The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne

PERSISTENCE OF HIV-1 TRANSCRIPTION IN PERIPHERAL-BLOOD

MONONUCLEAR CELLS IN PATIENTS RECEIVING POTENT
ANTIRETROVIRAL THERAPY

MANOHAR R. FURTADO, PH.D., DUNCAN S. CALLAWAY, B.S., JOHN P. PHAIR, M.D., KEVIN J. KUNSTMAN, B.S.,
JENNIFER L. STANTON, B.S., CATHERINE A. MACKEN, PH.D., ALAN S. PERELSON, PH.D., AND STEVEN M. WOLINSKY, M.D.

ABSTRACT function, delays progression of disease, and prolongs

Background and Methods Although potent anti- survival.7,8 Despite the apparent success of antiretro-
retroviral therapy can control infection with human viral therapy in suppressing plasma HIV-1 RNA for
immunodeficiency virus type 1 (HIV-1), a long-lived long periods, a long-lived reservoir of infectious virus
reservoir of infectious virus persists in CD4+ T cells. remains in CD4+ T cells and perhaps other types of
We investigated this viral reservoir by measuring the cells, suggesting the need for continued long-term
levels of cell-associated viral DNA and messenger treatment.9-12 It is important to determine the extent
RNA (mRNA) that are essential for HIV-1 replication. of this reservoir of HIV-1, whether it is being re-
Approximately every 6 months, we obtained sam-
ples of peripheral-blood mononuclear cells from five
newed, and the length of time that cells containing
men with long-standing HIV-1 infection who had had replication-competent HIV-1 proviral DNA remain.
undetectable levels of plasma HIV-1 RNA for 20 Quantitative assessments of HIV-1 and HIV-1–
months or more during treatment with potent anti- infected cells in blood and lymphoid tissue after treat-
retroviral drugs. ment have provided information on the kinetics of
Results Before treatment, plasma levels of HIV-1 viral replication and clearance in vivo and the rapidity
RNA correlated with the levels of cell-associated un- of the turnover of affected cells and virus.1,2,4-6,12,13
integrated HIV-1 DNA and unspliced viral mRNA. Mathematical modeling suggests that the decrease in
After treatment, plasma levels of HIV-1 RNA fell by plasma HIV-1 RNA levels in response to combina-
more than 2.7 log to undetectable levels. The de- tions of antiretroviral drugs that block new rounds of
crease in cell-associated integrated and unintegrated infection occurs in two phases.13 In the first phase,
HIV-1 DNA and mRNA occurred in two phases. The
first phase occurred during the initial 500 days of
as drug therapy extinguishes viral replication, levels
treatment and was characterized by substantial de- of viral RNA in plasma rapidly diminish and CD4+
creases in the levels of DNA and mRNA, but not to T cells infected with actively replicating HIV-1 die.1,2
undetectable levels. The concentrations of cell-associ- The second phase is marked by a slower rate of de-
ated unintegrated viral DNA, integrated proviral DNA, crease in plasma HIV-1 RNA, reflecting the presence
and unspliced viral mRNA decreased by 1.25 to 1.46 of long-lived infected CD4+ T cells that contain rep-
log. The second phase occurred during the subse- lication-competent virus in unintegrated and integrat-
quent 300 days or more of treatment and was char- ed forms, as well as free virus associated with lymph-
acterized by a plateau in the levels of HIV-1 DNA and oid-tissue follicular dendritic cells.6-13
unspliced mRNA. After an initial rapid decline, the Much of the integrated proviral DNA within
ratio of unspliced to multiply spliced viral mRNA (a
measure of active viral transcription) stabilized and
CD4+ T cells is unable to replicate. Replication-
remained greater than zero at each measurement. competent virus, however, persists in long-lived rest-
Conclusions Despite treatment with potent antiret- ing memory CD4+ T cells despite one or two years
roviral drugs and the suppression of plasma HIV-1 of apparently effective, potent antiretroviral thera-
RNA to undetectable levels for 20 months or more, py.9-11 HIV-1 may also be derived from chronically
HIV-1 transcription persists in peripheral-blood mono- infected cells, including CD4+ T cells and possibly
nuclear cells. Unless the quasi–steady state levels of macrophages, that were infected before therapy was
HIV DNA and mRNA eventually disappear with longer initiated and continue to survive and produce virus.5
periods of therapy, these findings suggest that HIV-1 Treatment with a protease inhibitor should render
infection cannot be eradicated with current treat- most newly produced virions noninfectious. None-
ments. (N Engl J Med 1999;340:1614-22.)
theless, some infectious particles may still be gener-
©1999, Massachusetts Medical Society.
ated. Consequently, it is not known to what extent
the stable viral reservoir represents infected cells that

C
OMBINATIONS of drugs that inhibit viral
reverse transcriptase and protease control
From the Departments of Pathology (M.R.F.) and Medicine (J.P.P.,
infection with human immunodeficiency K.J.K., J.L.S., S.M.W.), Northwestern University Medical School, Chicago;
virus type 1 (HIV-1) in many people by re- and the Theoretical Division, Los Alamos National Laboratory, Los Alamos,
ducing the levels of viral RNA in plasma and depleting N.M. (D.S.C., C.A.M., A.S.P.). Address reprint requests to Dr. Wolinsky at
the Division of Infectious Diseases, Department of Medicine, Tarry 3-735,
the pools of virus in lymphoid tissue.1-6 This sus- Northwestern University Medical School, 303 E. Chicago Ave., Chicago,
tained reduction in viral replication improves immune IL 60611, or at [email protected].

1614 · May 2 7 , 19 9 9

The New England Journal of Medicine

Downloaded from nejm.org at TEXAS STATE UNIV on May 10, 2013. For personal use only. No other uses without permission.
Copyright © 1999 Massachusetts Medical Society. All rights reserved.
 PERSISTENCE OF HIV-1 TRANSCRIP TION IN PATIENTS RECEIVING POTENT ANTIRETROVIRAL THERAPY

are replenished by ongoing low-level replication, AGAGCCAAC3') and pol-R outer primer (nucleotides 2205 to
chronically infected CD4+ T cells, or resting mem- 2234; 5'ACAGCTGGCTACTATTTCTTTTGCTACTA3') and a
pol-F inner primer (nucleotides 69 to 90; 5'CAACAGCCCCACC-
ory CD4+ T cells with replication-competent, inte- AGAAGAGA3') and a pol-R inner primer (nucleotides 1952 to
grated proviral DNA.10,11 Therefore, we examined 1976; 5'GATCTGGTTGTGCTTGAATGATT-C3'). The positions
the characteristics of the reservoir of HIV-1 in pe- of the oligonucleotide primers are numbered according to the pol
ripheral-blood mononuclear cells of patients who were gene of the HXB2 isolate.19 After extraction and amplification,
the DNA was sequenced and analyzed with a sequencing system
receiving potent antiretroviral therapy. (Prism 377, Applied Biosystems, Foster City, Calif.) as described
previously.18 Our assay allowed us to detect mutations coding for
METHODS drug resistance in at least 25 percent of the viral population.
Study Design and Subjects
Statistical Analysis
Over a period of up to 31 months, we measured the levels of
cell-associated viral DNA and viral messenger RNA (mRNA) and To assess the correlation between pairs of variables for individual
tracked changes in the nucleotide sequences of the HIV-1 pol gene patients and between pairs of measurements in all patients at base
that occurred in concert with continued suppression of plasma line, we calculated the linear correlation coefficient for pairs of
HIV-1 RNA to undetectable levels in blood samples from five measurements obtained at different times. The data obtained dur-
HIV-1–infected men. All five patients had confirmed HIV-1 in- ing the course of therapy were log-transformed before analysis with
fection, were enrolled in a multicenter study of the acquired im- a paired, two-sided t-test.20 To calculate the rates of decrease in
munodeficiency syndrome,14 and were selected on the basis of the amount of viral DNA and HIV-1 mRNA, we visually inspect-
their compliance with a potent multidrug regimen and the con- ed the data points and, when appropriate, separated them into
tinued suppression of plasma HIV-1 RNA to undetectable levels two groups: one representing the early phase of the decrease and
(<50 RNA copies per milliliter). Plasma HIV-1 RNA levels were one representing the later phase.5 We estimated the rates for each
measured approximately every six months by a quantitative reverse- patient using least-squares regression analysis, beginning with the
transcriptase–polymerase-chain-reaction (RT-PCR) assay (Roche data point obtained closest to the initiation of treatment.5
Molecular Diagnostic Systems, Branchburg, N.J.). All patients pro-
vided written informed consent according to the guidelines of the RESULTS
human subjects–protection committee of Northwestern University. Characteristics of the Patients
Analysis of Cell-Associated Unintegrated Viral DNA The five patients ranged in age from 38 to 56
and Integrated Proviral DNA years. Four had been infected with HIV-1 for at least
Total viral DNA was measured by a quantitative PCR assay as 10 years before the study began, and one had been
described previously.15 To assess the ability of the cellular DNA infected for 2 years. The study began in February
in a sample to be amplified by PCR and to ensure that samples 1996 and continued until December 1998. Table 1
would yield roughly equal amounts of DNA before amplification,
we measured the region coding for the HLA-DQ a chain using
shows the clinical characteristics of the patients. All
a PCR assay.15 The numbers of proviral DNA molecules that were five patients had been receiving a potent triple-drug
randomly integrated into cell DNA during the replication cycle antiretroviral regimen that included two of four nucle-
were assessed by a semiquantitative nested PCR with the use of oside analogues (zidovudine, stavudine, didanosine,
internal control standards of modified target DNA.16 Covalently and lamivudine) — or a non-nucleoside reverse-
closed circular forms of viral DNA molecules containing either
one or two long terminal repeats were measured by a semiquan- transcriptase inhibitor (nevirapine) in the case of Pa-
titative PCR assay with the use of appropriate primers.17 The tient 2 — and a protease inhibitor (ritonavir, indin-
number of copies was determined by comparison with a dilution avir, or saquinavir) for 20 to 31 months (Table 1).
series of the target DNA standard. Duplicate tubes with no target Patient 1 had received no prior therapy. Patient 3
DNA were included in each assay to detect contamination. The
limit of sensitivity of each DNA assay was approximately 20 cop-
had received zidovudine and lamivudine, Patient 4
ies per 106 peripheral-blood mononuclear cells. had received stavudine and didanosine, and Patient
5 had received zidovudine before the institution of
Analysis of Cell-Associated Unspliced and potent antiretroviral therapy. Patient 2 had received
Multiply Spliced Viral mRNA zidovudine in the past but not immediately before
Levels of cell-associated HIV-1 mRNA were measured by an beginning combination antiretroviral-drug therapy.
internally controlled quantitative RT-PCR assay with use of ap-
propriate, precisely matched oligonucleotide primer pairs to iden- Plasma HIV-1 RNA Levels and CD4+ T-Cell Counts
tify unspliced viral mRNA and viral mRNA with multiple splices
encoding Tat, Rev, and Nef, as described previously.18 To adjust The mean level of plasma HIV-1 RNA before the
for the total cellular RNA and verify the integrity of the RNA in initiation of potent antiretroviral therapy was 64,237
each sample, we quantified human ribosomal protein S17 mRNA copies per milliliter (range, 11,720 to 54,973). After
by RT-PCR using appropriate primers.15 Tubes with no reverse
transcriptase and no target complementary DNA were included to
treatment, it decreased by an average (±SD) of more
detect contamination. The limit of sensitivity of the assay was ap- than 2.7±0.3 log to less than 50 copies per millili-
proximately 50 copies per 106 peripheral-blood mononuclear cells. ter. The average half-life of HIV-1 RNA in plasma
was 30±18.2 days before the levels became unde-
Sequencing of the HIV-1 pol Gene
tectable. Infrequent sampling of blood precluded us
We used direct sequencing of cell-associated viral DNA to as- from measuring the length of the previously report-
sess the frequency of mutations coding for drug resistance in the
reverse-transcriptase and protease regions of the HIV-1 pol gene.18
ed rapid first-phase decline in plasma HIV-1 RNA.5
Cell-associated viral DNA was amplified by nested PCR with a Nonetheless, a half-life of 30 days is in general
pol-F outer primer (nucleotides 52 to 73; 5'TCAGAGCAGACC- agreement with the half-life reported for the second

Vol ume 340 Numb e r 21 · 1615

The New England Journal of Medicine

Downloaded from nejm.org at TEXAS STATE UNIV on May 10, 2013. For personal use only. No other uses without permission.
Copyright © 1999 Massachusetts Medical Society. All rights reserved.
 The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne

Patient 1
TABLE 1. CHARACTERISTICS OF THE FIVE PATIENTS.
106 CD4+ T cellsF
HIV-1 RNA

CD4+ T Cells (no./mm3)

800

HIV-1 RNA (copies/ml)

PLASMA
TIME OF CD4+ HIV-1 105
DRUG REGIMEN MEASUREMENT* T CELLS RNA†
600
mo cells/mm3 copies/ml 104
Patient 1
400
None ¡2 415 11,720 103
Zidovudine, lamivudine, indinavir 5 493 <50
Zidovudine, lamivudine, indinavir 10 778 <50
Zidovudine, lamivudine, indinavir 16 797 <50 102 200
Zidovudine, lamivudine, indinavir 22 724 <50
Patient 2
0
None ¡1 247 24,321
Lamivudine, stavudine, nevirapine 5 158 <50

DNA or RNA (copies/106 PBMC)

Lamivudine, stavudine, nevirapine 11 301 <50 Unspliced RNAF
Lamivudine, stavudine, nevirapine 17 520 <50 105 Total DNAF 105
Lamivudine, stavudine, nevirapine 23 337 <50 Integrated DNAF
Patient 3 Circular DNA
Lamivudine, zidovudine ¡1 12 45,000 104 104
Lamivudine, zidovudine, ritonavir 5 131 541
Lamivudine, zidovudine, ritonavir 10 204 <50
Lamivudine, zidovudine, ritonavir 17 305 <50 103 103
Lamivudine, zidovudine, ritonavir 21 306 <50
Lamivudine, zidovudine, saquinavir 27 414 <50
Lamivudine, zidovudine, saquinavir 31 274 <50
102 102
Patient 4
Didanosine, stavudine ¡1 230 54,973
Lamivudine, stavudine, indinavir 5 141 <50
Lamivudine, stavudine, indinavir 11 262 <50
Lamivudine, stavudine, indinavir 17 352 <50
Lamivudine, stavudine, indinavir 23 493 <50
105 Total MS mRNAF 105
nef mRNAF
mRNA (copies/106 PBMC)

Patient 5
Zidovudine ¡5 449 22,720 tat mRNAF
Zidovudine, lamivudine, indinavir 1 601 <50 rev mRNAF
104 F 104
Zidovudine, lamivudine, indinavir 7 398 <50
Zidovudine, lamivudine, indinavir 13 517 <50
Zidovudine, lamivudine, indinavir 20 558 <50
103 103
*The times indicate the duration of potent antiretroviral treatment. Neg-
ative numbers indicate measurements made before the start of this therapy.
†Although the formal limit of detection is 50 copies per milliliter for the 102 102
RT-PCR that we used, values ranging from 0 to 20 copies per milliliter
were measured.

0 200 400 600 800 1000

phase of the decline.5,21 CD4+ T-cell counts ranged Treatment (days)
from 12 to 449 cells per cubic millimeter of blood
before treatment and slowly rose with the continued Figure 1. Effect of Treatment on the Levels of Viral DNA and
mRNA in Peripheral-Blood Mononuclear Cells from Patients 1,
suppression of plasma HIV-1 RNA (final range, 274 2, and 3.
to 724 cells per cubic millimeter) (Table 1). For each patient, the top panel shows the CD4+ T-cell count
and plasma level of HIV-1 RNA; the middle panel shows the lev-
Temporal Changes in Cell-Associated HIV-1 DNA els of unspliced mRNA, total viral DNA, integrated proviral
and Viral mRNA DNA, and circular forms of unintegrated viral DNA; and the bot-
Figure 1 shows the temporal changes in the mean tom panel shows the total levels of multiply spliced (MS) viral
mRNA, as well as the levels of nef mRNA, tat mRNA, and rev
levels of HIV-1 RNA in plasma, CD4+ T-cell counts, mRNA. The arrows indicate the first time points after the start
and the concentrations of cell-associated viral DNA of treatment with potent antiretroviral therapy at which data
and HIV-1 mRNA in Patients 1, 2, and 3. These data were recorded. The horizontal dotted lines indicate the limits of
are representative of the results for all the patients. detection of the assays. PBMC denotes peripheral-blood mono-
Once treatment was begun, there was a slow, two- nuclear cells.
phase decrease in the levels of integrated proviral
DNA, unintegrated proviral DNA, and unspliced vi-
ral mRNA that differed in timing and extent from
the previously described pattern of decrease in plasma

1616 · May 2 7 , 19 9 9

The New England Journal of Medicine

Downloaded from nejm.org at TEXAS STATE UNIV on May 10, 2013. For personal use only. No other uses without permission.
Copyright © 1999 Massachusetts Medical Society. All rights reserved.
 PERSISTENCE OF HIV-1 TRANSCRIP TION IN PATIENTS RECEIVING POTENT ANTIRETROVIRAL THERAPY

Patient 2 Patient 3
106 106

CD4+ T Cells (no./mm3)

CD4+ T Cells (no./mm3)

800 800
HIV-1 RNA (copies/ml)

HIV-1 RNA (copies/ml)

105 105
600 600
104 104

400 400
103 103

102 200 102 200

0 0
DNA or RNA (copies/106 PBMC)

DNA or RNA (copies/106 PBMC)

105 105 105 105

104 104 104 104

103 103 103 103

102 102 102 102

105 105 105 105

mRNA (copies/106 PBMC)

mRNA (copies/106 PBMC)

104 104 104 104

103 103 103 103

102 102 102 102

0 200 400 600 800 1000 0 100 300 500 700 900 1100
Treatment (days) Treatment (days)

HIV-1 RNA.5 The first phase occurred during the tration of integrated proviral DNA dropped by
initial 500 days of treatment and was characterized 1.46±0.65 log during the first phase, reflecting the
by substantial decreases in cell-associated HIV-1 DNA loss of a population of cells presumably containing
and unspliced viral mRNA. In contrast, there was both short-lived cells infected with actively replicat-
little change during the second phase: values remained ing virus and long-lived infected cells. The half-life
at a quasi–steady state for an additional 300 days or of the integrated proviral DNA ranged from 29 to
more (Fig. 1). 108 days (mean, 53) (Table 2). The phase 1 decrease
in proviral DNA was followed by stable levels in
Detection of Cell-Associated Integrated Proviral DNA phase 2, which ranged from 28 to 49 copies per 106
As shown in Table 2, before the initiation of po- peripheral-blood mononuclear cells.
tent antiretroviral treatment, the mean level of inte-
grated proviral DNA was 2687 copies per 106 pe- Detection of Cell-Associated Unintegrated Viral DNA
ripheral-blood mononuclear cells (range, 542 to Before treatment, the levels of cell-associated un-
4274). After the initiation of treatment, the concen- integrated viral DNA correlated with the plasma lev-

Vol ume 340 Numb e r 21 · 1617

The New England Journal of Medicine

Downloaded from nejm.org at TEXAS STATE UNIV on May 10, 2013. For personal use only. No other uses without permission.
Copyright © 1999 Massachusetts Medical Society. All rights reserved.
 The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne

TABLE 2. BASE-LINE AND FIRST-PHASE VALUES AND HALF-LIVES OF HIV-1 DNA AND mRNA
IN PERIPHERAL-BLOOD MONONUCLEAR CELLS.*

PATIENT NO. INTEGRATED PROVIRAL DNA UNINTEGRATED VIRAL DNA UNSPLICED VIRAL mRNA TOTAL MULTIPLY SPLICED VIRAL mRNA

BASE LINE FIRST PHASE BASE LINE FIRST PHASE BASE LINE FIRST PHASE BASE LINE FIRST PHASE

final half- final half- final half- final half-

value life value life value life value life

copies/106 PBMC days copies/106 PBMC days copies/106 PBMC days copies/106 PBMC days

1 4017 36 47 784 20 61 8,628 435 75 5,110 2516 447

2 2240 28 29 2360 30 29 50,342 235 50 12,868 395 236
3 4274 28 37 7624 63 39 38,671 <50 75 27,723 1489 183
4 2362 49 108 7324 30 76 65,231 425 100 41,452 2574 195
5 542 32 44 1672 22 28 10,724 <50 23 13,258 674 275
Mean 2687 35 53 3953 33 47 34,719 239 65 20,082 1530 267
±SD ±1516 ±8.7 ±31.5 ±3264 ±17.3 ±21.1 ±24,734 ±190 ±29.2 ±14,473 ±1011 ±106.8

*PBMC denotes peripheral-blood mononuclear cells.

els of HIV-1 RNA (r=0.98, P<0.002) and with the species over time (declines of 0.53±0.37, 0.56±0.58,
levels of cell-associated unspliced mRNA (r=0.83, and 0.66±0.35 log, respectively) (Fig. 1). The levels
P<0.08). During the first phase of the decrease, the of multiply spliced and unspliced HIV-1 mRNA spe-
concentration of unintegrated circular forms of viral cies were correlated in individual patients over time,
DNA dropped by 1.46±0.68 log (Fig. 1). The num- with correlation coefficients ranging from 0.90 to
bers of copies of unintegrated and integrated forms 0.98 (P<0.04). During phase 1, the half-life of mul-
of viral DNA were correlated in each patient over tiply spliced HIV-1 mRNA ranged from 183 to 447
time (r>0.97, P<0.01), suggesting that these forms days (mean, 267) (Table 2).
decrease at similar rates. Indeed, during the first We used the ratio of unspliced viral mRNA to
phase, the half-life of unintegrated forms of viral multiply spliced viral mRNA as a measure of active
DNA ranged from 28 to 76 days (mean, 47), similar viral transcription and as an indicator of the propor-
to that for integrated proviral DNA (Table 2). tion of cells that contained high levels of full-length
viral RNA. The higher the ratio, the greater the pro-
Detection of Cell-Associated Unspliced HIV-1 mRNA portion of cells that are producing unspliced HIV-1
The levels of cell-associated unspliced HIV-1 RNA, which is indicative of replicating virus. The
mRNA correlated with the plasma levels of HIV-1 decline in the ratio resembled the decline in inte-
RNA at base line (r=0.89, P<0.009). During phase grated forms of proviral DNA. Throughout the sec-
1, the concentration of unspliced HIV-1 mRNA de- ond phase, these ratios remained stable and greater
creased by an average of 1.25±0.90 log, with a half- than zero at each point (Fig. 2), despite accompany-
life ranging from 23 to 100 days (mean, 65) (Table ing decreases in the levels of both unspliced and
2). Subsequently, the concentration stabilized in the multiply spliced viral mRNA species. The ratios be-
absence of detectable levels of HIV-1 RNA in plasma. fore treatment and 20 months or more after the sup-
The levels of cell-associated unintegrated forms of pression of plasma HIV-1 RNA to undetectable lev-
viral DNA and unspliced viral mRNA were highly cor- els were significantly different (P<0.01).
related in each patient, with correlation coefficients
ranging from 0.76 to 0.98 (P<0.08). Identification of Mutations That Confer Drug Resistance
At base line, we found a single mutation (R211K)
Detection of Cell-Associated Multiply Spliced HIV-1 mRNA coding for resistance to zidovudine in the pol gene19
The rates of decline among individual cell-associ- of the cell-associated viral DNA in Patient 3, who
ated multiply spliced viral mRNA species were similar. had previously received zidovudine monotherapy.
As a measure of the accuracy of the measurements After 31 months of therapy, three additional muta-
for tat, rev, and nef mRNA species, we calculated sam- tions coding for resistance to zidovudine (M41L,
ple correlation coefficients for each measurement in D67N, and L210W)18 were found in the pol gene in
individual patients over time and found correlation Patient 3. No mutations coding for drug resistance
coefficients ranging from 0.77 to 0.99 (P<0.07). Un- were detected in the cell-associated viral DNA from
like the large reduction in unspliced HIV-1 mRNA the other patients, indicating that treatment sup-
that occurred after the initiation of treatment, there pressed both discernible viral replication in plasma
were smaller decreases in the tat, rev, and nef mRNA and selection for drug-resistant variants.

1618 · May 2 7 , 19 9 9

The New England Journal of Medicine

Downloaded from nejm.org at TEXAS STATE UNIV on May 10, 2013. For personal use only. No other uses without permission.
Copyright © 1999 Massachusetts Medical Society. All rights reserved.
 PERSISTENCE OF HIV-1 TRANSCRIP TION IN PATIENTS RECEIVING POTENT ANTIRETROVIRAL THERAPY

We assessed the viral reservoir by measuring chang-

es in the concentration of cell-associated HIV-1
10 DNA and viral mRNA over time and found that the
Unspliced Viral mRNA:Multiply Spliced Viral mRNA

levels of virus decreased slowly in two phases, which

differed from the previously described pattern of de-
crease for plasma HIV-1 RNA.5 During the first
Patient 1F
8 Patient 2F phase, the concentrations of unintegrated viral DNA,
Patient 3F integrated proviral DNA, and unspliced viral mRNA
Patient 4F
Patient 5
decreased by approximately 1.5 log, consistent with
the hypothesized decreases in short-lived cells in-
fected with actively replicating virus and long-lived
6
infected cells.5 The half-life of integrated proviral
DNA during the first phase averaged 53 days and
ranged from 29 to 108 days, a range that was similar
to those for unintegrated circular forms of viral DNA
4 and unspliced viral mRNA. These estimates repre-
sent minimal values because the cell populations
would be renewed by ongoing viral replication dur-
ing the first phase. The values at the upper ends of
2 the ranges, however, are similar to the estimated life
span of resting memory CD4+ T cells in people with-
out HIV-1 infection.25 We did not identify the spe-
cific phenotypes of the cells that contained HIV-1
DNA and viral mRNA. Nonetheless, on the basis of
0
0 200 400 600 800 1000 previous work,13 our findings reflect the decrease in
two groups of HIV-1–infected peripheral-blood mon-
Treatment (days)
onuclear cells: short-lived cells infected with actively
Figure 2. Changes in the Ratio of Cell-Associated Unspliced replicating virus, which have a half-life of one to two
Viral mRNA to Multiply Spliced Viral mRNA in the Five Patients. days,13 followed by long-lived infected cells.
After the initial decrease during the first 500 days
of treatment, the levels of cell-associated HIV-1 DNA
and unspliced viral mRNA reached a plateau, char-
DISCUSSION acterizing the second phase. Persistently infected rest-
In patients infected with HIV-1, compartments of ing memory CD4+ T cells with integrated proviral
replication-competent virus persist despite treatment DNA and labile, unintegrated linear viral DNA rep-
with potent antiretroviral drugs that reduce plasma resent stable10 and inducible 23,24 viral reservoirs, re-
HIV-1 RNA to undetectable levels.9-11,16 To investi- spectively, that can contribute to the maintenance of
gate the characteristics of this viral reservoir, we stud- the plateau. Occasional stimulation by antigens may
ied the essential steps in the replication of HIV-1 provide an opportunity for persistently infected cells
(Fig. 3). The life cycle begins with the reverse tran- harboring replication-competent HIV-1 DNA to pro-
scription of genomic viral RNA into linear uninte- duce progeny virus transiently and then either die as
grated HIV-1 DNA.22 Once the viral DNA has been a direct or indirect result of viral replication or be-
synthesized and incorporated into a preintegration come quiescent again, thus replenishing the pool of
complex, it enters the nucleus.22-24 This is followed HIV-1–infected CD4+ T cells. Infectious virus pro-
by the integration of the linear forms of viral DNA duced in cells or areas of the body where medica-
into the host-cell genome3,23,24 and expression of the tions do not reach may also have a role in sustaining
virus.15 There are also circular forms of viral DNA in this pool.26,27 Both possibilities are consistent with
the nucleus that are byproducts of the integration the absence in our patients of mutations coding for
process. We used quantitative PCR assays to measure drug resistance other than those that arose during
cell-associated unintegrated circular forms of viral previous monotherapy that failed to suppress viral
DNA, as a surrogate marker of nuclear entry, inte- replication completely.
grated proviral DNA, and viral mRNA levels. All five In transformed cell lines harboring integrated provi-
patients whom we studied were receiving potent an- ral DNA that have been used as a model of persist-
tiretroviral therapy and had had undetectable levels ent HIV-1 infection, there are low levels of incom-
of HIV-1 RNA in plasma for 20 months or more. plete viral replication that result from a block early in
They also all had evidence of viral replication in their the life cycle of the virus.28-32 Such persistently infect-
peripheral-blood mononuclear cells, suggesting the ed CD4+ T cells, which contain integrated proviral
persistent presence of reservoirs of HIV-1. DNA that is transcriptionally active but translation-

Vol ume 340 Numb e r 21 · 1619

The New England Journal of Medicine

Downloaded from nejm.org at TEXAS STATE UNIV on May 10, 2013. For personal use only. No other uses without permission.
Copyright © 1999 Massachusetts Medical Society. All rights reserved.
 The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne

Mature
Retroviral retroviral
particle particle

Attachment Budding
CD4

Cell
membrane
Chemokine
receptor

Penetration Assembly

Structural
Cytoplasm proteins

Uncoating

Reverse Translation
transcription

Two-LTR
circular viral DNA Regulatory
proteins
Linear unintegrated
viral DNA Unspliced
mRNA

Preintegration One-LTR
complex circular viral Multiply
DNA
spliced
mRNAs
Integrated Transcription
proviral DNA Nucleus

Figure 3. The Essential Steps in the Life Cycle of HIV-1.

The first step is the attachment of the virus particle to receptors on the cell surface. The HIV-1 RNA genome then enters the cyto-
plasm as part of a nucleoprotein complex. The viral RNA genome is reverse-transcribed into a collinear DNA duplex, which has
terminal duplications known as long terminal repeats (LTRs) that are not present in HIV-1 RNA. Once the viral DNA has been syn-
thesized, the linear viral DNA molecule is incorporated into a preintegration complex that enters the nucleus. In the nucleus, unin-
tegrated viral DNA is found in both linear and circular forms. The unintegrated circular forms of viral DNA have either one or two
long terminal repeats, are byproducts of the integration process, and are found exclusively in the nucleus. The linear unintegrated
viral DNA is the precursor of integrated proviral DNA, which is a stable structure that remains indefinitely in the host-cell genome
and serves as a template for viral transcription. Transcription of the proviral DNA template and alternative mRNA splicing creates
spliced viral mRNA species encoding the viral accessory proteins, including Tat, Rev, and Nef, and the unspliced viral mRNA en-
coding the viral structural proteins, including the gag–pol precursor protein. A shift in the transcriptional pattern from the expres-
sion of predominantly multiply spliced viral mRNA to predominantly unspliced viral mRNA is indicative of active viral replication.
All the viral transcripts are exported into the cytoplasm, where translation and assembly and processing of the retroviral particle
take place. The cycle is completed by the release of infectious retroviral particles from the cell.

1620 · May 2 7 , 19 9 9

The New England Journal of Medicine

Downloaded from nejm.org at TEXAS STATE UNIV on May 10, 2013. For personal use only. No other uses without permission.
Copyright © 1999 Massachusetts Medical Society. All rights reserved.
 PERSISTENCE OF HIV-1 TRANSCRIP TION IN PATIENTS RECEIVING POTENT ANTIRETROVIRAL THERAPY

ally nonproductive, may also contribute to the low, We are indebted to John Coffin and Paulina Essunger for critical
stable ratio of unspliced HIV-1 mRNA to multiply review, to Jamie Drew and Joan Chmiel for data management, and
to Sam Wu and Hiaying Li for technical assistance.
spliced HIV-1 mRNA and the stable levels of cell-
associated integrated proviral DNA we found dur- REFERENCES
ing the second phase of the viral decrease. Although
1. Ho DD, Neumann AU, Perelson AS, Chen W, Leonard JM, Markowitz
our findings cannot be used to establish a half-life M. Rapid turnover of plasma virions and CD4 lymphocytes in HIV-1 in-
for the unintegrated forms of viral DNA in vivo, they fection. Nature 1995;373:123-6.
are consistent with two hypotheses: that the unin- 2. Wei X, Ghosh SK, Taylor ME, et al. Viral dynamics in human immu-
nodeficiency virus type 1 infection. Nature 1995;373:117-22.
tegrated forms of viral DNA have a rather short half- 3. Coffin JM. HIV population dynamics in vivo: implications for genetic
life and are regenerated by constant reinfection, or variation, pathogenesis, and therapy. Science 1995;267:483-9.
that they have a long half-life and persist as long 4. Haase AT, Henry K, Zupancic M, et al. Quantitative image analysis of
HIV-1 infection in lymphoid tissue. Science 1996;274:985-9.
as does integrated proviral DNA in long-lived in- 5. Perelson AS, Essunger P, Cao Y, et al. Decay characteristics of HIV-1-
fected cells. infected compartments during combination therapy. Nature 1997;387:
188-91.
Mathematical modeling of HIV dynamics has sug- 6. Cavert W, Notermans SW, Staskus K, et al. Kinetics of response in
gested that at least two to three years of a completely lymphoid tissues to antiretroviral therapy of HIV-1 infection. Science
inhibitory antiretroviral regimen would be required 1997;276:960-4. [Erratum, Science 1997;276:1321.]
7. Autran B, Carcelain G, Li TS, et al. Positive effects of combined anti-
to eliminate virus from long-lived infected CD4+ retroviral therapy on CD4+ T cell homeostasis and function in advanced
T cells.5,21 This estimate was not entirely accurate HIV disease. Science 1997;277:112-6.
because it was based on data that did not adequately 8. Palella FJ Jr, Delaney KM, Moorman AC, et al. Declining morbidity
and mortality among patients with advanced human immunodeficiency
reveal the persistence of a population of cells carry- virus infection. N Engl J Med 1998;338:853-60.
ing replication-competent integrated forms of provi- 9. Chun TW, Carruth L, Finzi D, et al. Quantification of latent tissue res-
ervoirs and total body viral load in HIV-1 infection. Nature 1997;387:183-
ral DNA.10,11,16 This model also did not address the 8.
effects of less-than-complete suppression of viral rep- 10. Finzi D, Hermankova M, Pierson T, et al. Identification of a reservoir
lication. Both these factors are important in attempts for HIV-1 in patients on highly active antiretroviral therapy. Science 1997;
278:1295-300.
to determine the duration of therapy required to erad- 11. Wong JK, Hezareh M, Gunthard HF, et al. Recovery of replication-
icate the virus.33 competent HIV despite prolonged suppression of plasma viremia. Science
Our data suggest that even after two or more 1997;278:1291-5.
12. Zhang ZQ, Notermans DW, Sedgewick G, et al. Kinetics of CD4+
years of complete suppression of plasma levels of T cell repopulation of lymphoid tissues after treatment of HIV-1 infection.
HIV-1 RNA, viral transcription continues and the Proc Natl Acad Sci U S A 1998;95:1154-9.
13. Perelson AS, Neumann AU, Markowitz M, Leonard JM, Ho DD.
decrease in the levels of integrated forms of proviral HIV-1 dynamics in vivo: virion clearance rate, infected cell life-span, and
DNA plateaus. Thus, the long-lived populations of viral generation time. Science 1996;271:1582-6.
persistently infected cells may not be decreasing at a 14. Kaslow RA, Ostrow DG, Detels R, Phair JP, Polk BF, Rinaldo CR Jr.
The Multicenter AIDS Cohort Study: rationale, organization and selected
simple exponential rate; this finding suggests that characteristics of the participants. Am J Epidemiol 1987;126:310-8.
there are not yet sufficient data to allow estimates of 15. Furtado MR, Kingsley LA, Wolinsky SM. Changes in the viral mRNA
the duration of therapy required to eradicate the in- expression pattern correlate with a rapid rate of CD4+ T-cell number de-
cline in human immunodeficiency virus type 1-infected individuals. J Virol
fection. Interestingly, our finding of a quasi–steady 1995;69:2092-100.
state in the levels of the virus indicates that the rate of 16. Chun TW, Stuyver L, Mizell SB, et al. Presence of an inducible HIV-1
latent reservoir during highly active antiretroviral therapy. Proc Natl Acad
replenishment of these cells is approximately equal Sci U S A 1997;94:13193-7.
to the natural depletion rate or that the rate is so slow 17. Teo I, Veryard C, Barnes H, et al. Circular forms of unintegrated hu-
that it is unnoticeable. Unless this quasi–steady state man immunodeficiency virus type 1 DNA and high levels of viral protein
expression: association with dementia and multinucleated giant cells in the
eventually disappears with longer periods of therapy brains of patients with AIDS. J Virol 1997;71:2928-33.
or can be overcome by the use of more potent ther- 18. Wolinsky SM, Korber BTM, Neumann AU, et al. Adaptive evolution
apies or alternative approaches that block the poten- of human immunodeficiency virus type 1 during the natural course of
infection. Science 1996;272:537-42.
tial spread of virus within tissues, HIV-1 may never 19. Hammond JA, Larder B, Schinazi RF, Mellors JW. Mutations in retro-
be eradicated. viral genes associated with drug resistance. In: Korber B, Hahn B, Foley B,
et al., eds. Human retroviruses and AIDS 1997: a compilation and analysis
In summary, persistently infected peripheral-blood of nucleic acid and amino acid sequences. Los Alamos, N.M.: Theoretical
mononuclear cells with integrated proviral DNA can Biology and Biophysics Group, Los Alamos National Laboratory, 1997.
be found in some patients who are receiving combi- 20. Sokal RR, Rohlf FJ. Biometry, the principles and practice of statistics
in biological research. 2nd ed. San Francisco: W.H. Freeman, 1981.
nations of drugs that inhibit viral reverse transcriptase 21. Notermans DW, Goudsmit J, Danner SA, de Wolf F, Perelson AS, Mit-
and protease and in whom plasma levels of HIV-1 tler J. Rate of HIV-1 decline following antiretroviral therapy is related to
RNA have been undetectable for 20 months or more. viral load at base line and drug regimen. AIDS 1998;12:1483-90.
22. Coffin JM, Hughes SH, Varmus HE. Retroviruses. Cold Spring Har-
Our results suggest that this reservoir represents a se- bor, N.Y.: Cold Spring Harbor Laboratory Press, 1997.
rious impediment to the long-term goal of the erad- 23. Zack JA, Arrigo SJ, Weitsman SR, Go AS, Haislip A, Chen IS. HIV-1
entry into quiescent lymphocytes: molecular analysis reveals a labile, latent
ication of HIV-1. viral structure. Cell 1990;61:213-22.
24. Bukrinsky MI, Stanwick TL, Dempsey MP, Stevenson M. Quiescent
T lymphocytes as an inducible virus reservoir in HIV-1 infection. Science
Supported by grants from the National Institutes of Health (HD37356- 1991;254:423-7.
01, AI035039, and RR06555) and by a gift from an anonymous foun- 25. Mclean AR, Michie CA. In vivo estimates of division and death rates
dation. of human T lymphocytes. Proc Natl Acad Sci U S A 1995;92:3707-11.

Vol ume 340 Numb e r 21 · 1621

The New England Journal of Medicine

Downloaded from nejm.org at TEXAS STATE UNIV on May 10, 2013. For personal use only. No other uses without permission.
Copyright © 1999 Massachusetts Medical Society. All rights reserved.
 The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne

26. Kepler TB, Perelson AS. Drug concentration heterogeneity facilitates 30. Michael NL, Morrow P, Mosca J, Vahey M, Burke DS, Redfield RR.
the evolution of drug resistance. Proc Natl Acad Sci U S A 1998;95:11514- Induction of human immunodeficiency virus type 1 expression in chroni-
9. cally infected cells is associated primarily with a shift in RNA splicing pat-
27. Zhang H, Dornadula G, Beumont M, et al. Human immunodeficiency terns. J Virol 1991;65:1291-303. [Erratum, J Virol 1991;65:7084.]
virus type 1 in the semen of men receiving highly active antiretroviral ther- 31. Clouse KA, Powell D, Washington I, et al. Monokine regulation of hu-
apy. N Engl J Med 1998;339:1803-9. man immunodeficiency virus-1 expression in a chronically infected human
28. Pomerantz RJ, Trono D, Feinberg MB, Baltimore D. Cells nonproduc- T cell clone. J Immunol 1989;142:431-8.
tively infected with HIV-1 exhibit an aberrant pattern of viral RNA expres- 32. Butera ST, Roberts BD, Lam L, Hodge T, Folks TM. Human immu-
sion: a molecular model for latency. Cell 1990;61:1271-6. nodeficiency virus type 1 RNA expression by four chronically infected cell
29. Seshamma T, Bagasra O, Trono D, Baltimore D, Pomerantz RJ. lines indicates multiple mechanisms of latency. J Virol 1994;68:2726-30.
Blocked early-stage latency in the peripheral blood cells of certain individ- 33. Wein LM, D’Amato RM, Perelson AS. Mathematical analysis of anti-
uals infected with human immunodeficiency virus type 1. Proc Natl Acad retroviral therapy aimed at HIV-1 eradication or maintenance of low viral
Sci U S A 1992;89:10663-7. loads. J Theor Biol 1998;192:81-98.

1622 · May 2 7 , 19 9 9

The New England Journal of Medicine

Downloaded from nejm.org at TEXAS STATE UNIV on May 10, 2013. For personal use only. No other uses without permission.
Copyright © 1999 Massachusetts Medical Society. All rights reserved.