`
`
`Rapid turnover of plasma virions and CD4
`lymphocytesin HIV-1 infection
`David D. Ho, Avidan U. Neumann‘, Alan S. Perelson’, Wen Chen,
`John M. Leonard’ & Martin Markowitz
`Aaron Diamond AIDS Research Center, NYU School of Medicine, 455 First Avenue, New York, New York 10016, USA
`* Santa Fe Institute, Santa Fe, New Mexico 87501, USA
`+ Theoretical Division, Los Alamos National Laboratory, Los Alamos, New Mexico 87545, USA
`{ Pharmaceutical Products Division, Abbott Laboratories, Abbott Park,Illinois 60064, USA
`
`Treatment of infected patients with ABT-538, an inhibitor of the protease of human immuno-
`deficiency virus type 1 (HIV-1), causes plasma HIV-1 levels to decrease exponentially (mean
`half-life, 2.1 + 0.4 days) and CD4 lymphocyte counts to rise substantially. Minimum estimates
`of HIV-1 production and clearance and of CD4 lymphocyte turnover indicate that replication of
`HIV-1 in vivo is continuous and highly productive, driving the rapid turnover of CD4 lymphocytes.
`
`
`In HIV-1 pathogenesis, an increased viral load correlates with
`CD4 lymphocyte depletion and disease progression' °, but rela-
`tively little information is available on the kinetics of virus and
`CD4 lymphocyte turnoverin vivo. Here we administer an inhibi-
`tor of HIV-1 protease, ABT-538 (refs 10, 11), to twenty infected
`patientsin order to perturb the balance between virus production
`and clearance. From serial measurements of the subsequent
`changes in plasma viraemia and CD4 lymphocyte counts, we
`have been able to infer kinetic information about the pretreat-
`ment steady state.
`ABT-538 has potent antiviral activity in vitro and favourable
`pharmacokinetic and safety profiles in vivo'®. It was adminis-
`tered orally (600-—1,200 mg per day) on day 1 and daily thereafter
`to twenty HIV-1-infected patients, whose pretreatment CD4
`lymphocyte counts and plasmaviral levels ranged from 36 to
`490 per mm* and from 15x 10° to 554 16° virions per ml,
`respectively (Table 1). Post-treatment CD4 lymphocyte counts
`were monitored sequentially, as were copy numbersofparticle-
`associated HIV-! RNAin plasma, using an ultrasensitive assay
`(Fig.
`1 legend) based on a modification of the branched DNA
`signal-amplification technique'~'’. The trial design and clinical
`findings of this study will be reported elsewhere (M.M. efal.,
`manuscript in preparation).
`
`Kinetics of HIV-1 turnover
`Following ABT-538 treatment, every patient had a rapid and
`dramatic decline in plasma viraemia over the first two weeks.
`As shown using three examples in Fig.
`la, the initial decline
`in plasma viraemia was-always exponential, demonstrated by a
`straight-line fit to the data on a log plot. The slope ofthis line,
`as defined by linear regression, permitted the half-life (t,,.) of
`viral decay in plasma to be determined (Fig.
`|
`legend): for
`example, patient 409 was found to have a viral decay slope of
`—0.47 per day, yielding a t,,2 of 1.5 days (Fig. 1a). Hence the
`rate and extent of decay of plasma viraemia was determined for
`each patient. As summarized in Fig. 1b, in every case there was
`a rapid decline, the magnitude of which ranged from 11- to 275-
`fold, with a mean of 66-fold (equivalent to 98.5% inhibition).
`The residual viraemia may be attributable to inadequate drug
`concentration in certain tissues, drug resistance, persistence of a
`small long-lived virus-producing cell population (such as macro-
`phages), and gradual activation of a latently infected pool of
`cells. As summarized in Table 1, the viral decay slopes varied
`from —0.21 to —0.54 per day, with a mean of —0.34+0.06 per
`day; correspondingly, t),2 varied from 1.3 to 3.3. days, with a
`mean of 2.140.4 days. The latter value indicates that, on aver-
`NATURE - VOL 373 - 12 JANUARY 1995
`
`age, half of the plasma virions turn over every two days, showing
`that HIV-1 replication in vivo must be highly productive.
`The exponential decline in plasma viraemia following ABT-
`538 treatment reflects both the clearance of free virions and the
`loss of HIV-1-producing cells as the drug substantially blocks
`new rounds of infection. But although drug inhibition is prob-
`ably incomplete and virus-producing cells are not lost immedi-
`ately, a minimum valueforviral clearance canstill be determined
`(Fig.
`1
`legend) by multiplying the absolute value of the viral
`decay slope by the initial viral load. Assuming that ABT-538
`administration does not affect viral clearance, this estimate is
`also valid before treatment. As the viral load varies little during
`the pretreatmentphase (Fig. 1a, and data not shown), we assume
`there exists a steady state and hence the calculated clearance
`rate is equal to the minimum virion production rate before drug
`therapy. Factoring in the patient’s estimated plasma and extra-
`cellular fluid volumes based on body weight, we determined the
`minimum daily production and clearance rate of HIV-1 particles
`for each case (Table 1). These values ranged from 0.05 to
`
`2.07 x 10° virions per day with a mean of 0.68 + 0.13 x 10” virions
`per day. Although these viral turnover rates are already high,
`true values may be up to a few-fold higher, depending on the
`ty. of virus-producing lymphocytes. The precise kinetics of this
`additional parameter remains undefined. However, the mean ¢;/2
`of virus-producing cells is probably less, or in any case cannot
`be much larger, than the mean 1,2 of 2.1 days observed for
`plasma virion elimination, demonstrating that
`turnover of
`actively infected cells is both rapid and continuous. It could also
`be inferred from our data that nearly all (98.5%) of the plasma
`virus must come from recently infected cells.
`Examination of Fig. 1b shows that the viral decay slopes
`(clearance rate constants) are independent of the initial viral
`loads. The slopes do not correlate with the initial CD4 lympho-
`cyte counts (Fig. 2a), another indicator of the disease status of
`patients. Therefore these observations strongly suggest that the
`viral clearance rate constant is not dependent on the stage of
`HIV-1 infection. Instead, they indicate that viral load is largely
`a function of viral production, because clearance rate constants
`vary by about 2.5-fold whereas the initial loads vary by almost
`40-fold (Table 1).
`
`Kinetics of CD4 lymphocyte turnover
`After ABT-538 treatment, CD4 lymphocyte counts rose in each
`of the 18 patients that could be evaluated. As shownin three
`examples in Fig. 3, some increases were dramatic (patient 409,
`for example) whereas others (such as patient 303) were modest.
`Based on the available data, it was not possible to determine
`123
`
`Ambry Exhibit 1018
`
`Ambry Exhibit 1018
`
`
`
`ARTICLES
`
`
`
`
`lope:
`—0.
`
`ty: 2.2 days
`42
`‘
`
`
`
`Slope: 0.47 -
`fy: 1.5 days
`ri2ea:
`:
`
`409
`
`& 1,000
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`100
`
`10
`
`1
`
`
`
`
`
`
`
`— x E o
`
`oy
`
`A &Q
`
`a.
`Q0
`<
`~%
`
`Zm
`
`0.1
`-10 -5
`
`0
`
`5
`
`10 15 20 25 30
`
`-10 -5
`
`0
`
`5
`
`10 15 20 25 30
`
`-10 -5
`
`0
`
`5
`
`10 15 20 25 30
`
`Days
`
`
`
`load before and after ABT-538 treatment was
`FIG. 1 a, Plasma viral
`begun on day 1 for three representative cases. Plasma samples were
`tested with the branched DNAsignal-amplification assay as previously
`described*?**. Those samples with RNA levels below the detection sen-
`sitivity of 10,000 copies per ml were then tested using a modified
`assaydiffering from the original in two ways: hybridization of the bDNA
`amplification system is mediated by binding to overhangs on contiguous
`target probes; and the enzymatic amplification system has been
`enhanced by modification of wash buffers and the solution in which the
`alkaline phosphatase probeis diluted. The results of these changes
`are a diminution of background signals, an enhancementof alkaline
`phosphatase activity, and thus a greater detection sensitivity (500
`copies per ml). Linear regression was used to obtain the best-fitting
`straight line for 3-5 data points between day 1 andtheinflection point
`before the plateau of the new steady-state level. The slope, S, of each
`line represents the rate of exponential decrease; that is, the straight-
`line fit
`indicates that the viral
`load decreases according to Vit)=
`V(O) exp (—St). Given
`the
`exponential
`decay, VUts,2)=WO)//2=
`VO) exp (—St1,2), and hencetheviral half fife, t1,2=1n (2)/S. Before drug
`administration, the change in viral load with time can be expressed by
`1
`the differential equation, dV/dt= P—cV, wherePis the viral production
`rate, c is the viral clearance rate constant, and V is the number of
`plasma virions. During the pretreatment steady state, dV/dt=0, and
`hence P=cV. We havealso tested moreintricate models, in which the
`viral decay is governed by two or three exponential rates, namely the
`viral clearance rate, the decay rate of virus-producing cells, and the
`decay rate of latently infected cells. But there are insufficient data at
`this time to estimate the multiple parameters separately. Nonetheless,
`using the model in ref. 14, we find that if the death rate of latently
`infected cells is very small compared with the other two rates, then
`viral decay follows a simple exponential decline, with S=c, because
`the slow activation of a large numberof latently infected celis offsets
`the loss of actively infected cells. Irrespective of the model, on a log
`plot, S=—d(In V)/dt=c— P/V.If drug inhibition is complete and virus-
`
`0.1
`-10 5 0
`
`5
`
`10 15 20 25 30
`
`Days
`
`producing ceils are rapidly lost (so P=0), then S=c.If virat production
`continues, S is still <c, so the slope is a minimum estimate of the
`viral clearance efficiency. b, Decline of plasma viral load after ABT-538
`treatmentin all 20 patients. The slope for each case was obtained as
`already discussed, and the length of each line was determined by the
`initial viral load and the new steady-state level.
`
`10,000
`
`b
`
`S 1,000
`
`100
`
`10
`
`x =33
`
`6
`‘a°o
`Z
`
`with confidence whetherthe rise wasstrictly exponential (Fig. 3,
`top) or linear (Fig. 3, bottom). An exponential increase would
`be consistent with proliferation of CD4 lymphocytes in the per-
`iphery, particularly in secondary lymphoid organs, whereas a
`linear increase would indicate cellular production from a pre-
`cursor source such as the thymus'*. Given that the thymusinvo-
`lutes with age and becomes further depleted with HIV-1
`infection'®, it is more likely that the rise in CD4 lymphocytes is
`largely due to proliferation. Nevertheless, as both components
`may contribute, we analysed the obseved CD4 lymphocyte data
`by modelling both exponential and linear increases.
`124
`
`The slope of the line depicting the rise in CD4 lymphocyte
`counts on a log plot was determined for each case (Fig. 3, top).
`Individual slopes varied considerably, ranging from 0.004 to
`0.088 per day, with a mean of 0.047 per day (Table 1), corre-
`sponding to a mean doubling time of ~15 days (Fig. 3 legend).
`On average, the entire population of peripheral CD4 lympho-
`cytes was turning over every 15 days in our patients during the
`pretreatment steady state when CD4 lymphocyte production and
`destruction were balanced. Moreover, the slopes were inversely
`correlated with baseline CD4 lymphocyte counts (Fig. 26) in
`that patients with lower initial! CD4 cell counts had more pro-
`NATURE - VOL 373 - 12 JANUARY 1995
`Ambry Exhibit 1018
`
`Ambry Exhibit 1018
`
`
`
`TABLE 1 Summary data of H!V-1 and CD4 lymphocyte turnover during the pretreatment steady state
`
`Baseline values
`Kinetics of HIV-1 turnover
`Kinetics of CD4 lymphocyte turnover*
`Minimum
`Minimum
`
`production and destruction
`production and
`clearancet
`(virions per
`day x 10°)
`0.56
`0.26
`0.11
`0.54
`0.50
`1.27
`1.48
`1.85
`0.05
`0.51
`0.30
`0.88
`0.47
`0.21
`0.74
`1.06
`0.08
`0.08
`2.07
`0.53
`
`ARTICLES
`
`Slope
`0.070 (6.9)
`0.004 (0.5)
`0.005 (1.4)
`0.019 (1.9)
`0.055 (21.5)
`0.058 (25.7)
`0.020 (9.1)
`0.088 (11.8)
`0.038 (15.6)
`0.064 (11.0)
`0.048 (4.5)
`0.077 (5.9)
`NA
`0.014 (3.1)
`0.015 (2.4)
`0.080 (8.5)
`NA
`0.059 (3.4)
`0.073 (15.9)
`0.051 (5.6)
`
`Blood§
`(cells per day x 10°)
`21.7 (28.1)
`4.3
`(2.7)
`99
`(9.5)
`22.2 (13.0)
`108.0 (157.0)
`105.0 (150.0)
`55.9 (65.8)
`20.7 (56.6)
`71.0 (81.9)
`38.9 (62.8)
`17.0 (26.9)
`24.7 (40.5)
`NA
`13.4 (17.4)
`13.8 (18.7)
`24.6 (57.5)
`NA
`12.5 (19.7)
`35.3 (115.0)
`32.4 (34.5)
`
`Total ||
`(cells per day x 10°)
`1.1 (1.4
`0.2 (0.1)
`0.5 (0.5
`1.1 (0.6)
`5.4 (7.8
`5.3 (7.5)
`2.8 (3.3
`1.0 (2.8)
`3.6 (4.1
`2.0 (3.1
`0.8 (1.4
`1.2 (2.0
`NA
`0.7 (0.9
`0.7 (0.9
`1.2 (2.9
`NA
`0.6 (1.0)
`1.8 (5.7
`1.6 (1.7)
`
`
`
`cb4
`ceil count
`(mm7*)
`76
`209
`293
`174
`269
`312
`386
`AQ
`357
`107
`59
`47
`228
`169
`120
`46
`490
`36
`67
`103
`
`Piasma
`viraemia
`(virions per
`ml x 10°)+
`193
`80
`41
`121
`88
`475
`185
`554
`15
`130
`70
`100
`101
`55
`126
`244
`18
`23
`256
`99
`
`Patient
`301
`302
`303
`304
`305
`306
`308
`309
`310
`311
`312
`313
`401
`402
`403
`404
`406
`408
`409
`410
`
`tye
`(days)
`2.3
`2.6
`3.3
`2.5
`2.1
`1.3
`1.5
`2.4
`2.7
`2.4
`2.3
`1.3
`1.7
`2.5
`2.2
`2.6
`2.2
`2.8
`1.5
`1.9
`
`Slope
`0.30
`—0.27
`—0.21
`—0.28
`—0.33
`—0.52
`—0.46
`—0.29
`—0.26
`—0.29
`—0.30
`—0.54
`—0.40
`-0.28
`—0.32
`—0.27
`—0.31
`—0.25
`—0.47
`—0.36
`—0.214
`-0.54
`—0.34+0.06
`
`to
`
`Range
`Mean
`
`36-490
`
`180+46
`
`15-554
`134+440
`
`1.3-3.3
`
`2.1404
`
`0.05-2.07
`
`0.684 0.13
`
`0.004-0.088 (0.5-25.7) 4.3-108.0 (2.7-157.0)
`0.047 (8.6)
`35.1 (53.2)
`
`0.2-5.4 (0.1-7.8)
`1.8 (2.6)
`
`*The results for the kinetics of CD4 lymphocyte turnover generated by an exponential growth model are shown without parentheses; results generated by a linear
`production model are shownin parentheses.
`¥ Each virion contains two RNA copies.
`t Calculated using plasma and extracellular fluid volumes estimated from body weights, and assuming that plasma and extracellular fluid compartments are in
`equilibrium.
`§ Calculated using blood volumes estimated from body weights.
`|| Calculated on the assumption that the lymphocyte pool in blood represents 2% of the total population*®. NA, not analysed owing to large fluctuations in CD4cell
`counts.
`
`minentrises. This demonstrates convincingly that the CD4 lym-
`phocyte depletion seen in AIDS is primarily a consequence of
`the destruction of these cells induced by HIV-1 not a lack of
`their production.
`As ABT-538 treatment reduces virus-mediated destruction of
`CD4 lymphocytes, the observed increase in CD4 cells provides
`a minimum estimate (Fig. 3 legend) of the pretreatment CD4
`lymphocyte production rate, which in turn equals the destruction
`rate during the steady state. Minimum production (destruction)
`
`rates were calculated for each case by multiplying the slope,
`the initial CD4 cell count, and the estimated blood volume.
`The minimum numbers of CD4 cells in blood produced or
`destroyed each day ranged from 4.3 x 10° to 108 x 10°, with
`a mean of 35.1 x 10° (Table 1). Given that the blood lympho-
`cyte pool
`is about 2% of the total population'®,
`the overall
`CD4 lymphocyte turnover in our patients was calculated to
`vary from 0.2 x 10° to 5.4% 10° cells per day, with a mean of
`1.8 x 10 cells per day.
`
`constant) 0
`—Slope(HIVclearancerate
`
`
`
`300
`
`400
`
`500
`
`100
`
`200
`
`increase 0
`ExponentialslopeofCD4
`
`
`100
`
`200
`
`300
`
`400
`
`Baseline CD4 cell count
`
`Baseline CD4 cell count
`
`FIG. 2 a, Lack of correlation between viral decay slopes and disease
`status as
`indicated by baseline CD4
`cell
`counts. Correlation
`coefficient =0.05 (P value >0.1). b, Inverse correlation between the
`exponential CD4 increase slopes and baseline CD4cell counts. Correla-
`tion coefficient =—0.57 (P value <0.01). Such an inverse correlation
`
`would be expected if T-cell proliferation were governed by a density-
`dependent growth function (logistic, for example), in which the growth
`rate decreases with increasing population level, if T cells were produced
`from precursors at a constant rate or from a combination of these two
`effects.
`
`NATURE - VOL 373 - 12 JANUARY 1995
`
`Ambry Exhibit 1018
`
`125
`
`Ambry Exhibit 1018
`
`
`
`ARTICLES
`
`
`FIG. 3 Increase in CD4 cell counts after ABT-538 treatment plotted on
`a logarithmic (top) or linear (bottom) scale. Each slope was obtained
`from the best-fit line derived from linear regression on 2-4 data points.
`In the model for exponential
`increase, the doubling time was deter-
`mined by dividing In (2) by the slope. From the slope, we obtained
`minimum estimates of the CD4 lymphocyte production rate. The change
`in CD4 cell number over time can be described by the equation,
`dT/dt=P—pT, where T is the cell count, P is the cell production rate,
`and yu
`is the cell decay rate. The slope, S, on a Jog plot is thus =
`d(In T)/dt = P/T— py. Hence, S x T must be less than P, showing that our
`estimates indeed represent minimum CD4 lymphocyte productionrates.
`Using a similar argument, slopes derived from a model of linear
`increases are also minimum estimates of CD4 lymphocyte production.
`
`CD4cellcount(permm3)
`
`
`
`CD4cellcount
`
`409
`403
`303
`
`
`
`
`
`
`
`
`
`
`10 15 20 25 30
`
`
`
`
`(permm3)
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`oti :|||islope : 15.9:
`
`10-5 0
`5
`10 15 20 25 30
`-10-5 0
`5S
`10 15 20 25 30 -10 5 0
`5
`
`
`
`Days
`
`the CD4 lymphocyte depletion
`regenerative process. Second,
`seen in advanced HIV-1 infection may belikened to a sink con-
`taining a low waterlevel, with the tap and drain both equally
`wide open. As the regenerative capacity of the immune system
`is not infinite, it is not difficult to see why the sink eventually
`empties. It is also evident from this analogy that our primary
`strategy to reverse the immunodeficiency ought to be to target
`virally mediated destruction (plug the drain) rather than to
`emphasize lymphocyte reconstitution (put in a second tap).
`
`The increase in CD4 lymphocyte counts following ABT-538
`administration was also modelled linearly (Fig. 3, bottom). The
`slope of the line depicting the increase for each case was deter-
`mined, and the values varied from 0.5 to 25.7 cells per mm°*per
`day, with a mean of8.6 cells per mm* per day (Table 1). Using
`the same argument as for the exponential case, minimum esti-
`mates of total CD4 production (or destruction) rates at baseline
`were determined to vary from 0.1 x 10° to 7.8 x 10” cells per day,
`with a mean of 2.6 x 10° cells per day.
`Although our two sets of CD4 lymphocyte analyses do not
`Discussion
`yield identical numericalresults, they are in close agreement and
`Webelieve our new kinetic data have important implications for
`emphasize the same qualitative points about HIV-1 pathogen-
`HIV-1 therapy and pathogenesis. It is self evident that, with
`esis. The number of CD4 lymphocyte destroyed and replenished
`rapid turnover of HIV-1, generation of viral diversity and the
`each day is of the order of 10’, which is strikingly close to
`attendant increased opportunities for viral escape from thera-
`estimates of the total number of HIV-1 RNA-expressing lym-
`peutic agents are unavoidable sequelae’””°. Treatmentstrategies,
`phocytes in the body determined using in situ polymerase chain
`if they are to have a dramatic clinical impact, must therefore be
`reaction and hybridization methods*'’. In addition, CD4 replen-
`initiated as early in the infection course as possible, perhaps even
`ishment appears to be highly stressed in many patients in that
`during seroconversion. The rapid turnover of HIV-1 in plasma
`the faster production rates are ~25-78-fold higher than the slow-
`also suggests that current protocols for monitoring the acute
`est rate (Table 1), which is presumably still higher than the as-
`antiviral activity of novel compounds must be modified to focus
`yet-undefined normal CD4 turnover rate. The precise mechan-
`onthefirst few days following druginitiation. Ourinterventional
`isms of CD4 lymphocyte repopulation, however, will have to be
`approach to AIDS pathogenesis has shown that HIV-1 produc-
`addressed in the future by studies on phenotypic markers and
`tion and clearance are delicately balanced but highly dynamic
`functional status of the regenerating cells. Nonetheless, the rapid
`processes. Taken together, our findings strongly support the view
`CD4 lymphocyte turnover has several implications. First, the
`that AIDSis primarily a consequence of continuous, high-level
`apoptosis commonly observedin the setting of HIV-1 infection"
`replication of HIV-1, leading to virus- and immune-mediated
`O
`killing of CD4 lymphocytes.
`may simply be an expected consequenceof an active lymphocyte
`
`
`IeSOBNDARWNE
`
`Received 16 November; accepted 15 December 1994.
`Weiss, R. A. Science 260, 1273-1279 (1993).
`Ho, D. D., Moudgil, T. & Alam, M. New Engi. J. Med. 321, 1621-1625 (1989).
`. Simmonds,P. et al. J. Virol. 64, 864-872 (1990).
`Connor, R.
`|., Mohri, H., Cao, Y. & Ho, D. D. J. Virol, 67, 1772-1777 (1993).
`. Patterson, B. K. et al. Science 260, 976-979 (1993).
`Piatak, M. et al. Science 259, 1749-1754 (1993).
`Pantaleo, G. et al. Nature 362, 355-359 (1993).
`. Embretson, J. et al. Nature 362, 359-362 (1993).
`. Wain-Hobson, S. Nature 366, 22 (1993).
`. Kempf, D. et af. Proc. natn. Acad. Sci. U.S.A. {in the press).
`11. Markowitz, M. et a/. J. Virol. (in the press).
`412. Pachi, C. et ai. J. AIDS (in the press).
`13. Cao, Y. et al. AIDS Res. Human Retroviruses (in the press).
`
`126
`
`14. Perelson, A. S., Kirschner, D. E. & De Boer, R. J. Math. Biosci. 114, 81-125 (1993).
`415. Grody, W. W., Fligiel, S. & Naeim, F. Am. J. clin. Path. 84, 85-95 (1985).
`16. Paul, W. E. Fundamental immunology (Raven, New York, 1985).
`17. Haase, A. T. Ann. N. Y. Acad. Sci. 724, 75-86 (1994),
`18. Ameisen, J. C. & Capron, A. immun. Today 12, 102-105 (1991).
`19. Wain-Hobson, S. Curr. Opin. Genet. Dev. 3, 878~883 (1993).
`20. Coffin, J. M. Curr. Top. Microbiol. immun. 176, 143-164 (1992).
`
`ACKNOWLEDGEMENTS. Wethank the patients for their participation, Y. Cao and J. Wilbur for
`assistance with branched DNA assays, A. Hsu and J. Valdes for input on trial design, and J.
`Moore and R. Koup for helpful discussions. This work was supported by grants from Abbott
`Laboratories, the NIH and NYU Center for AIDS Research, the Joseph P. and Jeanne M.Sullivan
`Foundation, Los Alamos National Laboratory LRDR Program and The Aaron Diamond
`Foundation.
`
`NATURE - VOL 373 - 12 JANUARY 1995
`Ambry Exhibit 1018
`
`Ambry Exhibit 1018
`
`