`Myriad Genetics, Inc. et al. (Petitioners) v. The Johns Hopkins University (Patent Owner)
`IPR For USPN 7,824,889
`
`Page 1 of 6
`
`
`
`ARTICLES
`
`FIG. 1 Plasma viral RNA determina-
`tions
`in
`representative
`subjects
`treated with the HIV-1 protease inhibi-
`tors ABT-538 (a) and L-735,524 (b).
`Subjects had not received other anti-
`retroviral agents for at least 4 weeks
`before therapy. Treatment was initi-
`ated
`at week
`O-
`with
`400-
`1,200 mg d™* of ABT-538 or 1,600-
`2,400 mg d* of L-735,524 and was
`continued throughout the study. Viral
`RNA was determined by modified
`branched DNA (bDNA)'® (a) or RT-
`PCR?’ (b) assay and confirmed by QC-
`PCR®. Shown are the least-squaresfit
`data,
`linear-regression
`curves
`for
`points between days O and 14indicat-
`viral
`ing
`exponential
`(first-order)
`elimination.
`
`104 -|
`
`i
`$=
`|
`ze
`Be 1084
`>= 5 107 =~
`£ 8
`b
`\
`a6
`.
`PE 101 ++
`7
`|
`105
`
`104 q
`j
`103
`|
`
`\
`
`‘
`
`\
`
`»
`
`\
`
`
`
`
`
`
`
`
`
`
`106 +
`—- -
`— a
`—.
`_
`—
`pe
`a \
`Subject 5001
`Subject 5002
`Subject 5003]
`.
`Subject 5004
`ee
`\ee
`ot
`“
`
`-
`
`Subject 6001
`
`\
`°
`
`~
`Subject 6002
`
`|
`|
`
`‘e
`
`|
`}
`|
`
`e
`
`°
`
`°
`ep
`a
`Subject 6003
`
`|
`;
`
`@
`
`“\e
`
`.
`
`Subject 6004;
`
`e
`
`5
`
`e
`‘.
`\e.
`poe ee ee en ae el
`10123465 61012345 6 -1012345 6-+01234656
`
`Time (weeks)
`
`plasmaandeither baseline CD4* lymphocyte count or baseline
`that takes into account the two separate processes ofelimination
`viral RNA level.
`offree virus and virus-producingcells, as described. Method (1)
`.
`gives a t1,. of 1.8+0.9 days; method (2) gives a t,/2 of 3.041.7
`Virus turnover
`days; and method (3) gives a t1 2 of 2.0+0.9 days for the slower
`Direct population sequencing. As an independent approach
`of the two decay processes and a very similar value, 1.5+0.5
`days, for the faster one. These are averages (+1s.d.) for all|for determining virus turnover and clearanceof infected cells,
`22 patients. Method (3) arguably provides the most complete|we quantified serial changes in viral genotype and phenotype
`assessment of the data, whereas method (2) provides a simpler
`with respect to drug resistance in the plasma and PBMCsof
`interpretation (but slightly slower estimate) for virus decline
`four subjects treated with NVP (Fig. 2). NVP potently inhibits
`becauseit fails to distinguish the initial delay in onset of antiviral
` HIV-1 replication but selects for one or more codon substitu-
`activity due to the drug accumulation phase, and the time
`tionsin the reverse transcriptase (RT) gene***!”’. These muta-
`required for very recently infected cells to initiate virus expres-
`tions result in dramatic decreases (up to 1,000-fold) in drug
`sion, from the subsequent phase of exponential virus decline.
`susceptibility and are associated with a corresponding loss of
`There were nosignificant differences in the viral clearance rates
`viral suppression in vivo**. Genetic. changes resulting in NVP
`in subjects treated with ABT-538, L-735,524 or NVP, and there
`resistance can thus serve as a quantifiable molecular marker of
`wasalso no correlation between the rate of virus clearance from
`virus turnover. A rapid decline in plasma viral RNA was
`
`FIG. 2 Plasma viral RNA deter-
`minations(a), CD4* lymphocyte
`counts (b), and percentages of
`
`‘Subject 1619
`Subject 1624
`Subject 1625
`mutant
`viral
`genomes
`in
`106
`408
`08 -
`108
`
`'
`|
`plasma and PBMCs(c) of sub-
`jects initiating treatment with
`1 4\ [ne
`105
`NVP. Subjects were partici-
`\
`av
`pants
`in a clinical protocol
`i
`assessing the effects of NVP
`104
`104 |
`when added to existing treat-
`:
`:
`ment with ddl (subject 1605) or
`
`103
`eT
`:
`
`ddI plus zidovudine (subjects
`1619, 1624, 1625). Treatment
`-
`100-
`——
`| BBO
`b
`7
`with NVP wasinitiated at week
`iN
`8
`go-
`| 200-4 A
`|
`‘
`j/
`O using 200 mg per day and
`120-1/
`3S300-//
`co NA
`150] d \ey
`was increased to 400 mg per
`T
`— oS
`i \
`|
`|
`i
`day after 2 weeks. ddl and
`, 80
`ey.
`53 200-4
`s 40-) |
`too2
`NO
`zidovudine
`dosages
`were
`oe
`zl
`wie |
`400 mg per day and 300-
`i
`|
`
`
`QO ~*|50100-4 * 1 40- | 20
`
`
`
`600 mg per day, respectively.
`
`O-
`-
`-
`0
`—
`ol
`0
`-
`Viral RNA (@) was determined
`
`by QC-PCR assay®. CD4* lym-
`eT
`7
`i
`|
`\ OT —
`phocytes (@) were quantified
`*—.—___. |
`100-4
`_- = —as
`| 100 2. 100 as 100 +
`/
`'
`f
`|
`i ©
`|
`by flow cytometry. Frequencies
`|
`/
`5 7 wf
`|
`7.
`|
`| 75 f
`Pt] wel
`of viral genomes containing
`<
`|
`G
`!
`/
`UB
`i
`=
`o |
`i
`f
`“Oo
`NVP-resistance-associatedmut-
`Bg st | 50 |
`| 50,
`50} {po —0 |
`
`ations
`in plasma
`(M)
`and
`
`27
`ly F
`'
`it
`|
`|
`|
`/
`!
`
`
`§
`ie
`{i
`|
`if
`PBMCs(LL) were determined by
`S
`84 /
`og
`254
`| 2)
`ae);
`automated
`DNA
`sequence
`=
`i)
`/
`;
`| /
`/
`|
`ae
`ae
`— “
`0d
`analysis (Fig. 3,
`legend), with
`0
`4
`0
`0
`4
`0
`4
`42
`each data point representing
`Time (weeks)
`the average of 3-6 indepen-
`dent PCR amplifications and
`sequence determinations.
`Page 2 of 6
`
`f
`
`/
`
`J
`
`a
`
`#8
`
`12
`
`oe
`32
`
`4
`
`#8
`
`12
`
`-
`
`«8B
`
`12 2
`
`+B
`
`12 2
`
`118
`
`NATURE - VOL 373 - 12 JANUARY 1995
`
`a
`
`|
`|
`eo
`|
`|
`|
`|
`
`a
`$=
`ae
`Se
`Sg
`36
`bs
`z=
`
`Cc
`
`:
`
`f
`
`\.
`
`—
`
`
`
`
`
`Page 2 of 6
`
`
`
`
`
`ARTICLES
`
`
`observed following the institution of NVP therapy and this was
`associated with a reciprocal increase in CD4” lymphocyte counts
`(Fig. 2a and 5). Both responses were of limited duration, return-
`ing to baseline within 6-20 weeks in these four patients. The
`proportion of virus in uncultured plasma and PBMCsthatcon-
`tain NVP-resistance-conferring mutations (Fig. 2c) was deter-
`mined by direct automated nucleotide sequencing ofviral nucleic
`acid (Fig. 3), as previously described**. We first validated this
`method by reconstitution experiments, confirming its sensitivity
`for detecting RT mutants that comprise as little as 10% of the
`overall virus population. Defined mixtures of wild-type and mut-
`ant HIV-1 RT cDNAclones (differing only at the second base
`position of codon 190) were amplified and sequenced (Fig. 3a).
`Varying proportions of wild-type and mutant viral sequences
`present in the original DNA mixtures (mutant composition: 0,
`10, 25, 50, 75 and 100%) were faithfully represented in the rela-
`tive peak-on-peak heights (and in the relative peak-on-peak
`areas) of cytosine (C) and guanine (G) residues at the second
`base position within
`this
`codon. Ratios of
`(mutant)/
`(mutant + wild type) nucleotide peak heights expressed in arbi-
`
`trary fluorescence units were as follows (predicted/observed):
`0/< 10%; 10/18%; 25/29%; 50/49%; 75/71% and 100/94%.
`Wenext determined the ability of direct population sequenc-
`ing to quantify wild-type and mutant viral RNA genomesin
`clinical specimens. Figure 3b shows the sequence chromatograms
`of RT codons 179-191 from virionspelleted directly from uncul-
`tured plasma specimens of subject 1625 before (day —7) and
`after (days +28 and +140) the initiation of NVP therapy. At
`day —7, all codons within the amino-terminal half of the RT
`gene (codons 1-250), including those shown, were wild-type at
`positions associated with NVP resistance’’*?. However, after
`only 28 days of NVPtherapy, the wild-type plasmavirus popula-
`tion was completely replaced by a NVP-resistant mutant popula-
`tion differing from the wild-type at codon 190 (glycine-to-serine
`substitution). After 140 days of drug therapy, this codon had
`evolved further such that the plasma virus population consisted
`of an equal mixture of two drug-resistant strains, one containing
`G190S and the other containing G190A. There were no other
`NVP-resistance-conferring mutations detectable within the viral
`RT gene.
`
`
`
`
`
`100% (wt)
`0% (mut)
`
`90%
`10%
`
`75%
`25%
`
`50%
`50%
`
`25%
`75%
`
`0%
`100%
`
`FIG. 3 Quantitative detection of HIV-1 drug-resistance mutations by
`automated DNA sequencing. a, DNA sequence chromatograms of RT
`codon 190 from a defined mixture of wild-type (wt) and mutant (mut)
`HIV-1 cDNA clones differing only at the second base position of the
`codon. Sequences shown were obtained from, and therefore are
`presented as, the minus (non-coding) DNA strand. For example, the
`minus-strand TCC sequence shown corresponds to the plus-strand
`codon GGA (glycine, G). Similarly, the minus-strand TGC sequence cor-
`responds to the plus-strand codon GCA (alanine, A). The single-letter
`amino-acid code correspondsto the plus-strand DNA sequence. Mixed
`bases approximating a 50/50 ratio are denoted as N. b, DNA sequence
`chromatograms of RT codons 179-191 (again displayed as the minus-
`strand sequence) derived from plasma-virion-associated RNAof subject
`1625 before (day —7) and after (days +28 and +140) starting NVP
`therapy. Codon changesresulting in amino-acid substitutions at position
`190 are indicated for the plus strand. For example, the GCC minus-
`strand sequence at position 190 (day —7) corresponds to GGC (glycine,
`G), and the GCT minus-strand sequence at position 190 (day +28)
`corresponds to AGC (serine, S) in the respective plus strands.
`METHODS.Mixtures of wild-type and mutant cDNAclones(a) were pre-
`pared and diluted such thatfirst-round PCR amplifications were done
`with 1,000 viral cDNA target molecules per reaction. HIV-1 RNA was
`isolated from virions pelleted from uncultured plasma specimens (b),
`as described*®. cDNA was prepared using Moloney murine leukaemia
`virus reverse transcriptase (GIBCO BRL)® and an oligonucleotide primer
`corresponding to nucleotides 4,283 to 4,302 of the HXB2 sequence**
`Thefull-length viral reverse transcriptase gene (1,680 bp) was amplified
`by means of a nested PCR using conditions and oligonucleotide primers
`(outer primers: nt 2,483-2,502 and 4,283-4,302; inner primers: nt
`2,549-2,565 and 4,211-4,229), previously reported*®. Subgenomic
`fragments of the RT gene were also amplified using combinations of
`the following oligonucleotide primers:
`(5’) 2,585-2,610; (5) 2,712-
`2,733;
`(3') 2,822-2,844;
`(3’) 3,005-3,028;
`(3’) 3,206-3,228;
`(3’)
`3,299-3,324;
`(3’) 3,331-3,350;
`(3’) 3,552-3,572; and (3’) 3,904—
`3,921. All 3’ primers incorporated the universal primer sequence for
`subsequent dye-primer sequence analysis. The HIV-1 copy numberin
`every PCR reaction was determined (100-10,000 copies). A total of
`three to six separate PCR amplifications of primary patient material was
`done on each sample using different combinations of primers, and
`representative chromatograms are shown. Rarely, codon interpretation
`was ambiguous.
`in the day +140 plasma sample from subject 1625
`(bottom of panel b),
`the complementary (plus) strand could read:
`AGC(serine), GCN(alanine), ACN(threonine), AGA/AGG(arginine)
`or
`GGN(glycine).
`In this case, we sequenced 7 full-length RT molecular
`clones and found that they encoded only serine or alanine. For sequ-
`encing, an automated ABI 373A sequenator and the Taq Dye Primer
`Cycle Sequencing Kit (ABI) were used. Sequences were analysed using
`Sequencher (Gene Codes Corp.) and Microgenie (Beckman) software
`packages, and base-pair mixtures were quantified by measuring relative
`peak-on-peak heights”®.
`
`Day -7
`
`1906.
`AGAGCCTACATATAAA CATCCATGTIA
`
`TGATA
`
`TA AC
`
`NATURE -AAGE 3942. 12 JanuaRY 1995
`
`119
`
`Page 3 of 6
`
`
`
`ARTICLES
`
`
`°
`o
`Gcc
`Subject 1625 - Plasma (Codon me
`Bo
`3
`
`d 28ae——
`=3
`<r
`zi*:
`IS
`
`Ss
`Subject 1625 - pone (Codon tee
`
`a N
`d14
`Subject 1624 - Plasma (Codon 190)
`Tec
`TGec
`
`;
`
`REC
`Subject 1624 - PEMe (Codon pee)
`©
`©)
`
`B:
`
`
`
`
`four subjects evaluated by direct viral population
`In all
`sequencing (Fig. 4), specific NVP-resistance-conferring muta-
`tions within the RT gene could be unambiguously identified and
`subsequently confirmed by molecular cloning, expression and
`drug susceptibility testing. In all cases, mutant virus increased
`rapidly in the plasmaandvirtually replaced wild-type virus after
`only 2-4 weeks of NVP therapy (Fig. 2c). By analysing the rate
`of accumulation of resistant mutants in the plasma population,
`we could obtain an independentestimate of the turnoverrate of
`free virus. The rise of drug-resistant mutant virus is influenced
`substantially by the precéding increase in the CD4" cell popula-
`tion (which provides additional resourcesfor virus production*’)
`and therefore follows complex dynamics. However, we could
`obtain an estimate of these dynamics by making simplifying
`assumptions. We assumethat wild-type virus declines exponen-
`tially with a decay rate a, and that the drug-resistant mutant
`increases exponentially with the rate 8. Thus, the ratio of mutant
`to wild-type virus increases exponentially at the combined rate
`a+. Our genetic RNA (cDNA)data allow us to estimate this
`sum. Knowing @ from our data on virus decline, we get B ~0.27,
`or a 32% daily virus production (average over 4 patients).
`Assuming that mutantvirusrises exponentially, this corresponds
`to a doubling time of ~2 days, which is in excellent agreement
`
`with the measured elimination half-life of 2.040.9 days for
`plasmavirus (Figs 1 and 2a). Turnoverof viral DNA from wild-
`type to drug-resistant mutant in PBMCs was delayed and less
`complete compared to plasmavirus, reaching levels of only 50-
`80% of the total PBMC-associated viral DNA population by
`week 20 (Fig. 2c). Measurementofthe time required for resistant
`virus to spread in the PBMC population allowed us also to
`estimate the half-life of infected PBMCs. After complete turn-
`over of mutant virus in the plasma pool, we may assume that
`PBMCsinfected with wild-type virus decline exponentially at a
`rate d, whereascells infected by mutant virus are generated at a
`constant rate, but also decline exponentially at rate d. With these
`simplifying assumptions,
`the rate at which the frequency of
`resistant virus in the PBMC population increases provides an
`estimate for the parameter d and hence for the half-life of
`infected PBMCs. We obtaineda half-life of ~50-100 days. This
`meansthat the average half-life of infected PBMCsis very long
`and of the same order ofmagnitude as the half-life ofuninfected
`PBMCs**?°. Based on the long half-life of PBMCs,and thefact
`that these cells harbour predominantly wild-type virus at a time
`(days 14-28) when mostvirus in plasma is mutant, we conclude
`that most PBMCscontribute comparatively little to plasma virus
`load. Instead, other cell populations, most probably in the lym-
`phoreticular system'''?°, must be the major source of virus
`A a production.
`Direct sequence analysis of viral nucleic acid revealed not only
`“a
`=
`rapid initial turnover in viral populations but also continuing
`viral evolution with respect to drug resistance mutations. In sub-
`ject 1625 (Fig. 4, top panel), wild-type virus in plasma was com-
`pletely replaced after 28 days of NVP therapy by mutantvirus
`
`
`Subject 1624 - Plasma (Codon 181)
`ATA
`ATA
`ATA
`™
`™”
`)
`
`|
`
`Seeeeee
`~
`™
`™
`{
`
`4.28
`14
`4-7
`Subject 1619 - Plasma (Codon190)
`Yee
`
`TAA
`Subject 1619: Flesma (Codon 188)
`
`i (YAFAL)
`| [|
`Hy
`| “wm,
`
`|
`
`Subject 1619- Plasma (Codon 181)
`
`™
`
`
`
`er
`m
`
`42
`
`Ne
`fee
`“ “
`
`|
`
`AK
`
`(V/F/L)
`\
`
`ATA
`
`(FL)
`ss
`
`>
`
`z °
`
`T
`
`™
`
`
`
`ae
`()
`
`acid
`@
`
`ATA
`
`aT a
`
`i™
`| |
`
`|
`
`:
`
`||
`|
`
`T
`
`n™
`
`
`>
`
`\
`
`120
`
`Page 4 of 6
`
`NATURE - VOL 373 - 12 JANUARY 1995
`
`d 84
`
`d 140
`
`d 288
`
`
`
`ot
`ton
`on day 0. As in Fig. 3, minus-strand sequences are shown together
`automated DNA sequencing in plasma viral RNA (cDNA) and PBMC-
`
`FIG. 4 Quantitative detection of HIV-1 drug resistance mutations by
`
`associated viral DNA populations before and after the initiation of NVP
`
`with single-letter amino-acid codes of the corresponding plus-strand
`sequence. Mixed bases approximating a 50/50 ratio are denoted
`as
`N.
`METHODS.HIV-1 cDNA wasprepared from virions pelleted from uncul-
`tured plasmaas described for Fig. 3. Viral DNA wasisolated from uncul-
`tured PBMCs,as described“. The full-length viral reverse transcriptase
`genes as well as subgenomic fragments were amplified and sequenced
`as described for Fig. 3. The HIV-1 copy number in every PCR reaction
`was determined (L00-10,000 copies). Some sequences were deter-
`mined from both coding and non-coding DNA strands to ensure the
`accuracy of quantitative measurements.
`
`Page 4 of 6
`
`
`
`
`
`ARTICLES
`
`
`
`
`
`
`
`1625
`
`PBMC
`
`1624
`
`Plasma
`
`1624
`
`PBMC
`
`1605
`
`1619
`
`PBMC
`
`Plasma
`
`cloned into pLG18-1, expressed in Escherichia coli, and tested
` TABLE 1 Jn situ functional analysis of HIV-1 RT clones
`
`individually for enzymatic function and NVP susceptibility by
`Functional NVP-sensitive
`NVP-resistant
`in situ assay”? (Table 1), For subject 1625 at day —7, 100%
`clones
`clones
`clones
`Specimen
`Subject
`(80/80) of RT clones from plasma and 100%(163/163) of RT
`0
`80
`80 (100%)
`(0%)
`Plasma
`day —7
`1625
`clones from PBMCsexpressed enzymethat wassensitive to NVP
`
`
`
`+14 27=(38%)72 45 (62%
`
`inhibition. By day 14, however, 62% of plasma-derived clones
`
`+28 57=(100%)57 0 (0%)
`
`
`
`expressed enzyme that wasresistant to NVP, and by days 28,
`+84
`67
`0
`(0%)
`67
`(100%
`+140
`86
`0
`(0%)
`86
`(100%
`84 and 140, 100% were resistant. Conversely, at day 14, 0% of
`—7
`163
`163 (100%)
`0
`{0%)
`PBMC-derived clones expressed NVP-resistant enzyme, and
`+14
`121
`121 (100%)
`0
`(0%)
`even after 28, 84 and 140 days, only 48-75% of clones were
`+28
`258
`134
`(52%)
`124
`(48%
`+84
`133
`43
`resistant. Similar results were obtained for the other study sub-
`(32%)
`90
`(68%
`+140
`261
`65
`(25%)
`196
`(75%
`jects (Table 1). Thus, the kinetics of virus population turnover
`-7
`19
`19 (100%)
`0
`(0%)
`determined by a quantitative RT in situ expression assay corre-
`+14
`34
`4
`(12%)
`30
`(88%)
`sponded closely with those determined by direct population
`+28
`79
`6
`(8%)
`73
`(92%
`sequencing (Fig. 2c).
`+140
`27
`0
`(0%)
`27
`(100%
`-7
`24
`24 (100%)
`0
`(0%)
`Infectious virus drug susceptibility testing. Plasma and
`+414
`34
`29
`(85%)
`5
`(15%
`PBMCsare knownto harboursubstantial proportions of defec-
`+28
`52
`42
`(81%)
`10
`(19%)
`tive or otherwise non-infectious virus**’. To determine whether
`+140
`87
`26
`(30%)
`61
`(70%)
`the viral genomes represented in total viral nucleic acid (Fig. 4
`-7
`31
`31 (100%)
`0
`(0%)
`+140
`31
`41
`(35%)
`20
`(65%
`and Table 1) corresponded to infectious virus with respect to
`~14
`79
`79 (100%)
`0
`(0%)
`NVP-resistance-conferring mutations, we co-cultivated PBMCs
`+28
`41
`0
`(0%)
`41
`(100%
`from three of the study subjects (1605, 1624, 1625) with normal
`+140
`38
`0
`(0%)
`38
`(100%)
`donor lymphoblasts in order to establish primary virusisolates.
`The RT genes of these cultured viruses, obtained before and
`after therapy, were cloned (Fig. 3 and Table {1
`legends) and
`sequenced in their entirety (V.A.J. and G.M.S., submitted). RT
`codons associated with NVP susceptibility were completely con-
`cordantin cultured and uncultured virus strains. Furthermore,
`the virus isolates exhibited NVP susceptibility profiles** consist-
`ent with their genotypes.
`
`Full-length RT genes were amplified by PCR from uncultured plasma and uncul-
`tured PBMCs as described in Fig. 3 legend. DNA products were cloned into the
`EcoRI! and Hindill sites of the bacterial expression plasmid pLG18-1 (refs 29, 30).
`The expression plasmids were screened for the presence of functional RT and
`tested in situ for susceptibility to NVP inhibition at 3,000 nM (~50-75 fold greater
`than the ICg0)**""5*. To ensure accuracy in distinguishing RT genes encoding
`NVP-resistant versus sensitive enzymes, and to confirm the identification of speci-
`fic NVP-resistance-conferring RT mutations obtained by direct sequencing (Figs 3
`and 4), we determined the complete nucleotide sequences of 21 cloned RT genes
`which had been phenotyped in the jn situ assay (V.A.J. and G.M.S., submitted).
`There was complete concordance between the phenotypes and genotypesof these
`21 clones with respect to NVP-resistance-conferring mutations, as well as com-
`plete concordance between direct viral population sequences and clone-derived
`sequences at NVP-resistance-conferring codons.
`
`(G190S), which in turn evolved by day 140 into a mixture of
`G190S and G190A. In subject 1624 (Fig. 4, middle panel), two
`codon changes conferring NVP resistance occurred. A G190A
`substitution appeared in plasma virus at day 14 and a Y181C
`appeared at day 42. Similarly, in subject 1605 (not shown), a
`Y181C mutation appeared in plasma at day 14 and a YI88L
`mutation at day 28. The sequential changes in plasma virus were
`mirrored by similar changes in PBMCsat later timepoints. In
`subject 1619, the pattern of resistance changes was even more
`complex (Fig. 4, bottom panel). By day 14, approximately 70%
`of plasma virus contained a G190A mutation. By day +28, this
`mutant population was largely replaced by virus containing a
`Y188F/L substitution. By day 84, still another majorshift in the
`viral quasispecies occurred, this time resulting in a population of
`viruses containing mutations at both YI8IC and GI90A.
`Finally, by day 288 the viral population in plasma consisted
`exclusively of a mutant exhibiting a single tyrosine-to-isoleucine
`substitution at position 181 (Y1811); mutations at codons 188
`and 190 were not.present in this virus population. All of these
`amino-acid substitutions at RT codons 181, 188 and 190 were
`shown in our in situ expression studies and by others*!**”* to
`confer high-level NVP resistance. The direct sequence analyses
`thus demonstrate that major changes in the HIV-1 quasispecies
`occur quickly and continuously in responseto selection pressures
`and that these changesare reflected first and most prominently
`in the plasma virus compartment.
`in situ RT gene expression and drug susceptibility testing.
`Because direct sequence analysis of viral mixtures provides only
`semiquantitative information and does not distinguish between
`viruses with functional rather than defective RT genes, we
`employed another method for quantifying virus turnover in
`uncultured plasma and PBMC compartments. Full-length RT
`genes were amplified by polymerase chain reaction (PCR),
`Rage 3 of 6
`NATURE -
`VOL
`373 - 12 JANUARY 1995
`
`CD4* lymphocyte dynamics
`Changes in CD4* lymphocyte counts during thefirst 28 days of
`therapy could be assessed in 17 of our patients (Fig. 2b and data
`not shown), CD4" cell numbers increased in every patient by
`between 41 and 830 cells per mm’. For the entire group, the
`average increase was 186+199 cells per mm® (mean +s.d.), or
`268 + 319% from baseline. As CD4* lymphocytes increase in
`numbers because of (1) exponential proliferation of CD4* cells
`in peripheral tissue compartments, and/or (2) constant (linear)
`production of CD4°cells from a pool of precursors, we analysed
`our data based on each of these assumptions. The average per-
`centage increase in cell number per day (assumption (1)) was
`5.0+3.1% (mean +1 s.d.). The average absolute increase in cell
`number per day (assumption (2)) was 8.0+7.8 cells mm7 d7'.
`Given that peripheral blood contains only 2% of the total body
`lymphocytes** and that the average total blood volume is ~5
`litres, an increase of 8 cells mm~* d~' implies an overall steady-
`state CD4* cell turnover rate (where increases equal losses) of
`(50) x (5 x 10° mm*) x (8 cells mm~* d7'), or 2x 10° CD4*cells
`produced and destroyed each day.
`
`Discussion
`Previously, it was shown that lymphoreticular tissues serve as
`the primary reservoir andsite of replication for HIV-1 (refs 11,
`19, 20) and that virtually all HIV-1-infected individuals, regard-
`less of clinical stage, exhibit persistent plasma viraemia in the
`range of 10° to 10” virions per ml®. However, the dynamic contri-
`butions of virus production and clearance, and of CD4"* cell
`infection and turnover, to the clinical ‘steady-state’ were obscure,
`although not unanticipated**”**?. We show by virus quantita-
`tion and mutation fixation rates that the composite lifespan of
`plasma virus and of virus-producing cells is remarkably short
`(4/2=2.0+0.9 days). This holds true for patients with CD4*
`lymphocyte counts as low as 18 cells per mm* and as high as
`355 cells per mm® (Figs 1 and 2; G.M.S., unpublished). These
`findings were made in patients treated with three different anti-
`retroviral agents having two entirely different mechanisms of
`action and using three different experimental approaches for
`assessing virus turnover. The viral kinetics thus cannot be
`121
`
`Page 5 of 6
`
`
`
`ARTICLES
`
`
`
`explained by a unique or unforeseen drug effect or a peculiarity
`of any particular virological assay method. Moreover, when new
`cycles of infection are interrupted by potent antiretroviral ther-
`apy, plasmaviruslevels fall abruptly by an average of 99%, and
`in some cases by as much as 99.99% (10,000-fold). This result
`indicates that
`the vast majority of circulating plasma virus
`derives from continuous rounds of de novo virus infection,
`replication and cell turnover, and not from cells that produce
`virus chronically or are latently infected and becomeactivated.
`The identity and location of this actively replicating cell popula-
`tion is not known, but appears notto reside in the PBMC pool,
`consistent with prior reports'!’!*°. Nevertheless, PBMCstraffic
`through secondary lymphoid organs and to some extent are in
`equilibrium with these cells*°. It is thus possible that a small
`fraction of PBMCs®**'*"'”, like a small fraction of activated lym-
`phoreticular cells*’, could make an important contribution to
`viraemia.
`The magnitude of ongoing virus infection and production
`required to sustain steady-state levels of viraemia is extraordin-
`ary: based on a virus t),2 of 2.0 days andfirst-order clearance
`kinetics (v(t) = 0(0)e"*, where @ =0.693/t,,2), 30% or more of
`the total virus population in plasma must be replenished daily.
`For a typical HIV-1-infected individual with a plasmavirustitre
`equalling the pretreatment geometric mean in this study (10°°
`RNA molecules per ml/2 RNA molecules per virion= 10°?
`virions per ml) and a plasma volumeof3 litres, this amounts
`to (0.30) x (10°?) x (3 x 10?) = 1.1 x 10° virions per day (range
`for all 22 subjects, 2 x 10’ to 7 x 10°), Even this may be a substan-
`tial underestimate of virus expression because virions may be
`inefficiently transported from the interstitial extravascular spaces
`into the plasma compartment andviral protein expression alone
`(short of mature particle formation) may result in cytopathy
`or immune-mediated destruction. Because the half-life of cells
`producing the majority of plasma virus cannot exceed 2.0 days,
`at least 30% of these cells must also be replaced daily. In our
`patients, we estimated the rate of CD4* lymphocyte turnoverto
`be, on average, 2 10” cells per day, or about 5% ofthe total
`CD4" lymphocyte population, depending on clinical stage. This
`rapid and ongoing recruitment of CD4*cells into a short-lived
`virus-expressing pool probably explains the abrupt increase in
`CD4* lymphocyte numbersthat is observed immediately follow-
`ing the initiation of potent antiretroviral therapy, and suggests
`the possibility of successful immunological reconstitution even
`
`in late-stage disease if effective control of viral replication can
`be sustained.
`The kinetics of virus and CD4* lymphocyte production and
`clearance reported here have a numberofbiological and clinical
`implications. First,
`they are indicative of a dynamic process
`involving continuous rounds of de novo virus infection, replica-
`tion and rapid cell turnover that probably represents a primary
`driving force underlying HIV-1 pathogenesis. Second,
`the
`demonstration of rapid and virtually complete replacement of
`wild-type virus by drug-resistant virus in plasma after only 14~
`28 days of drug therapy is a striking example of the capacity of
`the virus for biologically relevant change. In particular,
`this
`implies that HIV-1 must have enormous potential to evolve in
`response to selection pressures as exerted by the immune
`system*’. Although other studies*”*? have provided some evi-
`dence that virus turnover occurs sooner in plasma than in
`PBMCs,our data show this phenomenon most clearly. A similar
`experimental approach involving the genotypic and phenotypic
`analysis of plasma virus could be helpful in identifying viral
`mutations andselection pressures involved in resistance to other
`drugs, immune surveillance and viral pathogenicity. Third, the
`difference in lifespan between virus-producing cells and latently
`infected cells (PBMCs) suggests that virus expression per se is
`directly involved in CD4™ cell destruction. The data do not sug-
`gest an ‘innocent bystander’ mechanism ofcell killing whereby
`uninfected or latently infected cells are indirectly targeted for
`destruction by adsorption of viral proteins or by autoimmune
`reactivities.
`Although we have emphasized that most virus in plasma
`derives from an actively replicating short-lived population of
`cells, latently infected cells that becomeactivated or chronically
`producingcells that generate proportionately less virus (and thus
`do not contribute substantially to the plasma virus pool) may
`nonetheless be important in HIV-1 pathogenesis. Based on in
`situ analysis’°, these cells far outnumberthe actively replicating
`pool and the diversity of their constituent viral genomes repre-
`sents a potentially important source of clinically relevant vari-
`ants,
`including those conferring drug resistance.
`In future
`studies, it will be important not only to discern the specific elimi-
`nation rates of free virus and of the most actively producing
`cells, but also the dynamicsof virus replication and cell turnover
`in othercell populations andin patients at earlier stages of infec-
`tion. Such information will be essential to developing a better
`understanding of HIV-1 pathogenesis and a more rational
`approach to therapeutic intervention.
`im)
`
`Received 22 November; accepted 16 December 1994.
`1. Ho, D. D., Moudgil, T. & Alam, M. New Engi. J. Med. 321, 1621-1625 (1989).
`2. Coombs, R. W. et af. New Engl. J. Med. 321, 1626-1631 (1989).
`3. Saag, M. S. et al. J. infect. Dis 164, 72-80 (1991).
`4. Clark, S. J. et al. New Engl. J. Med. 324, 954-960 (1991).
`5. Daar, E. S., Mougdil, T., Meyer, R. D. & Ho, 0.D. New Engl. J. Med. 324, 961-964 (1991).
`6. Piatak, M. Jr et al. Science 259, 1749-1754 (1993).
`7. Piatak, M. et af. Lancet 341, 1099 (1993).
`8. Schnittman, S. M., Greenhouse, J. J., Lane, H. C., Pierce, P. F. & Fauci, A. S. AIDS Res.
`Hum. Retrovir. 7, 361-367 (1991).
`9. Michael, N. L., Vahey, M., Burke, D. S. & Redfield, R. R. J. Virol. 66, 310-316 (1992).
`10. Winters, M. A., Tan, L. B., Katzenstein, D. A. & Merigan, T. C. J. clin. Microbiol. 31, 2960-
`2966 (1993).
`11. Pantaleo, G. et af. Proc. natn. Acad. Sci. U.S.A. 88, 9838-9842 (1991).
`12. Connor, R.
`|., Mohri, H., Cao, Y. & Ho, D. D. J. Virol. 67, 1772-1777 (1993).
`13. Bagnarelli, P. et af. J. Virol. 66, 7328-7335 (1992).
`14. Bagnarelli, P. et al. J. Virol. 68, 2495-2502 (1994).
`15. Graziosi, C. et al. Proc. natn. Acad. Sci. U.S.A. 90, 6405-6409 (1993).
`16. Patterson, B. K. et al. Science 260, 976-979 (1993).
`17. Saksela, K., Stevens, C., Rubinstein, P. & Baltimore, D. Proc. natn. Acad. Sci. U.S.A. 91,
`1104-1108 (1994).
`18. Cao, Y. et al. AIDS Res. Hum. Retrovir. {in the press).
`19. Pantaleo, G. et a/. Nature 362, 355-358 (1993).
`20. Embretson, J. et ai. Nature 362, 359-362 (1993).
`21. Aoki-Sei, S. et al. AIDS Res. Hum. Retrovir. 8, 1263-1270 (1992).
`22. Coffin, J. M. Curr. Top. Microbiol. immun. 176, 143-164 (1992).
`23. Wain-Hobson, S. Curr. Opin. Genet. Dev. 3, 878-883 (1993).
`24. Merluzzi, V. J. et al. Science 250, 1411-1413 (1990).
`25. Kempf, D. et al. Proc. natn. Acad. Sci. U.S.A. {in the press).
`
`26, Vacca, J. P. et al. Proc. natn. Acad. Sci. U.S.A 91, 4096-4100 (1994).
`27. Mulder, J. et al. J. clin. Microbiol. 32, 292-300 (1994).
`28. Larder, B. A. et al, Nature 365, 671-675 (1993).
`29. Prasad, V. R. & Goff, S. P. J. biol. Chem. 264, 16689-16693 (1989).
`30. Saag, M. S. et al. New Engl. J. Med. 329, 1065-1072 (1993).
`31. Richman, D. D. et af. Proc. natn. Acad. Sci. U.S.A. 88, 11241-11245 (1991).
`32. Richman,D. D.et al. J. Virol. 68, 1660-1666 (1994).
`33. McLean, A. R. & Nowak, M. A. AIDS 6, 71-79 (1992).
`34. Michie, C., McLean, A., Alcock, C. & Beverley, P. C. L. Nature 360, 264-265 (1992).
`35. Sprent, J. & Tough, D. F. Science 265, 1395-1400 (1994).
`36. Balzarini, J. et al. Proc. natn. Acad. Sci. U.S.A. 91, 6599-6603 (1994).
`37. Meyerhans,A. et a/. Cel! 58, 901-910 (1989).
`38. Japour, A. J. et al. Antimicrob. Agents Chemother. 37, 1095-1101 (1993).
`39. Nowak, M. A. et al. Science 284, 963-969 (1991).
`40. Simmonds,P. et af. J. Virol. 65, 6266-6276 (1991).
`41. Smith, M. S., Koerber, K. L. & Pagano, J. S. J. infect. Dis. 167, 445-448 (1993).
`42. Znang, Y.-M., Dawson, S. C., Landsman, D., Lane, H. C.& Salzman, N. P. J. Virol. 68, 425—
`432 (1994).
`43. Myers, G. K., Korber, B., Berzofsky, J. A. & Smith, R.F. Human Retroviruses and AIDS 1993
`{Los Alamos Nationa! Laboratory, New Mexico, 1993).
`44. Shaw,G. M.et al. Science 226, 1165-1171 (1984).
`ACKNOWLEDGEMENTS. Wethankthe study participants; K. Squires, J. M. Kilby, M. Trechsel,
`L. DeLoach and the UAB 1917Clinic staff; Abbott Laboratories, Merck & Co. and Boehringer
`Ingelheim Pharmaceuticals Inc. (BIP!); J. Coffin, R. May and F. Gao for discussion; J. Decker, S.
`Campbell-Hill, Y. Niu and S. Yin Jiang for technica! assistance; and J. Wilson for artwork. This
`study was supported by the NIH, the US Army Medical Research Acquisition Activity, BIPI, the
`Wellcome Trust, Keble College and Boehringer