytidine deaminase from Bacillus subtilis at 2.0 A resolution." Biochemistry 41:
2563-70.

Khan, M., C. Aberham, et al. (2001). "Human immunodeficiency virus type 1 Vif
protein is packaged into the nucleoprotein complex through an interaction with viral
genomic RNA." Journal of Virology 16:7252-65.

Kinter T., N.E. Sherman, et al. (2000). "Molecular cloning of Apo1
complementation factor (ACF), a novel RNA- binding protein involved in the editing
of apolipoprotein B mRNA." Molecular and Cellular Biology 2000:1846–1854.

Ko, T.-P., J. Lin, et al. (2003) "Crystal structure of yeast cytosine deaminase.
Insights into enzyme mechanism and evolution." Journal of Biological Chemisty 278: 19111-
19117.

Lellek, H., R. Kirsten, et al. (2000). "Purification and molecular cloning of a novel
essential component of the apolipoprotein B mRNA editing enzyme-complex."
Journal of Biological Chemistry 275: 19848–19856.

Longerich, S., U. Basu, et al. (2006). “AID in somatic hypermutation and class
switch recombination." Currernt Opinions in Immunology 2:164-74.

Lu, W., Ou, J. et al. (2002). "Phosphorylation of hepatitis C virus core protein by
protein kinase A and protein kinase C." Virology 300: 20–30.

Matsumoto,T., H. Marusawa, et al. (2006). "Expression of APOBEC2 is
transcriptionally regulated by NF-kappaB in human hepatocytes." FEBS Letters
3:731-5.

Mikl, M., I. Watt, et al. (2005). "Mice Deficient in APOBEC2 and APOBEC3."
Molecular and Cellular Biology 16:7270-7.

Minegishi, Y., A. Lavoie, et al. (2000). "Mutations in Activation-Induced Cytidine
Deaminase in Patients with Hyper IgM Syndrome." Clinical Immunology 97: 203-
210.

Muckenfuss, H., M. Hamdorf et al. (2006). "APOBEC3 proteins inhibit human
LINE-1 retrotransposition." Journal of Biological Chemistry 31:22161-72.

Navaratnam N, Fujino T, et al. (1998). "Escherichia coli cytidine deaminase provides
a molecular model for ApoB RNA editing and a mechanism for RNA substrate
recognition." Journal of Molecular Biology 275:695-714.

Navaratnam, N., J. Morrison, et al. (1993) "The p27 catalytic subunit of the
apolipoprotein B mRNA editing enzyme is a cytidine deaminase." Journal of
Biological Chemistry 268: 20709–20712.

79
80

Navarro, F., B. Bollman, et al. (2005). "Complementary function of the two catalytic
domains of APOBEC3G." Virology 333: 374-386.

OhAinle,

M., J. Kerns, et al. (2006). "Adaptive Evolution and Antiviral Activity of
the Conserved Mammalian Cytidine Deaminase APOBEC3H." Journal of Virology
80: 3853-3862.

Okazaki, I.M., A. Kotani, et al. (2007). "Role of AID in Tumorogenesis." Advances
in Immunology 94, 245-73.

Opi, S., H. Takeuchi, et al. (2006). "Monomeric APOBEC3G is catalytically active
and has antiviral activity." Journal of Virology 80: 4673-4682.

Petersen-Mahrt, S. K., and M. S. Neuberger (2003). "In vitro deamination of
cytosine to uracil in single-stranded DNA by apolipoprotein B editing complex
catalytic subunit 1 (APOBEC1). Journal of Biological Chemistry 278" 19583–
19586.

Pham, P., R. Bransteitter, et al. (2003). "Processive AID-catalysed cytosine
deamination on single-stranded DNA simulates somatic hypermutation." Nature 242:
103-7.

Pham, P., R. Bransteitter, et al. (2005). "Reward versus Risk: DNA Cytidine
Deaminases Triggering Immunity and Disease." Biochemistry 44: 2703-15.

Rogozin, I. B., M. Basu, et al. (2005). "APOBEC4, a new member of the
AID/APOBEC family of polynucleotide (Deoxy)cytidine deaminases predicted by
computational analysis." Cell Cycle 4: 1281–1285.

Schumacher, A., D. Nissley, et al. (2005). "APOBEC3G hypermutates genomic
DNA and inhibits Ty1 retrotransposition in yeast." Proc. Natl. Acad. Sci. 28:9854-9.

81
Shindo, K., A. Takaori-Kondo, et al. (2003). "The enzymatic activity of
CEM15/Apobec-3G is essential for the regulation of the infectivity of HIV-1 virion
but not a sole determinant of its antiviral activity." Journal of Biological Chemistry
278: 44412-6.

Smith, A. A., D. C. Carlow, et al. (1994). "Mutations Affecting Transition-State
Stabilization by Residues Coordinating Zinc at the Active Site of Cytidine
Deaminase." Biochemistry 33: 6468-74.

Stenglein, M. and R. Harris. (2006). "APOBEC3B and APOBEC3F inhibit L1
retrotransposition by a DNA deamination-independent mechanism." Journal of
Biological Chemistry 25:16837-41.

Tanaka,Y., H. Marusawa et al. (2006). "Anti-viral protein APOBEC3G is induced by
interferon-alpha stimulation in human hepatocytes." Biochem. Biophys. Res.
Commun. 3:314-9.

Teh, A., M. Kimura, et al. (2006). "The 1.48 A Resolution Crystal Structure of the
Homotetrameric Cytidine Deaminase from Mouse." Biochemistry 45: 7825-33.

Teng, B., S. Ochsner, et al. (1999). "Mutational analysis of apolipoprotein B mRNA
editing enzyme (APOBEC1): structure-function relationships of RNA editing and
dimerization." Journal of Lipid Research 40: 623-35.

Wang, J., R. Shinkura, et al. (2006). "Identification of a Specific Domain Required
for Dimerization of Activation-induced Cytidine Deaminase." Journal of Biological
Chemistry 28: 19115-23.

Wedekind, J., R. Gillilan, et al. (2006). "Nanostructures of APOBEC3G support a
hierarchical assembly model of high molecular mass ribonucleoprotein particles
from dimeric subunits." Journal of Biological Chemistry 50: 38122-6.

Wiegand, H. L., B. P. Doehle, et al. (2004). "A second human antiretroviral factor,
APOBEC3F, is suppressed by the HIV-1 and HIV-2 Vif proteins." The EMBO
Journal 23: 2451-8.

82
Xie, K., M. P. Sowden, et al. (2004). "The structure of a yeast RNA-editing
deaminase provides insight into the fold and function of activation-induced
deaminase and APOBEC-1." Proc. Natl. Acad. Sci. 101: 8114-9.

Yamanaka, S., M. Balestra., et al. (1995). “Apolipoprotein B mRNA-editing protein
induces hepatocellular carcinoma and dysplasia in transgenic animals.” Proc. Natl.
Acad. Sci. 18:8483-7.

Abstract (if available)

Abstract APOBEC-2 (Apo2) belongs to the Apolioprotein B (APOB) mRNA-editing enzyme catalytic polypeptide (APOBEC) family of cytidine deaminases that modify genes by deaminating cytosines in mRNA coding sequences and cytidines in ssDNA (Franca et al., 2006). APOBEC deaminases have remarkably diverse functions comprising antibody maturation, inactivation of viral genomes including Human Immunodeficiency virus-1 (HIV-1) and Hepatitis B virus (HBV), inhibition of retrotransposition, and RNA editing. The repertoire of APOBECs and their respective activities are continuously expanding to include new functions and sub-classes of the family.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

A structure based study of the HIV restriction factor APOBEC3G

PDF

Biochemical studies of APOBEC protein family

PDF

Structural and biochemical analyses on substrate specificity and HIV-1 Vif mediated inhibition of human APOBEC3 cytidine deaminases

PDF

Structural and biochemical studies of two DNA transaction enzymes

PDF

How phosphorylation affects the biochemical regulation and targeting of activation induced cytidine deaminase (AID)

PDF

Exploring roles of human APOBEC-mediated RNA editing activity

PDF

Structural and biochemical determinants of APOBEC1 substrate recognition and enzymatic function

PDF

Scanning and catalytic properties of AID with structural comparisons to APOBEC3A

PDF

APOBEC RNA mutational signatures and the role of APOBEC3B in SARS-CoV-2 infection

PDF

In silico enzyme modeling: A study of entropic contributions to enzyme catalysis and the design of artificial kemp eliminase

PDF

Structural studies of two key factors for DNA replication in eukaryotic cells

PDF

Structure and regulation of lymphoid tyrosine phosphatase (LYP) in autoimmune response

PDF

AID scanning & catalysis and the generation of high-affinity antibodies

PDF

Structure and function of archaeal McM helicase

PDF

Mechanistic basis for chromosomal translocations at the E2A gene

PDF

Discovery of mature microRNA sequences within the protein- coding regions of global HIV-1 genomes: Predictions of novel mechanisms for viral infection and pathogenicity

PDF

Biochemical characterization and structural analysis of two hexameric helicases for eukaryotic DNA replication

PDF

Simulating the helicase motor of SV40 large tumor antigen

PDF

Differential role of two coactivators, CCAR1 and CARM1, for dysregulated beta-catenin activity in colorectal cancer cell growth and gene expression

Asset Metadata

Creator Prochnow, Courtney (author)

Core Title The crystal structure of APOBEC-2 and implications for APOBEC enzymes

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Molecular Biology

Publication Date 02/21/2008

Defense Date 01/25/2008

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag activation induced cytidine deaminase,AID,APOBEC,APOBEC-2,cytidine deaminase,cytidine deamination,HIV,OAI-PMH Harvest

Language English

Advisor Chen, Xiaojiang S. (committee chair), Bau, Robert (committee member), Chen, Lin (committee member), Goodman, Myron F. (committee member)

Creator Email cprochno@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m1025

Unique identifier UC1430925

Identifier etd-Prochnow-20080221 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-39667 (legacy record id),usctheses-m1025 (legacy record id)

Legacy Identifier etd-Prochnow-20080221.pdf

Dmrecord 39667

Document Type Dissertation

Rights Prochnow, Courtney

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email cisadmin@lib.usc.edu