Emerging complexities of APOBEC3G action on immunity and viral fitness during HIV infection and treatment.

Abstract

The enzyme APOBEC3G (A3G) mutates the human immunodeficiency virus (HIV) genome by converting deoxycytidine (dC) to deoxyuridine (dU) on minus strand viral DNA during reverse transcription. A3G restricts viral propagation by degrading or incapacitating the coding ability of the HIV genome. Thus, this enzyme has been perceived as an innate immune barrier to viral replication whilst adaptive immunity responses escalate to effective levels. The discovery of A3G less than a decade ago led to the promise of new anti-viral therapies based on manipulation of its cellular expression and/or activity. The rationale for therapeutic approaches has been solidified by demonstration of the effectiveness of A3G in diminishing viral replication in cell culture systems of HIV infection, reports of its mutational footprint in virions from patients, and recognition of its unusually robust enzymatic potential in biochemical studies in vitro. Despite its effectiveness in various experimental systems, numerous recent studies have shown that the ability of A3G to combat HIV in the physiological setting is severely limited. In fact, it has become apparent that its mutational activity may actually enhance viral fitness by accelerating HIV evolution towards the evasion of both anti-viral drugs and the immune system. This body of work suggests that the role of A3G in HIV infection is more complex than heretofore appreciated and supports the hypothesis that HIV has evolved to exploit the action of this host factor. Here we present an overview of recent data that bring to light historical overestimation of A3G's standing as a strictly anti-viral agent. We discuss the limitations of experimental systems used to assess its activities as well as caveats in data interpretation.

Full text

Page 1

Emerging complexities of APOBEC3G action on
immunity and viral fitness during HIV infection
and treatment
Monajemi et al.
Monajemi et al. Retrovirology 2012, 9:35
http://www.retrovirology.com/content/9/1/35

Page 2

REVIEWOpen Access
Emerging complexities of APOBEC3G action on
immunity and viral fitness during HIV infection
and treatment
Mahdis Monajemi1†, Claire F Woodworth2†, Jessica Benkaroun2, Michael Grant3*and Mani Larijani3*
Abstract
The enzyme APOBEC3G (A3G) mutates the human immunodeficiency virus (HIV) genome by converting
deoxycytidine (dC) to deoxyuridine (dU) on minus strand viral DNA during reverse transcription. A3G restricts viral
propagation by degrading or incapacitating the coding ability of the HIV genome. Thus, this enzyme has been
perceived as an innate immune barrier to viral replication whilst adaptive immunity responses escalate to effective
levels. The discovery of A3G less than a decade ago led to the promise of new anti-viral therapies based on
manipulation of its cellular expression and/or activity. The rationale for therapeutic approaches has been solidified
by demonstration of the effectiveness of A3G in diminishing viral replication in cell culture systems of HIV infection,
reports of its mutational footprint in virions from patients, and recognition of its unusually robust enzymatic
potential in biochemical studies in vitro. Despite its effectiveness in various experimental systems, numerous recent
studies have shown that the ability of A3G to combat HIV in the physiological setting is severely limited. In fact, it
has become apparent that its mutational activity may actually enhance viral fitness by accelerating HIV evolution
towards the evasion of both anti-viral drugs and the immune system. This body of work suggests that the role of
A3G in HIV infection is more complex than heretofore appreciated and supports the hypothesis that HIV has
evolved to exploit the action of this host factor. Here we present an overview of recent data that bring to light
historical overestimation of A3G’s standing as a strictly anti-viral agent. We discuss the limitations of experimental
systems used to assess its activities as well as caveats in data interpretation.
The role of APOBEC3G in HIV restriction
APOBEC3G (A3G) is a recently discovered primate-spe-
cific member of the apolipoprotein B mRNA-editing en-
zyme, catalytic polypeptide-like editing complex family of
cytidine deaminase enzymes with potential to inhibit
propagation of the human immunodeficiency virus (HIV)
[1,2]. The APOBEC family includes eleven members in
humans: activation-induced cytidine deaminase (AID),
APOBEC1, APOBEC2, APOBEC3A-H, and APOBEC4
[3,4]. These enzymes convert deoxycytidine (dC) to deox-
yuridine (dU) in single stranded DNA (ssDNA) or RNA of
human and viral genomes, thereby affecting a variety of
physiological functions [5–7].
through the study of heterokaryons generated between
A3Gwasdiscovered
cells permissive and non-permissive to infection by virion
infectivity factor (Vif)-deficient HIV that were used to de-
termine the action of the HIV protein Vif [1,8,9]. A3G is
primarily expressed in CD4+T lymphocytes, macrophages,
and dendritic cells, which are all the natural targets of HIV
infection [2,10–14]; although expression in other tissues
may be induced by interferon(s) [15–18]. A3G mutates dC
in nascent viral minus strand DNA generated by reverse
transcription [17–24] and preferentially deaminates dC in
signature trinucleotides (CCC, TCC) often referred to as
hotspots [6,19–21]. The resulting dUs can trigger DNA
degradation through the action of DNA repair pathways,
such as those involving uracil DNA glycosylase and apuri-
nic-apyrimidinic endonuclease [25,26]. For viral genomes
that evade destruction, the consequent deoxyguanosine
(dG) to deoxyadenosine (dA) substitutions in plus strand
DNA can alter reading frames, introduce premature transla-
tion termination codons, and/or produce mutated viral pro-
teins [7,20–25]. In addition, A3G can disrupt propagation
* Correspondence: mgrant@mun.ca; mlarijani@mun.ca
†Equal contributors
3Division of Biomedical Sciences, Faculty of Medicine, Health Sciences Center,
MUN, 300 Prince Phillip Dr., St. John’s, NL, A1B 3V6, Canada
Full list of author information is available at the end of the article
© 2012 Monajemi et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative
Commons Attribution License ( http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and
reproduction in any medium, provided the original work is properly cited.
Monajemi et al. Retrovirology 2012, 9:35
http://www.retrovirology.com/content/9/1/35

Page 3

of HIV by binding viral RNA, interfering with the DNA
strand transfer acrobatics of reverse transcription, physically
blocking reverse transcriptase (RT), and obstructing integra-
tion into the host cell genome [24,26–31]. A3G has been
shown to block RT activity by decreasing tRNA priming,
competing for binding to templates, restricting strand trans-
fer during reverse transcription, and direct binding
[28,32,33]. Beyond the reverse transcription stage, incorpor-
ation of dU into minus strand DNA of the HIVgenome has
been shown to interfere with synthesis of the complemen-
tary plus strand [23]. These findings initially led to the no-
tion that A3G can inhibit viral propagation through
pathways dependent or independent of its deamination ac-
tivity; however, many studies supporting deaminase-inde-
pendent activities utilized A3G overexpression. It has
recently been appreciated that with low level A3G expres-
sion, which may be a more accurate representation of the
physiological case, deaminase activity is required for viral
restriction [34–39]. While the relative contribution of de-
amination independent activities to viral restriction remains
contentious, these may prove more relevant to the action of
A3G in restricting endogenous non-long terminal repeat
retrotransposons, such as long and short interspersed nu-
clear elements [40–45]. The anti-retroelement activity of
A3G may represent a host strategy to protect its genome
from the deleterious effects of transposable elements. A
possible mechanism could involve the binding of A3G to
retroelements resulting in blockage of their mobility [46].
The recent expansion of a single APOBEC3 gene in
mice to seven (APOBEC3A-H) in primates and the rela-
tively high divergence within APOBEC3 enzymes in pri-
mates are evidence for immense evolutionary pressure
on the locus suggested to possibly be concomitant with
the emergence of modern lentiviruses [3,4,47,48]. Con-
versely, the finding that the accelerated rate of A3G di-
vergence predates modern lentiviruses, together with the
lack of a clear correlation between human A3G poly-
morphisms and the progression of acquired immunodefi-
ciency syndrome (AIDS), suggest that lentiviral pressure
may be, at best, only partially responsible for expansion
of the APOBEC3 locus [49–52]. This manner of growth
in host defence capacity can reciprocally drive co-evolu-
tion of highly adaptable viruses. In this regard, we high-
light an emerging body of evidence suggesting that the
activity of A3G may be partially subverted by HIV for its
survival benefit. These data support a more complex sce-
nario in which the initial perception of A3G as a strictly
anti-viral agent may have been naïve.
Viral and cellular factors limiting APOBEC3G effectiveness
The view of A3G as a potent intrinsic anti-viral factor was
largely borne out of findings of high levels of dG to dA
hypermutated virus sequences in di- and tri-nucleotide
motifs targeted by A3G [53–57]. In stark contrast, the
previously recognized mutational machinery of HIV, RT,
only introduces approximately one mutation per viral gen-
ome during a replication cycle [58]. Supporting the po-
tency of A3G as a mutagenic agent is a wealth of
biochemical data showing that it is a highly processive en-
zyme able to mediate multiple mutations on a given
stretchofssDNA. Accordingly,
diminishes viral propagation in several cell culture experi-
mental systems of HIV infection [7,20,22,42,47].
To counteract these activities, lentiviruses have evolved
several strategies, primarily in the form of auxiliary pro-
teins such as Vif, which binds and targets newly synthe-
sized A3G for degradation via a ubiquitin-dependent
proteosomal pathway [59–69]. A3G is packaged into vir-
ions in infected virus-producing cells and it has been
shown that it is largely this virion-packaged fraction of
A3G rather than the pool of cytoplasmic A3G that is most
active on the viral genome in newly infected cells [70–74].
The number of A3G molecules incorporated into each vir-
ion is dependent on the level of A3G expression in the
producer cell [75]. On average, 3 to 11 molecules of A3G
are sufficient for effective viral restriction in the target cell
[76]. Besides lowering A3G levels through degradation, Vif
has also been suggested to directly interfere with A3G en-
capsidation and may impair its translation [66,74,77–80].
Vif utilizes other co-factors present in the target cell to
ubiquitinate A3G and it was recently shown that Core
binding factor (CBF)-β, a cellular transcription factor, is
required for Vif-mediated degradation of A3G [81,82]. As
a result, when Vif is present, the mutation levels induced
by A3G and its effectiveness in viral restriction are dimin-
ished. That Vif is essential for HIV replication in A3G-
expressing cells, and that the sole function of Vif was
thought to be A3G inactivation, lent credence to the no-
tion that A3G is a potent restrictor of HIV propagation
[83]. On the other hand, it is now appreciated that even in
the presence of Vif, A3G can still cause sub-lethal levels of
dG to dA mutations [19,84]. It is possible that the prefer-
ential targeting of newly synthesized A3G by Vif leaves a
fraction of previously synthesized A3G intact [85]. In
addition, it appears that Vif expression does not com-
pletely abolish A3G activity and the correlation between
the levels of viral infectivity and A3G inhibition by Vif is
not absolute [62,78]. Other functions for Vif and Vif-
mediated ubiquitination, besides A3G degradation, are
also coming to light. For instance, along with the auxiliary
protein Vpr, Vif can induce G2 cell-cycle arrest, which
may contribute to CD4+T lymphocyte depletion [86–89].
Vif thus mediates several functions that are independent
of its interaction with A3G and is a variable negative regu-
lator of A3G activity rather than a complete inhibitor.
A3G action is further limited by its entrapment in
high-molecular-mass ribonuclear complexes (HMM) that
may reach megadaltons in size, mediated by non-specific
A3Gsignificantly
Monajemi et al. Retrovirology 2012, 9:35
http://www.retrovirology.com/content/9/1/35
Page 3 of 12

Page 4

binding of cellular and/or viral RNA and proteins
[12,71,90–97]. The shuttling of A3G into newly synthe-
sized virions depends on binding viral RNA and/or pro-
teins [98–101]. The requirement for high affinity
interactions with RNA/DNA substrates may explain the
evolution of A3G (and other APOBEC enzymes, e.g.
AID) to contain an unusually high number of charged
residues on its surface [102–104]. Ironically, this same
attribute necessary to enact the anti-viral function of
A3G may also be a key contributor to limiting its anti-
viral function through HMM formation. Reversion of
HMM to low-molecular-mass (LMM) A3G can be ex-
perimentally mediated by treatment with RNase A/H
[70,71,105]. The RNase H activity of RT is thought to re-
lease viral RNA-bound A3G, allowing it to act on the
proximal minusstrand DNA during
[2,19,75]. Enzymatically active A3G able to be incorpor-
ate into newly synthesized virions is strictly found out-
side of the HMM complexes in the LMM fraction
[73,106]. The LMM form primarily resides in peripheral
blood-derived resting CD4+T cells and monocytes [12];
however, upon activation of CD4+T cells or differenti-
ation of monocytes into macrophages, a higher propor-
tion of A3G is shuttled to HMM complexes [2,91].
Although this was suggested to be a mechanism that
restricted the infection of resting T cells by HIV, subse-
quent knockout studies of LMM A3G in resting CD4+T
cells did not render these cells permissive to HIV infec-
tion, thus indicating that the difference in the LMM-
versus HMM-bound proportion of A3G is not the sole
mechanism for resistance of resting CD4+T cells to HIV
infection [107,108]. Beyond the induction of HMM for-
mation by HIV through cellular activation processes, Vif
has been shown to directly promote HMM production
[109]. Remarkably complex co-evolution is evident con-
sidering the intimate linkage between HIV infection and
HMM formation and the notable level of mechanistic in-
tegration between A3G function and the viral replication
machinery. The RNase H activity of RT is at once both
necessary and detrimental to viral propagation due to its
role in the release of active A3G.
The complexities surrounding regulation of Vif activity
and HMM formation notwithstanding, it is clear that both
result in diminished A3G efficacy. It is possible that muta-
tions introduced by A3G only succeed in restricting viral
replication at a sub-optimal level and conversely may assist
the virus by generating sequence variation [35,39,84]. Con-
sequently, an alternative view that A3G activity can con-
tribute to viral fitness has recently gained strong support.
In the following sections, we highlight evidence for the
pro-viral activities of A3G. At the same time, we discuss
caveats of experimental systems and data interpretation
that must henceforth be considered in development of a
revised and better-informed picture of A3G function.
its synthesis
The role of APOBEC3G in generation of anti-viral drug
resistant HIV
Gain of resistance to drug(s) used in the treatment of
HIV is a major determinant of viral evolution during the
course of disease. To date, almost a hundred drug resist-
ance mutation sites have been identified in the HIV gen-
ome [110]. These induce resistance to common anti-HIV
drugs acting as nucleoside/nucleotide analogue RT inhi-
bitors, such as 2',3'-dideoxy-3'-thiacytidine (3TC), abaca-
vir (ABC), and 2',3'-dideoxyinosine (DDI), as well as
non-nucleoside/nucleotide analogue RT inhibitors, in-
cluding Nevirapine (NVP), Delavirdine (DLV), and Efa-
virenz (EFV) [110]. Drug resistance mutations function
directly by altering drug targets or indirectly by modify-
ing pathways that contribute to drug escape. Many drug
resistance mutations have been shown to reside in A3G
hotspots [111].
A bioinformatics study assessed the probability of A3G
mutations in known drug resistance sites taking into con-
sideration the double-crested gradient of A3G-induced
mutational levels throughout the HIV genome. Out of
52,000 G to A mutations, only 695 (1.3%) were located in
drug resistance sites [112]. In this context, the investigators
reported a modest correlation between A3G activity and
the generation of drug resistance mutations relative to the
overall footprint of A3G on the HIV genome [19,112,113];
however, recent experimental evidence more strongly
implicates A3G in the generation of drug resistance muta-
tions. For example, the very common M184I(V) mutation
of RT that causes resistance to 3TC and, to a lesser extent,
ABC and DDI, is located in an A3G hotspot (TCCAT to
TCUAT) and is produced by A3G in vitro during HIV rep-
lication in cell culture systems [114]. Intriguingly, this was
observed in the absence of 3TC in as many as 40% of
sequenced proviruses, reflecting a pre-treatment pool of
resistant viruses poised for propagation after drug exposure
[29,115–118]. Because this mutation may in fact reduce
viral replication fitness in the absence of 3TC [119–122], it
is likely that this measurement actually underestimates the
role of A3G in the generation of this mutation. In support
of this notion, the M184I mutation emerges at significantly
higher rates when the virus is grown in A3G-expressing as
compared to A3G non-expressing host cells, indicating
that A3G activity is the major source of this mutation
[123]. This is a striking example of the parallel role of A3G
in simultaneously aiding host and virus: in the same man-
ner that it acts as a pre-existing innate immune factor that
fortifies host defenses prior to viral exposure, A3G boosts
the inherent ability of HIV to gain resistance even before
drug treatment. That this mutation is associated with a de-
cline in viral fitness may indicate that drug resistance pre-
sents a significant source of pressure in viral evolution
resulting in the gain or maintenance of A3G hotspots in
key positions in the viral genome.
Monajemi et al. Retrovirology 2012, 9:35
http://www.retrovirology.com/content/9/1/35
Page 4 of 12

Page 5

If the contribution of A3G action to drug resistance
and survival of HIV is a biologically considerable one,
the evolution of HIV during disease could involve active
relaxation of A3G inhibition. Indeed, direct evidence for
this phenomenon was provided by the prevalence of the
Vif K22H mutation in patients failing drug treatment, as
compared to treatment-naïve patients [124,125]. Vif K22
is a key residue for interaction with A3G, and Vif K22H
exhibits reduced effectiveness in neutralizing A3G [115].
Ex vivo infection of peripheral blood mononuclear cells
(PBMCs) with viral stocks harboring various other Vif
mutations that are unable to deactivate A3G (e.g. Vif
K22E) yielded a significant increase in the generation of
M184I mutants [114]. In addition, several drug resistance
mutations, including M184I in RT and G16E/M36I in
the protease, are significantly more common in patients
harboring elevated relative levels of K22H-mutated
viruses [125]. Like the M184I mutation, both G16E and
M36I mutation sites are located in A3G hotspots. Thus,
not only does HIV benefit from spontaneous pre-drug
treatment A3G-induced mutations in a passive, some-
what random manner, it appears that resistance sites for
some of the most commonly used drugs arose in A3G
hotspots. This in no way implies viral sentience, but
merely indicates a selective advantage derived from the
overlap of sites more susceptible to mutation (A3G hot-
spots) being able to confer drug resistance.
The contribution of APOBEC3G to the evasion of adaptive
immunity by HIV
Restrictions imposed on the activity of A3G by Vif and
HMM limit its effectiveness as an innate immune agent;
however, following the first weeks of HIV infection, devel-
opment of B and T cell mediated adaptive immunity par-
tially controls viremia [116–118,126]. A central facet of
the adaptive immune response is elimination of infected
target cells by cytotoxic T cells (CTL), as highlighted by
the close inverse association between robustness of the
CTL response with viremia levels and disease progression
[127–129]. Thus, evasion of the CTL response is thought
to be a powerful driving force for the evolution of HIV
during disease, as confirmed by several studies showing
the prevalence of CTL escape in HIV infection [130,131].
CTL evasion may result from alterations in CTL access to
infected cells. For instance, the auxiliary HIV protein Nef
modulates class I MHC expression to decrease the recog-
nition and killing of infected cells [132]. Alternatively,
CTL evasion may result from alterations in the interac-
tions between the CTL and infected target cell. Mutations
in CTL recognition epitopes have been shown to mediate
CTL evasion through modulating the efficacy of CTL acti-
vation [133–135].
As in the case of drug resistance, it is possible that
HIV can exploit the limited non-lethal action of A3G to
generate CTL escape mutants. In support of this model,
a study examining CTL escape during early infection
found that approximately a third of the rapidly mutating
sites mediating CTL escape were embedded in A3G hot-
spots, with more highly mutating sites being relatively
enriched in A3G hotspots [136]. Twenty-four rapidly di-
versifying sites were identified at which G to A muta-
tions were 2–3 fold more frequent than the overall G to
A mutation rate across the entire HIV genome (29 versus
12%). Fourteen of these sites located in or near CTL epi-
topes. These data suggest that it may be advantageous
towards immune escape for HIV to maintain A3G hot-
spots in areas where mutations can affect processing,
presentation or recognition of T cell epitopes, or con-
versely to establish T cell epitopes near A3G hotspots.
In contrast, another study reported that A3G muta-
tions enhance the virus-specific CTL response through
the introduction of premature stop codons into the HIV
genome that cause the generation of truncated or mis-
folded proteins [137]. In this study, Vif+ or Vif- HIV was
produced in the presence or absence of A3G in a cell
line and subsequently used to infect PBMCs followed by
assessing their susceptibility to MHC-matched peptide-
specific CTL clones. It is possible that the finding of
enhanced target cell killing as a result of A3G activity
reflects an inherent bias of the specificity of the CTL
clones examined. In addition, given the numbers, diver-
sity, and relative scarcity of CTL specific for each par-
ticular peptide in vivo, the general biological relevance of
this work remains to be determined. Therefore, although
A3G appears to play a role beyond innate immunity and
modulate adaptive immunity, further work is required to
elucidate the nature and extent of this activity.
Manipulation of APOBEC3G effectiveness: implications
and challenges for the design of therapeutic approaches
To date, multiple avenues have been suggested and/or
pursued towards exploitation of A3G as an antiviral ther-
apy. These approaches include the development of small
molecules that inhibit the interaction between Vif and
A3G and/or inhibit interactions with cellular factors that
act downstream of Vif, enhancement of LMM formation
over HMM formation, and increasing A3G levels by treat-
ment with interferons or gene-therapy delivery of A3G
along with the restriction factor Trim 5α [70,138–140].
Strategies to down-regulate the action of Vif and
HMM that were initially suggested as therapeutic
approaches have recently been questioned in light of the
increasingly apparent pro-viral activities of A3G [35,141].
Disturbing the Vif-APOBEC interaction presents a deli-
cate challenge because subtle adjustments to Vif activity
have been shown to modulate levels of A3G activity. For
instance, naturally occurring patient-derived virions har-
boring Vif mutations selectively exhibit viral genome
Monajemi et al. Retrovirology 2012, 9:35
http://www.retrovirology.com/content/9/1/35
Page 5 of 12