Accepted Manuscript
Modeling the Effect of Tat Inhibitors on HIV Latency. Luis U. Aguilera, Jesus ´ Rodr´ıguez-Gonzalez ´ PII: DOI: Reference:
S0022-5193(19)30159-6 https://doi.org/10.1016/j.jtbi.2019.04.018 YJTBI 9893
To appear in:
Journal of Theoretical Biology
Received date: Revised date: Accepted date:
14 August 2018 7 February 2019 16 April 2019
Please cite this article as: Luis U. Aguilera, Jesus Modeling the Ef´ Rodr´ıguez-Gonzalez, ´ fect of Tat Inhibitors on HIV Latency., Journal of Theoretical Biology (2019), doi: https://doi.org/10.1016/j.jtbi.2019.04.018
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Highlights • Mathematical study to understand Tat-inhibitors as virus-suppressing
CR IP T
agents.
• HIV-productive and latent cell phenotypes are described by stochastictransitions.
AN US
• Competitive and non-competitive inhibitors produce reversible viral suppression.
• Didehydro-Cortistatin A (dCA) achieves a more permanent viral inhi-
AC
CE
PT
ED
M
bition.
1
AC
CE
PT
ED
M
AN US
CR IP T
ACCEPTED MANUSCRIPT
2
ACCEPTED MANUSCRIPT
Modeling the Effect of Tat Inhibitors on HIV Latency. Luis U. Aguilera
, Jes´ us Rodr´ıguez-Gonz´alezb,
Department of Modeling of Biological Processes, COS Heidelberg / Bioquant, Heidelberg University, Im Neuenheimer Feld 267, 69120 Heidelberg, Germany. b Centro de Investigaci´ on y Estudios Avanzados del Instituto Polit´ecnico Nacional, Unidad Monterrey, Via del Conocimiento 201, Parque PIIT, CP 66600 Apodaca NL, M´exico.
CR IP T
a
1,a
AN US
Abstract
Even in the presence of a successful combination therapy stalling the progress of AIDS, developing a cure for this disease is still an open question. One of the major steps towards a cure would be to be able to eradicate latent HIV
M
reservoirs present in patients. During the last decade, multiple findings point to the dominant role of the viral protein Tat in the establishment of latency.
ED
Here we present a mathematical study to understand the potential role of Tat inhibitors as virus-suppressing agents. For this aim, we implemented
PT
a computational model that reproduces intracellular dynamics. Simulating an HIV-infected cell and its intracellular feedback we observed that remov-
CE
ing Tat protein from the system via inhibitors resulted in a temporary and reversible viral suppression. In contrast, we observed that compounds that
AC
interact with Tat protein and disrupt the integrated viral genome produced a more permanent viral suppression.
1
Email address:
[email protected] (Jes´ us Rodr´ıguez-Gonz´ alez) Present Address: Colorado State University.
Preprint submitted to Journal of Theoretical Biology
April 17, 2019
ACCEPTED MANUSCRIPT
Keywords: Viral suppression, Stochastic model, Tat circuit 1. Introduction
2
Treating patients infected with the Human Immunodeficiency Virus (HIV)
3
has strongly progressed during the last years. Although until today HIV in-
4
fection cannot be eradicated in patients, Highly Active Antiretroviral Ther-
5
apy (HAART) decreases viral loads during the infection and prevents clin-
6
ical progression to the Acquired Immunodeficiency Syndrome (AIDS) [38].
7
A major barrier to achieving HIV eradication is the existence of a reservoir
8
of resting memory CD4+ T-lymphocytes (T4-cells) carrying latent provirus
9
[7, 10]. HIV latency is a reversible non-virus-producing phenotype in an in-
AN US
fected cell [21].
M
10
CR IP T
1
11
It is not entirely clear what molecular processes dictate a latent or productive
13
cell infection. Multiple mechanisms have been suggested to contribute to HIV
14
latency [18, 20]. However, important findings point to the dominant role of
15
the viral protein Tat in latency [30]. Tat is a transactivating regulatory pro-
16
tein. Its activation is a key mechanism that promotes viral gene expression
17
[4]. Thus, once Tat protein is produced, it enhances HIV gene expression by
CE
PT
ED
12
binding to the trans-activating response element (TAR), a stem-loop that is
19
present in the 5 end of newly synthesized viral transcripts. Tat changes the
20
effective transcription rate of the respective genes up to 100-fold over basal
21
expression levels by recruiting the positive transcription elongation complex
AC 18
4
ACCEPTED MANUSCRIPT
(P-TEFb) made of Cyclin T1 and CDK9 [27]. Since this process also accel-
23
erates Tat transcription itself, it constitutes a positive feedback loop in the
24
system. The absence of the Tat protein results in premature termination of
25
transcription. Over-expression of Tat protein is sufficient to induce the pro-
26
duction and spread of HIV from resting memory T4-cells [9]. The complete
27
scheme of HIV gene expression is given in Fig. 1.
28
CR IP T
22
Altogether, the cellular fate of individual cells is determined mainly by the
30
number of molecules, e.g. of Tat protein present in the cell. Therefore,
31
like other cellular events which involve low molecule numbers, cellular fate
32
should be strongly affected by the stochasticity of discrete molecular events
33
[2]. The central role of stochasticity in the processes carried out by the
34
Tat protein was confirmed by studies conducted in the group of Weinberger.
35
They modeled the Tat-circuit and adjusting it to the experimental data they
36
determined a monostable architecture, which after introducing stochasticity
37
showed bimodal dynamics [37]. Subsequently, they proved that HIV expres-
38
sion is dictated by the time that the Tat protein remains in the inactive
39
(deacetylated) state, which is longer than the time that it remains in the
40
active state (acetylated) [36]. Recently, the same group proved that the
CE
PT
ED
M
AN US
29
transition between a latent to a productive infection depends solely on the
42
Tat circuit, independent of the cellular state [31].
AC 41
43
44
As previously stated, the biggest barrier to eradicating HIV infection is the 5
ACCEPTED MANUSCRIPT
persistence of HIV latent infected cells [33]. Two main strategies have been
46
discussed for achieving a functional cure. The first one is known as the
47
”shock and kill” strategy, and it involves the purge of latent reservoirs by up-
48
regulating HIV expression in combination with antiviral therapy [26]. The
49
second less explored approach involves the long-term control of the virus by
50
achieving a permanent HIV suppression in the absence of HAART [8, 29]. So
51
far, the shock and kill strategy is being actively studied and clinical studies
52
have been discussed [22]. In contrast, the only example of permanent inhibi-
53
tion is the use of the Tat inhibitor didehydro-cortistatin-A, a natural alkaloid
54
that restricts viral reactivation probably by the establishment of epigenetic
55
modifications at the viral promoter [24]. Permanent HIV suppression still
56
faces multiple challenges; the most important is the finding of stronger and
57
permanent virus-suppressing agents [11].
ED
M
AN US
CR IP T
45
58
Intense research has been done to develop specific compounds to affect the
60
Tat-dependent transcription. Those compounds include specific Tat com-
61
petitive inhibitors and indirect inhibitors to cell machinery used during the
62
Tat-dependent transcription. However, the effect of those inhibitors in viral
63
latency is not completely understood [25]. Little attention has been paid to
CE
PT
59
understand the potential role of Tat inhibitors as virus-suppressing agents.
65
To test our hypothesis, we developed a series of computational experiments
66
that allowed us to simulate the temporal HIV infection in the presence of Tat
67
inhibitors. Our simulation results showed that competitive, non-competitive
AC 64
6
ACCEPTED MANUSCRIPT
and irreversible Tat inhibitors confer a dose-dependent viral inhibition and
69
that removal of those inhibitors resulted in viral expression. Contrarily, by
70
simulating the suggested mechanism of action of didehydro-cortistatin-A, we
71
predicted a more permanent inhibition in the viral gene expression.
CR IP T
68
72
2. Materials and Methods
74
2.1. Model for intracellular HIV gene expression.
75
To reproduce the overall dynamics of the intracellular HIV infection, we used
76
the simple and tractable mathematical model developed by Razooky et al.
77
[31]. In the model the dynamics of viral biomolecules are lumped in the
78
following reactions: First, reaction (1) gives the viral promoter switching
79
between active (LT RON ) and inactive (LT ROF F ) states. Subsequently, in
80
reaction (2) a constant term represents the basal transcription rate of HIV
81
which is given by the TATA-binding protein, NF-κB, Sp1 and RNAPII [27].
82
Reaction (3) describes a complex formed by the viral promoter and Tat
83
protein, transactivation is represented in reaction (4). Translation of Tat
84
protein is shown in reaction (5), and finally, first order terms that represent
85
the degradation of the mRN A and Tat are specified in reactions (6) and (7),
AC
CE
PT
ED
M
AN US
73
86
respectively.
k
ON −−− −− * LT ROF F ) − LT RON ,
kOf f
7
(1)
ACCEPTED MANUSCRIPT
k
k
bind −− − −− * T at + LT RON − ) − − LT RON −T at ,
kunbind
k
k
AN US
trans −→ LT RON −T at + mRN A, LT RON −T at −−
T at mRN A −− → mRN A + T at,
δ
(3)
(4)
(5)
(6)
ED
M
mRN A −−m→ ∅,
(2)
CR IP T
LT RON −−m→ mRN A + LT RON ,
δ
T at T at −− → ∅.
(7)
Model initial conditions are given in Table 1, and all the parameters in the
88
reactions have ranges defined by experimental reports and are depicted in
89
Table 2. The stochastic model dynamics were solved using the Gillespie’s
90
algorithm [12] coded in COPASI [14].
CE
AC
91
PT
87
92
2.2. Tat dependent transcription inhibitors.
93
To describe the mechanism of action from Tat inhibitors, we borrowed termi-
94
nology from enzyme kinetics and defined three categories according to their 8
ACCEPTED MANUSCRIPT
mechanism of action [32]. Additionally, given the particular mechanism of
96
action of didehydro-cortistatin-A, it was considered in an independent cate-
97
gory. In the simulations, we considered inhibitor concentrations as constants.
98
This can be justified by assuming a large inhibitor concentration so that the
99
formation of complexes does not change the inhibitor dynamics [15]. The de-
100
scription of additional reactions and mechanisms of action for Tat inhibitors
101
are given as follows:
AN US
CR IP T
95
102
Competitive Inhibitors (CI) compete with the LTR for the active site of the
104
Tat. Examples of those inhibitors are oligonucleotides that mimic the TAR
105
structure [19, 1, 39] (see Fig. 2A). The inhibition constant (IC), for those
106
compounds, is reported to be of the order of 250nM. The effect of those
107
compounds was incorporated in the model by adding a reversible reaction
108
(8), that describes the binding of the CI to T at.
PT
ED
M
103
k
bCI −− −− * CI + T at − ) − CI/T at.
kubCI
(8)
Non-Competitive Inhibitors (NCI) reversibly bind to elements that are indis-
110
pensable for Tat-dependent transcription but not the active site. Examples
111
of NCI include Flavopiridol and its derivatives that bind and inactivate the
AC
CE
109
112
cellular factor P-TEFb that is indispensable on the Tat-dependent transcrip-
113
tion, with IC value of 10nM [6, 3], 6BIOder with an IC value of 40nM [17, 13],
114
and Tat peptide inhibitors with IC ranging from 0.1 to 3.5 µM [34]. A graph-
9
ACCEPTED MANUSCRIPT
ical representation for the mechanism of action of these compounds is given
116
in Fig. 2B. The effect of those compounds was incorporated in the model
117
by adding reversible reaction (9), describing the binding of the NCI to the
118
complex LT RON −T at and reversible reaction (10) that represents the binding
119
of the NCI to T at. k
CR IP T
115
bN CI −− −− −− * N CI + LT RON −T at − ) − N CI/LT RON −T at ,
AN US
kubN CI
k
2bN CI −− −− −− −− * N CI + T at − ) − N CI/T at.
k2ubN CI
(9)
(10)
Irreversible Inhibitors (II) exert their action by forming an irreversible com-
121
plex that promotes proteosomal degradation. Examples include Triptolide
122
(a natural diterpenoid epoxide) with IC ranging from 0.32 - 1.6 nM [35], and
123
dominant negative mutants of Cycling T1 that promote Tat protein degra-
124
dation [16] (see Fig. 2C). The effect of these compounds was incorporated
125
in the model by adding reaction (11) that describes the inactivation of T at
126
by binding the II, and reaction (12) that describes the degradation of the
127
complex.
AC
CE
PT
ED
M
120
128
k
bII II + T at −− → II/T at,
δII/T at
II/T at −−−−→ ∅.
(11)
(12)
Didehydro-cortistatin-A (dCA) is a natural alkaloid that dramatically re10
ACCEPTED MANUSCRIPT
stricts viral reactivation promoting a state of permanent latency. dCA mech-
130
anism of action is not fully understood, but it has been shown that it binds
131
specifically to the Tat protein. Additionally, it has been suggested that it
132
promotes the establishment of epigenetic modifications at the viral promoter
133
[23]. This compound has a IC ranging from 0.7 pM to 2.6 nM [25, 24] (see
134
Fig. 2D). The effect of dCA was incorporated into the model by adding a
135
reaction (13), describing the irreversible binding of the dCA to the complex
136
LT RON −T at .
AN US
CR IP T
129
k
dCA −→ dCA/LT RON −T at . dCA + LT RON −T at −−
(13)
3. Results
138
To reproduce the latent-productive decision at single-cell resolution stochas-
139
tic dynamics were taken into account. Here we performed multiple stochastic
140
simulations using the Direct Method from Gillespie [12] and the intracellular
141
model (reactions (1) to (7)) obtaining trajectories that represent the dynam-
142
ics of T at in one infected cell. In our simulations, we used the dynamics of
143
T at to define a threshold to differentiate latent and virus-producing states.
144
This threshold was assumed to be similar to the one used by Razooky et
AC
CE
PT
ED
M
137
145
al. [31]. Latency is characterized by low expression of viral gene products,
146
and for that reason, a low concentration of T at (T at < 100 Molecules/Cell)
147
was used to define the latent state. In contrast, the virus-producing state
148
is characterized by the active expression of viral gene products. For this, 11
ACCEPTED MANUSCRIPT
we assumed the virus-productiing state if T at reached values larger than
150
100 Molecules/Cell. The simulation results showed cellular activation prob-
151
abilistically reactivates latent virus starting in the initial condition (latent
152
state) and then - with increasing time - stochastically changing to the virus-
153
productive state, see Fig. 3.
CR IP T
149
154
To understand the effect of Tat inhibitors in the HIV intracellular dynam-
156
ics, we performed a set of simulations where the inhibitors were introduced
157
in two approaches. The first one represents the effect of having a constant
158
inhibitor in the system, and the second one represents the effect of applying
159
the inhibitor for an intermediate time of 8 days and then removing it for
160
the rest of the total simulation time. For those experiments, we used the
161
intracellular model (reactions (1) to (7)), and we observed the effect of the
162
different classes of inhibitors. Modeling the inhibitors in the system requires
163
binding and unbinding constants that were initially adjusted to obtain the
164
latent state. All simulations were performed using parameters values given
165
in Table 2. To model the effect of CI in the system, we introduced reaction
166
(8). We observed that CI applied at a constant concentration delays Tat
167
expression, but when it is removed, the Tat expression recovers a dynamics
CE
PT
ED
M
AN US
155
similar to a system without inhibitor, see Fig. 4A. Subsequently, we studied
169
the effect of NCI, and we found that a constant application of NCI reduced
170
most transitions to the virus-producing state. Removing NCI restores Tat
171
expression, see Fig. 4B. Then, we analyzed the effect of II in the system
AC 168
12
ACCEPTED MANUSCRIPT
through the introduction of reactions (11) and (12). Our results showed that
173
II constantly applied reduces the transitions to the virus-producing state.
174
Removing II restores Tat expression, see Fig. 4C. Finally, we studied the
175
effect of dCA in the system by using reaction (13). We observed that dCA
176
applied at a constant concentration dramatically impairs the virus expression
177
which is observed in the total inhibition of transitions to the virus-producing
178
state. Removing dCA from the system affects the observed permanent in-
179
hibitory effect. A permanent inhibition was observed for most of the cells,
180
but a small fraction (6 cells) were able to restart a productive infection, see
181
Fig. 4D.
AN US
CR IP T
172
182
As a further measure of inhibitor efficiency, we calculated the switching time
184
between the latent state and the virus-producing state. We observed that
185
in a simulation with no inhibitor this transition time is characterized by a
186
distribution with a peak close to the first three days of infection. NCI and
187
dCA deform this original distribution to a long-tail shaped distribution with
188
much smaller peaks during the first days of infection. For CI and II it was
189
obtained an almost uniform distribution with low probability of transition.
190
This result shows that Tat inhibitors act by reducing the total number of
CE
PT
ED
M
183
the transitions to the producing state during initial days of the infection and
192
keep a small probability of transition for long periods, see Fig. 5.
AC 191
193
194
It is important to remark that the results obtained maybe be parameter de13
ACCEPTED MANUSCRIPT
pendent. For this reason, we performed a parameter scan using ranges of
196
parameters one order of magnitude above and one order of magnitude be-
197
low the nominal values given in Table 2. In this simulation, we varied the
198
parameters and inhibitor concentrations and reported the average Tat con-
199
centration in the virus-producing state. Three independent simulations were
200
performed with 1000 trajectories representing individual cells. With this, we
201
proved the robustness of our model predictions and corroborated that the re-
202
sults obtained are consistent even using wide ranges of parameters. In Fig. 6
203
A-C, it can be observed the effect of Tat inhibitors applied in temporal (left
204
panel) and constant (right panel) schemes on the average expression of the
205
Tat protein. The results prove that CI, NCI, and II applied in a temporal
206
scheme result in a recovery of Tat expression after the inhibitor is removed,
207
and only a permanent inhibition in Tat protein expression is observed by
208
constantly applying those inhibitors. On the contrary, dCA showed a more
209
permanent Tat inhibition during the constant application scheme, and only
210
HIV expression was observed under the lowest concentration of dCA (see
211
Fig. 6 D).
AN US
M
ED
PT
CE
212
CR IP T
195
4. Discussion
214
The greatest barrier to eradicate HIV infection is the persistence of HIV la-
215
tent infected cells. Two main strategies have been discussed for achieving a
216
functional cure: The shock and kill strategy [26] and the less explored perma-
AC
213
14
ACCEPTED MANUSCRIPT
nent HIV suppression [25]. So far, the shock and kill strategy is being actively
218
studied and even clinical trails have been considered [33]. In contrast, per-
219
manent HIV suppression still faces multiple challenges, the most important
220
being the finding of stronger and permanent virus-suppressing agents [11]. In
221
this project, we presented a mathematical study to understand the potential
222
role of Tat inhibitors as virus-suppressing agents. In the presented model,
223
we combined the well-known feedback circuits in HIV gene expression with
224
the effect of different classes of Tat inhibitors. Our modeling strategy is sim-
225
plistic enough to allow us to thoroughly understand the general dynamics of
226
HIV gene expression and those of cells in latent and productive states. Our
227
results suggest that diverse compounds with different mechanisms of inhibi-
228
tion can achieve a substantial reduction in the virus-producing state.
M
AN US
CR IP T
217
229
Summarizing our simulation results, it is possible to detect the main differ-
231
ences between the studied Tat inhibitors. Removing Tat protein from the
232
cell via CI, NCI or II results in a decrease in the overall HIV gene expression.
233
Nevertheless, those compounds do not affect the viral genome integrated into
234
the host genome. Thus the virus maintains its potential to “re-awake” after
235
those inhibitors are removed. This re-awakening capacity can be understood
CE
PT
ED
230
at a system level by the low viral transcription that can be given even in
237
total absence of Tat [27]. This basal transcription rate eventually restores
238
the positive feedback loop in the system, by a slow but constant increase in
239
Tat concentration. In contrast, dCA by promoting epigenetic modifications
AC 236
15
ACCEPTED MANUSCRIPT
disrupts the integrated viral genome resulting in the loss of this re-awakening
241
capacity [25]. The disruption of the integrated viral genome does not affect
242
Tat protein instantaneously, but the virus-producing state is removed, re-
243
sulting in the permanent extinction of the viral gene products. Our results
244
suggest that a putative therapy with CI, NCI, and II would result in re-
245
versible dynamics that could cause an ineffective scenario as that obtained
246
with current HAART in latency [28]. On the contrary, we predict that a con-
247
stant application of dCA can achieve a permanent inhibition. Nevertheless,
248
it is important to remark the difference between a permanent Tat inhibi-
249
tion and total viral eradication. Our results show that dCA can achieve a
250
permanent Tat inhibition, but not an HIV-eradication. Analyzing Fig. 4D
251
it can be observed that some trajectories transition to the productive state
252
even after dCA application. Those cells escaped the effect of dCA because
253
this inhibitor needs to interact with an active Tat-LTR complex. This means
254
that dCA can only achieve total eradication of the virus if it is applied for
255
a period longer than the last latent-productive transition event. dCA is a
256
prominent candidate to achieve long-term Tat inhibition [25, 24].
258
AN US
M
ED
PT
CE
257
CR IP T
240
During the peer-review process of this paper, a similar work from Cao’s group was published, they studied the probability landscape of the Tat-circuit un-
260
der different perturbations [5]. Comparing their results with our findings, we
261
observed consistent outcomes for a ’shock and kill’ strategy and permanent
262
inhibition.
AC 259
16
ACCEPTED MANUSCRIPT
263
5. Conclusions
265
Overall, our results showed that removing Tat protein from the system via
266
inhibitors resulted in a temporary and reversible viral suppression. In con-
267
trast, compounds that interact with Tat protein and disrupt the integrated
268
viral genome produced a more permanent viral suppression. Future experi-
269
ments are still needed to substantiate dCA significance in latency, i.e., how
270
strong and for how long those epigenetic modifications would affect the HIV
271
expression. Improving dCA understanding in Tat inhibition could give light
272
in the design of novel therapeutic interventions to achieve the long searched
273
functional cure.
274
6. Funding
275
LUA has been supported by postdoctoral grant 263795 from Consejo Na-
276
cional de Ciencia y Tecnolog´ıa (CONACYT).
277
7. Acknowledgments
278
We gratefully thank Prof. Dr. Ursula Kummer and Christoph Zimmer
AC
CE
PT
ED
M
AN US
CR IP T
264
279
for her helpful comments to develop the biological model and revising the
280
manuscript. Finally, we thank Jocelyn Faberman for proofreading this manuscript.
17
ACCEPTED MANUSCRIPT
Figure captions
CR IP T
281
AC
CE
PT
ED
M
AN US
Figure 1: Graphical representation of the processes involved in HIV gene expression. HIV gene expression is given by viral and cellular machinery. HIV gene transcription without Tat protein results in abortive pauses that occur by the formation of a secondary structure (TAR) in the nascent viral transcripts. Tat protein acts as a transcriptional factor binding to the newly synthesized transcripts. This relaxes the abortive secondary structure in the nascent RNA allowing the expression of complete viral transcripts. Complete viral transcripts will produce all the viral elements starting a productive infection. One of those viral products is the same Tat protein. Tat protein forms a positive feedback circuit in the expression of HIV. This last process is termed transactivation.
18
ACCEPTED MANUSCRIPT
Table 1: Initial conditions
CR IP T
Initial Value 0 1 0 0 0
AC
CE
PT
ED
M
AN US
Variable LT ROF F LT RON mRN A T at LT RON −T at
19
ACCEPTED MANUSCRIPT
Table 2: Parameter values Description Nominal Range Promoter activation constant 1 × 10−4 fixed Promoter inactivation constant 1 × 10−2 fixed Transcription rate 1 × 10−1 fixed Translation rate 10 fixed Tat binding to promoter rate 1 × 10−2 fixed Tat unbinding to promoter rate 1 × 10−2 fixed Degradation rate of mRNA 1 fixed Degradation rate of Tat 0.125 fixed Transactivation rate 5 fixed
kbCI kubCI kbN CI kubN CI k2bN CI k2ubN CI kbII δII/T at kdCA
Binding constant for CI Unbinding constant for CI Binding constant for NCI Unbinding constant for NCI 2nd binding constant for NCI 2nd unbinding constant for NCI Irreversible binding by II Degradation of II/Tat Irreversible binding by dCA
CI N CI II dCA
Competitive Inhibitor Noncompetitive Inhibitor Irreversible Inhibitor Didehydro-Cortistatin-A
[1 × 10−4 fixed [1 × 10−4 fixed fixed fixed [1 × 10−4 fixed [1 × 10−4
, 1 × 10−2 ] Cell/(Molecules*h) 1/h , 1 × 10−2 ] Cell/(Molecules*h) 1/h Cell/(Molecules*h) 1/h , 1 × 10−2 ] Cell/(Molecules*h) Cell/(Molecules*h) , 1 × 10−2 ] Cell/(Molecules*h)
AN US
M
AC
CE
PT
ED
TS = This study.
1 × 10−3 1 × 10−5 1 × 10−3 1 × 10−1 1 × 10−4 1 × 10−1 1 × 10−3 1 × 10−3 1 × 10−3
Units 1/h 1/h 1/h 1/h 1/h 1/h 1/h 1/h 1/h
CR IP T
Parameter kON kOf f km kT at kbind kunbind δm δT at ktrans
20
1 × 103 1 × 103 1 × 103 1 × 103
[10 [10 [10 [10
, , , ,
2 × 103 ] 2 × 103 ] 2 × 103 ] 2 × 103 ]
Molecules/Cell Molecules/Cell Molecules/Cell Molecules/Cell
Ref. [31] [31] [31] [31] [31] [31] [31] [31] [31] TS TS TS TS TS TS TS TS TS TS TS TS TS
CE
PT
ED
M
AN US
CR IP T
ACCEPTED MANUSCRIPT
AC
Figure 2: Representation of the effect of different classes of Tat inhibitors. Multiple compounds have been described to inhibit Tat-dependent HIV transcription. Here we categorize them according to their mechanism of action: A) Competitive Inhibitors (CI) that reversibly bind to the LTR for the active site of Tat. B) Non-competitive Inhibitors (NCI) that reversibly bind to elements that are indispensable for Tat-dependent transcription. C) Irreversible Inhibitors (II) that promote proteasomal degradation of the Tat protein. D) Didehydro-Cortistatin-A (dCA) binds to the Tat protein and promotes the establishment of epigenetic modifications in the viral promoter.
21
M
AN US
CR IP T
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
Figure 3: Simulations for the intracellular model. A) Stochastic simulations were computed using reactions 1 to 7, initial conditions given in Table 1 and parameter values given in Table 2. 1,000 individual trajectories that represent single cell dynamics are shown for a simulated time-span of 24 days. A representative trajectory that presents a single cell dynamics is given by the blue line. At the right-hand side, a histogram of the final trajectories is given. Red line represents the threshold used to consider cells in the virus-producing state, the number of cells in each state is given next to the histogram.
22
AC
CE
PT
ED
M
AN US
CR IP T
ACCEPTED MANUSCRIPT
Figure 4: Effect of Tat inhibitors in Tat intracellular dynamics. The effect of different inhibitors in Tat expression was computed using 1,000 individual stochastic simulations. In the left panel, the inhibitors were applied for 8 days of simulation time and then removed from the simulation (the time of inhibitor application is inside the vertical black lines). In the right panel, the inhibitors were continuously applied during the 24 days of simulation time. A) Stochastic simulations with CI. B) Stochastic simulations with NCI. C) Stochastic simulations with 23 II. D) Stochastic simulations with dCA. The parameters used for the simulations are given in Table 2.
ED
M
AN US
CR IP T
ACCEPTED MANUSCRIPT
AC
CE
PT
Figure 5: Distribution of the switching time between latent and virus-producing states. Simulations were performed for 1,000 individual cells. The time when each cell transitioned between latent state to virus-producing state was recorded and used to build a probability distribution. The different distributions represent simulations for the system with no inhibitor and the corresponding inhibitor as given in the legend.
24
AC
CE
PT
ED
M
AN US
CR IP T
ACCEPTED MANUSCRIPT
Figure 6: Effect of varying parameters in Tat expression. In the left panel, the inhibitors were applied for 8 days of simulation time and then removed from the simulation. In the right panel, the inhibitors were continuously applied during the 24 days of simulation time. Simulations were computed using the initial conditions given in Table 1 and parameter ranges given in Table 2. Constants and inhibitor concentrations were varied according to the figure axis. Colormap represents the total concentration of Tat protein (Molecules/Cell) at the end of 24 days of simulation time. A) HIV intracellular dynamics 25NCI. C) HIV intracellular dynamics with II. with CI. B) HIV intracellular dynamics with D) HIV intracellular dynamics with dCA. Three independent simulations were performed with 1000 trajectories representing individual cells.
ACCEPTED MANUSCRIPT
282
References [1] Andrey Arzumanov, Dmitry A Stetsenko, Andrey D Malakhov, Ste-
284
fanie Reichelt, Mads D Sørensen, B Ravindra Babu, Jesper Wen-
285
gel, and Michael J Gait.
286
tion of hiv-1 tat-dependent trans-activation by mixmer 2’-o-methyl
287
oligoribonucleotides containing locked nucleic acid (lna), α-l-lna, or
288
2’-thio-lna residues.
289
10.1089/154545703322860762.
CR IP T
283
A structure-activity study of the inhibi-
AN US
Oligonucleotides, 13(6):435–453, 2003.
doi:
290
[2] G´abor Bal´azsi, Alexander van Oudenaarden, and J James Collins. Cel-
291
lular decision making and biological noise: from microbes to mam-
292
mals.
293
10.1016/j.cell.2011.01.030.
ISSN 1097-4172.
doi:
ED
M
Cell, 144(6):910–25, March 2011.
[3] Sebastian Biglione, Sarah A Byers, Jason P Price, Van Trung Nguyen,
295
Olivier Bensaude, David H Price, and Wendy Maury. Inhibition of hiv-1
296
replication by p-tefb inhibitors drb, seliciclib and flavopiridol correlates
297
with release of free p-tefb from the large, inactive form of the complex. Retrovirology, 4(1):47, 2007. doi: 10.1186/1742-4690-4-47.
[4] CA Bohan, F Kashanchi, B Ensoli, L Buonaguro, KA Boris-Lawrie, and
AC
299
CE
298
PT
294
300
JN Brady. Analysis of tat transactivation of human immunodeficiency
301
virus transcription in vitro. Gene expression, 2(4):391–407, 1991.
302
[5] Youfang Cao, Xue Lei, Ruy M Ribeiro, Alan S Perelson, and Jie Liang. 26
ACCEPTED MANUSCRIPT
Probabilistic control of hiv latency and transactivation by the tat gene
304
circuit. Proceedings of the National Academy of Sciences, 115(49):12453–
305
12458, 2018. doi: 10.1073/pnas.1811195115.
CR IP T
303
[6] Sheng-Hao Chao, Koh Fujinaga, Jon E Marion, Ran Taube, Ed-
307
ward A Sausville, Adrian M Senderowicz, B Matija Peterlin, and
308
David H Price. Flavopiridol inhibits p-tefb and blocks hiv-1 replica-
309
tion. Journal of Biological Chemistry, 275(37):28345–28348, 2000. doi:
310
10.1074/jbc.C000446200.
AN US
306
[7] Tae-Wook Chun, Lucy Carruth, Diana Finzi, Xuefei Shen, Joseph A.
312
DiGiuseppe, Harry Taylor, Monika Hermankova, Karen Chadwick,
313
Joseph Margolick, Thomas C. Quinn, Yen-Hong Kuo, Ronald Brook-
314
meyer, Martha A. Zeiger, Patricia Barditch-Crovo, and Robert F. Sili-
315
ciano. Quantification of latent tissue reservoirs and total body viral load
316
in HIV-1 infection. Nature, 387(6629):183–188, November 1997. ISSN
317
0028-0836. doi: 10.1038/246170a0.
319
ED
PT
hiv latency: the road to an hiv cure. Annual review of medicine, 66:407, 2015. doi: 10.1146/annurev-med-092112-152941.
AC
320
[8] Matthew Dahabieh, Emilie Battivelli, and Eric Verdin. Understanding
CE
318
M
311
321
[9] Daniel A Donahue, Bj¨orn D Kuhl, Richard D Sloan, and Mark A Wain-
322
berg. The viral protein tat can inhibit the establishment of hiv-1 latency.
323
Journal of virology, 86(6):3253–3263, 2012. doi: 10.1128/JVI.06648-11.
27
ACCEPTED MANUSCRIPT
[10] Diana Finzi, Monika Hermankova, Theodore Pierson, Lucy M Carruth,
325
Christopher Buck, Richard E Chaisson, Thomas C Quinn, Karen Chad-
326
wick, Joseph Margolick, Ronald Brookmeyer, et al. Identification of a
327
reservoir for hiv-1 in patients on highly active antiretroviral therapy. Sci-
328
ence, 278(5341):1295–1300, 1997. doi: 10.1126/science.278.5341.1295.
330
[11] Robert C Gallo. Shock and kill with caution. Science, 354(6309):177– 178, 2016. doi: 10.1126/science.aaf8094.
AN US
329
CR IP T
324
[12] Daniel T Gillespie. A general method for numerically simulating the
332
stochastic time evolution of coupled chemical reactions. Journal of Com-
333
putational Physics, 22(4):403–434, December 1976. ISSN 00219991. doi:
334
10.1016/0021-9991(76)90041-3.
M
331
[13] Irene Guendel, Sergey Iordanskiy, Rachel Van Duyne, Kylene Kehn-Hall,
336
Mohammed Saifuddin, Ravi Das, Elizabeth Jaworski, Gavin C Sampey,
337
Svetlana Senina, Leonard Shultz, et al. Novel neuroprotective gsk-3β
338
inhibitor restricts tat-mediated hiv-1 replication. Journal of virology, 88
339
(2):1189–1208, 2014. doi: 10.1128/JVI.01940-13.
CE
PT
ED
335
[14] Stefan Hoops, Sven Sahle, Ralph Gauges, Christine Lee, J¨ urgen Pahle,
341
Natalia Simus, Mudita Singhal, Liang Xu, Pedro Mendes, and Ursula
AC
340
342
Kummer. COPASI–a COmplex PAthway SImulator. Bioinformatics
343
(Oxford, England), 22(24):3067–74, December 2006. ISSN 1367-4811.
344
doi: 10.1093/bioinformatics/btl485.
28
ACCEPTED MANUSCRIPT
345
346
[15] Brian Ingalls. Mathematical modelling in systems biology: An introduction. Internet.[cited at p. 117], 2013. [16] Julie K Jadlowsky, Masanori Nojima, Antje Schulte, Matthias Geyer,
348
Takashi Okamoto, and Koh Fujinaga. Dominant negative mutant cy-
349
clin t1 proteins inhibit hiv transcription by specifically degrading tat.
350
Retrovirology, 5(1):63, 2008. doi: 10.1186/1742-4690-5-63.
CR IP T
347
[17] Kylene Kehn-Hall, Irene Guendel, Lawrence Carpio, Leandros Skaltsou-
352
nis, Laurent Meijer, Lena Al-Harthi, Joseph P Steiner, Avindra Nath,
353
Olaf Kutsch, and Fatah Kashanchi. Inhibition of tat-mediated hiv-1
354
replication and neurotoxicity by novel gsk3-beta inhibitors. Virology,
355
415(1):56–68, 2011. doi: 10.1016/j.virol.2011.03.025.
M
AN US
351
[18] Kara Lassen, Yefei Han, Yan Zhou, Janet Siliciano, and Robert F Sili-
357
ciano. The multifactorial nature of HIV-1 latency. Trends in molec-
358
ular medicine, 10(11):525–31, November 2004. ISSN 1471-4914. doi:
359
10.1016/j.molmed.2004.09.006.
361
PT
Lori, Andrey Louie, Phillip Markham, John Rossi, Marvin Reitz, and Robert C Gallo. Inhibition of human immunodeficiency virus type 1
AC
362
[19] Julianna Lisziewicz, Daisy Sun, Jason Smythe, Paolo Lusso, Franco
CE
360
ED
356
363
replication by regulated expression of a polymeric tat activation re-
364
sponse rna decoy as a strategy for gene therapy in aids. Proceedings
365
of the National Academy of Sciences, 90(17):8000–8004, 1993.
366
10.1073/pnas.90.17.8000. 29
doi:
ACCEPTED MANUSCRIPT
368
369
[20] Alessandro Marcello. Latency: the hidden HIV-1 challenge. Retrovirology, 3:7, January 2006. ISSN 1742-4690. doi: 10.1186/1742-4690-3-7. [21] David M Margolis. Mechanisms of hiv latency: an emerging picture
370
of complexity.
371
10.1007/s11904-009-0033-9.
CR IP T
367
Current HIV/AIDS Reports, 7(1):37–43, 2010.
doi:
[22] David M Margolis, J Victor Garcia, Daria J Hazuda, and Barton F
373
Haynes. Latency reversal and viral clearance to cure hiv-1. Science, 353
374
(6297):aaf6517, 2016. doi: 10.1126/science.aaf6517.
AN US
372
[23] Guillaume Mousseau and Susana Valente. Strategies to block hiv tran-
376
scription: focus on small molecule tat inhibitors. Biology, 1(3):668–697,
377
2012. doi: 10.3390/biology1030668.
M
375
[24] Guillaume Mousseau and Susana T Valente. Didehydro-cortistatin a: a
379
new player in hiv-therapy? Expert Review of Anti-infective Therapy, 14
380
(2):145–148, 2016. doi: 10.1586/14787210.2016.1122525.
PT
ED
378
[25] Guillaume Mousseau, Cari F Kessing, R´emi Fromentin, Lydie Traut-
382
mann, Nicolas Chomont, and Susana T Valente. The tat inhibitor didehydro-cortistatin a prevents hiv-1 reactivation from latency. MBio, 6(4):e00465–15, 2015. doi: 10.1128/mBio.00465-15.
AC
383
CE
381
384
385
[26] Geetha H Mylvaganam, Guido Silvestri, and Rama Rao Amara. Hiv
386
therapeutic vaccines: moving towards a functional cure. Current opinion
387
in immunology, 35:1–8, 2015. doi: doi: 10.1016/j.coi.2015.05.001. 30
ACCEPTED MANUSCRIPT
[27] Gary Nabel and David Baltimore. An inducible transcription factor
389
activates expression of human immunodeficiency virus in T cells. Nature,
390
326(711-713), 1987. doi: 10.1038/326711a0.
CR IP T
388
[28] S Palmer, F Maldarelli, A Wiegand, B Bernstein, G J Hanna, S C Brun,
392
D J Kempf, J W Mellors, J M Coffin, and M S King. Low-level viremia
393
persists for at least 7 years in patients on suppressive antiretroviral ther-
394
apy. Proc Natl Acad Sci USA, 105(10):3879–3884, 2008. doi: 0800050105
395
[pii]10.1073/pnas.0800050105.
396
397
AN US
391
[29] Vicente Planelles. An ounce of tat prevention is worth a pound of functional cure. mBio, 6(4):e01269–15, 2015. doi: 10.1128/mBio.01269-15. [30] Brandon S Razooky and Leor S Weinberger. Mapping the architecture
399
of the HIV-1 Tat circuit: A decision-making circuit that lacks bistability
400
and exploits stochastic noise. Methods (San Diego, Calif.), 53(1):68–77,
401
January 2011. ISSN 1095-9130. doi: 10.1016/j.ymeth.2010.12.006.
402
[31] Brandon S Razooky, Anand Pai, Katherine Aull, Igor M Rouzine, and
403
Leor S Weinberger. A hardwired hiv latency program. Cell, 160(5):
ED
PT
CE
990–1001, 2015. doi: 10.1016/j.cell.2015.02.009.
[32] Herbert M. Sauro. Enzyme kinetics for systems biology. Internet.[cited
AC
404
M
398
405
406
at p. 129], 2012.
407
[33] Janet D Siliciano and Robert F Siliciano. Recent developments in the
408
search for a cure for hiv-1 infection: targeting the latent reservoir for 31
ACCEPTED MANUSCRIPT
409
hiv-1. Journal of Allergy and Clinical Immunology, 134(1):12–19, 2014.
410
doi: 10.1016/j.jaci.2014.05.026. [34] Rachel Van Duyne, Irene Guendel, Elizabeth Jaworski, Gavin Sampey,
412
Zachary Klase, Hao Chen, Chen Zeng, Dmytro Kovalskyy, Mahmoud H
413
El Kouni, Benjamin Lepene, et al. Effect of mimetic cdk9 inhibitors
414
on hiv-1-activated transcription. Journal of molecular biology, 425(4):
415
812–829, 2013. doi: 10.1016/j.jmb.2012.12.005.
AN US
CR IP T
411
[35] Zhitao Wan and Xulin Chen. Triptolide inhibits human immunodefi-
417
ciency virus type 1 replication by promoting proteasomal degradation
418
of tat protein. Retrovirology, 11(1):88, 2014. doi: 10.1186/s12977-014-
419
0088-6.
M
416
[36] Leor S Weinberger and Thomas Shenk. An HIV feedback resistor: auto-
421
regulatory circuit deactivator and noise buffer. PLoS biology, 5(1):e9,
422
January 2007. ISSN 1545-7885. doi: 10.1371/journal.pbio.0050009.
423
[37] Leor S Weinberger, John C Burnett, Jared E Toettcher, Adam P
424
Arkin, and David V Schaffer. Stochastic gene expression in a lentivi-
PT
ral positive-feedback loop: HIV-1 Tat fluctuations drive phenotypic diversity.
AC
426
CE
425
ED
420
427
428
Cell, 122(2):169–82, July 2005.
ISSN 0092-8674.
doi:
10.1016/j.cell.2005.06.006.
[38] Samuel A. F. Williams and Warner C Greene. Host factors regulating
32
ACCEPTED MANUSCRIPT
429
post-integration latency of HIV. Trends in microbiology, 13(4):137–9,
430
April 2005. ISSN 0966-842X. doi: 10.1016/j.tim.2005.02.006. [39] Jessica E Wynn and Webster L Santos. Hiv-1 drug discovery: targeting
432
folded rna structures with branched peptides. Organic & biomolecular
433
chemistry, 13(21):5848–5858, 2015. doi: 10.1039/C5OB00589B.
AC
CE
PT
ED
M
AN US
CR IP T
431
33