JOURNAL OF INTERFERON RESEARCH 12:431^438 (1992)Mary Ann Liebert, Inc., Publishers

Acid Lability Is Not an Intrinsic Property of Interferon-a Inducedby HIV-Infected Cells

MARIA R. CAPOBIANCHI, PAOLO MATTANA, FRANCESCHINA MERCURI,GIANCARLO CONCIATORI, FRANCO AMEGLIO,' HELMUT ANKEL,2 and

FERDINANDO DIANZANI

ABSTRACT

Human immunodeficiency virus (HlV)-infected cells induce acid-labile interferon-a (al-IFN-a) in cultures ofmonon tic Iear cells from peripheral human blood. We have investigated the physicochemical properties of suchpreparations to elucidate the reasons for acid-lability of this IFN. AI-IFN-a is a mixture of both glycosylatedand unglycosylated molecules as shown by separation on Concanavalin-A Sepharose. Acid-lability is associatedonly with glycosylated molecules. Upon chromatography of the glycosylated fraction on Sepharose coupled toIFN-a-specific antibody, the portion of the IFN that is retained and eluted with guanidine-HCl is acid-stable,whereas the excluded antiviral activity is acid-labile, and is partially neutralized by antibodies to either IFN-aor IFN-7. Also, upon further purification of the unglycosylated fraction on the same antibody column, allantiviral activity remains indistinguishable from conventional IFN-a. Reconstitution experiments showed thatglycosylated material excluded from the anti-IFN-a column potentiates antiviral activity of the IFN that isspecifically retained by the column. This potentiation is abolished by acid treatment. Similar results areobtained with al-IFN-a from the serum of acquired immunodeficiency syndrome (AIDS) patients, indicatingthat its acid-lability is also the consequence of an acid-labile component that is capable of enhancing theantiviral activity. The potentiation of antiviral activity obtained by combining recombinant preparations ofIFN-a and IFN-7 suggests that the cooperating molecule is IFN-7. The presence of IFN-7 in al-IFN-a asdetermined by ELISA assay, by affinity chromatography on a column containing anti-IFN-7 antibody, and bydot-blot analysis of its mRNA in the induced cells, supports our conclusion that (i) there is no acid-labile IFN-aper se, (¡i) al-IFN-a is a mixture of conventional IFN-a and IFN-7, and (iii) the apparent acid-lability ofal-IFN-a is due to the synergistic antiviral effects of acid-stable IFN-a and acid-labile IFN-7 which are bothpresent in al-IFN-a.

IINTRODUCTION relation between disease progression and both prevalence and

titer of IFN.'37' These findings support the possibility that thisNTERFERONS (IFNs) are known to play an important role in IFN plays a role in the pathogenesis of such diseases. This IFNprimary defense against viral infections, and their presence is referred to as acid-labile IFN-a (al-IFN-a), because it is

in the circulation has been reported for several viral infec- neutralized by antibody specific for IFN-a, but, unlike all othertions.'" IFNs also possess immunoregulatory properties, and it known IFN-a subtypes, its antiviral activity is substantiallyis currently thought that development and maintenance of some inactivated by exposure to acidic pH. It has been suggested thatimmune disorders are correlated to abnormal activation of the al-IFN-a represents an unusual subspecies of the IFN-a family,IFN system.12' Significant levels of circulating IFN-a have but none of the separated or cloned IFN-a species that havebeen described for a number of autoimmune diseases and for been characterized is acid-labile. We have previously reportedacquired immunodeficiency syndrome (AIDS), with a time cor- that HIV-infected cells fixed with glutaraldehyde induce al-

The Institute of Virology, University "La Sapienza," Rome, Italy; 'Intitute San Gallicano, Rome, Italy; 2Medical College of Wisconsin,Milwaukee, WI 53226.

431

432 CAPOBIANCHI ET AL.

IFN-a in cultures of peripheral blood mononuclear cells,'8' andthat its production is triggered by a membrane interaction in-volving both HLA class II and CD4 antigens.'9'We have characterized this in v/rro-induced al-IFN-a to un-

derstand the reason for its acid-lability. In a previous study, weshowed that al-IFN-a is a mixture of both glycosylated andnonglycosylated molecules, and that acid-lability is only asso-

ciated with glycosylated molecules.'"" However, it was not

possible to define whether the glycosylated IFN is actuallyintrinsically acid-labile, or whether some other copurifying gly-coprotein is responsible for its acid-lability.This paper describes the characterization of al-IFN-a pro-

duced in vitro, and of the al-IFN-a present in the serum ofAIDS patients. Our data show that acid-lability is due to an

acid-labile molecule associated with, but distinct from IFN-a.We also present evidence for the coinduction of IFN-7 which ispresent in small amounts in crude al-IFN-a and which synergis-tically stimulates antiviral activity contained in al-IFN-a.

MATERIALS AND METHODS

Propagation of Cell Lines: H9 cells persistently infectedwith human immunodeficiency virus type 1 (HIV-1), strain IIIB(H9-HIV), were originally established by R.C. Gallo"" andwere kindly provided by M.W. Cloyd (Galveston, TX). Cellswere propagated in RPMI-1640 medium (Flow LaboratoriesInc., Irvine, Scotland), supplemented with 10% FBS and anti-biotics. A standard reverse transcriptase (RT) assay was per-formed at weekly intervals to monitor virus production by thesecells. Human amnion WISH cells were grown in RPMI-1640,supplemented with 10% FBS and antibiotics.

IFN Induction and Titration: Peripheral blood mononuclearcells (PBMC) from healthy donors, negative for anti-HIV anti-body, were isolated by Ficoll-Hypaque gradient centrifugation.They were incubated at 2.5 x 106/ml in RPMI containing anti-biotics and 10% human agamma serum obtained according toCantell et al.,'2 in the presence of the inducer cells. These were

H9-HIV cells fixed with glutaraldehyde as previously de-scribed,"3'used at a concentration of 1 x 105/ml.Under these conditions, no toxicity was observed, as cell

viability, assessed by trypan blue exclusion, exceeded 90% inall experiments. Supernatants were collected 24 h after seedingand clarified by low-speed centrifugation. IFN titration was

performed on human WISH cells by measuring inhibition ofSindbis virus hemagglutinin yield, after a single growth cycle,as previously described.114) The antiviral activity found was

characterized by neutralization with specific antisera. Neutral-ization was performed by incubating the IFN in different dilu-tions for 1 h at 37CC with a 10-fold excess of the followingantisera: polyclonal sheep anti-native IFN-a (Shering Co.,Bloomfield, NJ, titer 1:200,000 against 10 IU of IFN-a), anti-recombinant IFN-7 polyclonal rabbit antibody (obtained fromJ. Wietzerbin, Paris, France; titer 1:40,000 against 10 IU ofIFN-7), and mouse monoclonal antibody against IFN-7 (Hoff-man La-Roche; titer 1:32,000 against 10 IU of IFN-7). Resid-ual antiviral activity was evaluated on WISH cells.

To determine the stability of an IFN sample at acidic pH, 80p.1 of IFN solution were adjusted to pH 2 by addition of 80 p.1 of0.1 N HC1 in RPMI, incubated for 24 h at 4°C, and returned topH 7 with 40 p.1 of RPMI containing 0.2 N NaOH. Controlsreceived equivalent volumes of RPMI containing previouslycombined HC1 and NaOH. Recombinant human IFN-a andIFN-ß were completely stable during this procedure, whereasrecombinant human IFN-7 was inactivated more than 30-foldunder these conditions. Immunoreactive IFN-a was detected byan ELISA from Anawa International Bioscience (Zurich, Swit-zerland). This test utilizes two monoclonal antibodies againstIFN-a2. Because its cross-reactivity with other IFN-a subspe-cies is not known, results were not corrected according to themanufacturer's instructions, and are expressed in terms ofIFN-a2 units, instead of international units. ELISA for IFN-7was purchased from Med Genix Diagnostics (Brussels, Bel-gium).

Affinity chromatography on Concanavalin A-Sepharose:Concanavalin A (ConA) coupled to Scpharose was obtainedfrom Pharmacia and used according to the manufacturer's in-structions. Al-IFN-a preparations were partially purified byammonium sulfate precipitation between 30 and 65% saturationaccording to Capobianchi et al."0) to a specific activity of I04 7

IU/mg. One milliliter of the precipitate dissolved in phosphate-buffered saline (PBS), extensively dyalized against loadingbuffer (0.05 M NaCl in 0.02 M Tris-HCl, pH 7.4) was loadedon a 1-ml column previously equilibrated with loading buffercontaining also 0.1% bovine serum albumin. The protein elu-tion profile was determined by monitoring UV absorption, andafter extensive washing 0.5 M a-methyl-D-mannoside was

added to the elution buffer. Fractions of 2 ml were collected andtitrated as described. The antiviral activity present in both theexcluded and in the retained fractions after ConA-Sepharosechromatography accounted for 80-90% of the applied antiviralactivity. Protein concentration was assayed using the proteinreagent from Pierce (Oud Beijerland, The Netherlands).

Affinity Chromatography on Anti-IFN-a Columns: Cyano-gen bromide-activated Sepharose 4B from Pharmacia was con-

jugated according to the manufacturer's instructions, withpurified rabbit polyclonal IgG against purified native lympho-blastoid IFN-a (Welferon, Wellcome, Beckenam, UK) con-

taining most subspecies of the IFN-a family.'I5) This antibodypreparation had no neutralizing activity against human IFN-7.After extensive washing with PBS, the affinity gel suspen-

sion (1 ml for al-IFN-a produced in vitro or 0.2 ml for thatpresent in AIDS sera), was washed three times with 2 volumesof sodium citrate (0.1 M, pH 3) and three times with 2 volumesof PBS. The IFN samples were then incubated overnight withthe gel under gentle rotation at 4°C. Then the gels were loadedon disposable columns and eluted by the sequential addition ofthree volumes of the following: PBS, 8 M guanidine-HCl(R.M. Friedman, Bethesda, MD, personal communication);PBS, and 50% ethylene gl.ycol in PBS containing 1 M NaCl. Inprevious experiments, the eluting reagents have been shown tohave no effect on titer and acid-lability of the starting al-IFN-a.The column was regenerated with 0.1 M sodium citrate pH 3.To all fractions 0.1% bovine serum albumin in 0.1 M Tris-HCl,pH 7.2, was immediately added for stabilization. All samples

HIV-INDUCED IFN-a IS NOT INTRINSICALLY ACID-LABILE 433

Table 1. Awn-IFN-a Affinity Chromatography of ConA Non-retained (ConA )Antiviral Activity

Antiviral activity (units)Total Acid stable

Specific activity(log U/mg protein)

AppliedExcludedEluted

60,000180

55,000

60,000<20

55,000

4.11.3

>6.5

were extensively dialyzed against PBS and then titrated. Acid-lability of the fractions was tested as above described.

Affinity Chromatography on Anti-lFN-y Columns: The af-finity gel was prepared by coupling cyanogen bromide-acti-vated Sepharose 4B with purified rabbit polyclonal IgG againstrecombinant IFN-y as described for the anti IFN-a column.This antibody preparation had no neutralizing activity againsthuman IFN-a.

One hundred microliters of the IFN samples containing60,000 U of IFN before, and 18,000 U after treatment at pH 2,were adjusted to 1 ml with PBS and loaded on 1.5 ml of the gelas described for the anti IFN-a column. The chromatographywas monitored by UV absorbance, and buffers were changedafter stabilization of the UV absorbance curve. After extensivewashing with PBS, the elution was performed with 50% ethyl-ene gycol in PBS containing 1 M NaCl; then the column was

washed with PBS, and finally 0.1 M sodium citrate, pH 3, wasadded to regenerate the gel. Fractions of 1.5 ml were collectedand titrated on WISH cells starting at 1 : 10 dilution, withoutprevious dialysis.

Demonstration of IFN-y-Specific mRNA: The presence ofIFN-y-specific message was shown by dot-blot hybridization oftotal RNA extracted from PBMC induced with HIV-infectedcells with an IFN-y-specific cDNA probe. As positive control,RNA from staphylococcal enterotoxin B(SEB)-stimulatedPBMC was used. RNA extracted from unstimulated PBMCserved as negative control.Specifically, 3 x 107 PBMC were incubated overnight with

individual inducers and collected after repeated washing withPBS. Total RNA was isolated by acid guanidinium thiocyanatephenol extraction."6' Ten micrograms of denatured RNA fromthe various experimental conditions was immobilized onto a

nylon membrane (Amersham, UK), using a dot-blot apparatus(Bio-Rad Inc. CA) and was hybridized with a 32P-labeled probe

specific for human IFN-y. This was the 1.2-kb Pst I insert fromthe clone pRJ22,"7' kindly provided by C.W. Dieffenbach(Bethesda, MD). The purified cDNA fragment was labeledwith [32P]ATP by nick-translation to a specific activity of 50u.Ci/p.g. Hybridization was performed by standard procedures,and filters were exposed to a XAR.5 X-ray film (EastmanKodak, Rochester, NY) with an intensifier screen for 72 h at-70°C.

RESULTS

Characteristics of al-IFN-a upon anti-IFN-a affinitycolumn separationA pool of in v/rro-produced al-IFN-a was concentrated by

precipitation between 30 and 65% ammonium sulfate saturationas described under Materials and Methods. The precipitate wasdissolved in PBS and chromatographed on a ConA-Sepharosecolumn, which is known to retain most N-linked glycoproteinsby specifically binding to mannosyl residues of the glycosidicchains."8' As previously shown, about 60% of the starting IFNwas recovered in the excluded fractions (unglycosylated frac-tion), whereas about 30% of the starting material was elutedwith a-methyl-D-mannoside (glycosylated fraction). As previ-ously observed, only the glycosylated IFN was acid-labile.'""The peak fractions of either bound or unbound material werepooled separately and used for further purification on anti-IFN-a affinity columns. The results of representative experi-ments are shown in Tables 1 and 2. The IFN in the nonglycosy-lated fractions was almost totally recovered (>90% of the inputmaterial) after elution with guanidine-HCl and remained acid-stable, identical to the input material. However, the glycosy-lated material was only partially retained by the antibody col-umn (about 30% of the starting material), whereas about 10% of

Table 2. ANTi-IFN-a Affinity Chromatography of ConA Retained (ConA+)Antiviral Activity

Antiviral activity (units)

Total Acid stableSpecific activity

(log U/mg protein)AppliedExcludedEluted

18,0002,2005,700

4,500<2205,700

4.73.7

>6.5

434 CAPOBIANCHI ET AL.

Table 3. Acid-Labile Potentiation of Antiviral Activity of ANTi-IFN-a Affinity Column-Retained Fractions byTHE NONRETAINED FRACTION

Fraction

Antiviral activity (units/ml)

Measured Expected

TotalAcidstable Total

Acidstable

Ratio measured!expected

TotalAcidstable

Experiment 1a. ConA, Ab+a 4,000b. ConA +

, Ab_a 240c. ConA+, Ab+a 500

a + bb 4,000a + cb 2,000b + cb 800

Experiment 2d. ConA+, Ab~a 200e. ConA+,Ab++ 200

d + eb 400

4,000<30500

2,0002,000240

>20200100

2,1202,250370

200

2,0002,250250

100

1.90.92.2

2.0

1.00.91.0

1.0

"ConA * or ConA , retained or not retained by ConA; Ab+ or Ab , retained or not retained by anti-IFN-a.bEqual volumes of indicated fractions were mixed prior to titration of antiviral activity.

the starting antiviral activity was excluded. No additional IFNwas recovered in subsequent elutions, suggesting that no IFNwas retained on the antibody column after the guanidine-HClelution. Assessment of acid-lability of the various fractionsrevealed that, while the input IFN was acid-labile (25% residualactivity after treatment at pH 2), the IFN retained by the columnwas completely acid-stable. Conversely, the excluded antiviralactivity was very sensitive to acid treatment (<10% residualactivity after exposure to pH 2) as shown in Table 2. Theseresults were highly reproducible, the total recovery from anti-body columns ranging from 30 to 40% in several experiments.Furthermore, very comparable results were obtained by separat-ing the ammonium sulphate fraction of al-IFN-a directly on theantibody column without previous fractionation on ConA-Sepharose (not shown). Under these conditions total recoverywas about 30%, with 3-8% of the activity present in the un-

bound fraction and showing high acid-lability, and 25-30% ofthe activity recovered after elution with guanidine-HCl, show-ing complete resistance to acid treatment.

Restoration of acid lability byreconstitution experimentsThe above results suggested that the al-IFN-a produced

in vitro is not intrinsically acid-labile, and that this property isconferred to IFN-a by a glycoprotein component not retainedby the anti-IFN-a columns. The consistent loss of part of theinitial antiviral activity following antibody column separationled us to hypothesize that the antiviral activity of the startingmaterial is the result of a synergistic cooperation between theIFN-a molecules and an unidentified acid-labile component.To verify this hypothesis, we performed reconstitution experi-ments by combining the glycosylated, antibody-excluded frac-tion with IFN-a antibody-retained fractions. Antiviral activitywas measured before and after treatment at pH 2, and the

resulting titers were compared to the expected titers calculatedfor the combinations. Results are reported in Table 3. As can beseen, the excluded material containing acid-labile antiviral ac-tivity is able to potentiate both glycosylated and unglycosylatedacid-stable IFN-a, as the observed titers are about 200% of theexpected titers. This potentiation was abrogated by acid treat-ment. These results directly demonstrate that antiviral activityof crude al-IFN-a is the consequence of an overestimation of itstiter, which is due to a potentiating, acid-labile glycoproteinwith no affinity for anti IFN-a. Further experiments were aimedat identifying the cooperating molecule(s).

Evidence for the presence of IFN-y in al-IFN-a

It has previously been shown that IFN-7 can synergisticallyactivate the antiviral effects of IFN-a in antiviral assays."9'2"'To demonstrate the role of IFN-7 as the acid-labile glycoproteinresponsible for potentiating the antiviral effect of the acid-stableIFN-a present in al-IFN-a preparations, we used several ap-proaches.

In neutralization experiments, increasing dilutions of ammo-nium sulfate-precipitated al-IFN-a and of the material excludedby IFN-a antibody columns were incubated for 1 h at 37°C withantibodies against IFN-a and IFN-7, and the residual antiviralactivity was determined. The results are shown in Table 4.Antiviral activity of this al-IFN-a was strongly, although not

completely, inactivated by antibody to IFN-a; furthermore itwas not affected by antibody to IFN-7. The combination of bothantibodies completely suppressed antiviral activity. The IFNtiter of the material excluded from the anti-IFN-a column was

partially reduced by either of the two antibodies, but was com-

pletely suppressed by the combined treatment with anti-IFN-aand anti-IFN-7. Thus, although antibody to IFN-7 does notseem to affect antiviral activity of crude al-IFN-a preparations,it seems that al-IFN-a contains small amounts of IFN-7 that can

HIV-INDUCED IFN-a IS NOT INTRINSICALLY ACID-LABILE 435

Table 4. Effects of Various Antibodies on AntiviralActivity of Crude and Glycosylated FractionExcluded by ANTi-IFN-a Column (ConA+, Ab~)

Residual activity (units/ml)

Antibody added Starting" Cona+, Ab~

Nonea. anti-IFNab. anti-IFN-ya + b

1285

128<2

1283216<2

"After 35-60% ammonium sulfate precipitation.

Table 5.in ANTi-IFN-a Column Fractions

SamplelFN-a2

(total units)IFN-y

(total units)

Starting 6,800 145a. Nonglycosylated (ConA-)Applied 5,400 <10Excluded (Ab") 24 <10Eluted (Ab+) 3,900 <10

b. Glycosylated IFN (ConA+)Applied 1,170 164Excluded (Ab~) 20 123Eluted (Ab+) 1,400 <10

One hundred microliters ofammonium sulfate-precipitated al-IFN-awas separated with Con A-Sepharose in nonglycosylated (ConA-) andglycosylated (ConA + ) fractions. The two fractions were individuallyapplied on an anti-IFN-a antibody column, and the material retained bythe gel (Ab+) was eluted with guanidine-HCl as described under Mate-rials and Methods. Immunoreactive IFN-a2 and IFN-y in individualfractions were evaluated by ELISA.

be separated from the starting mixture by means of antibodychromatography. In fact, determination of immunoreactiveIFN-a and IFN-y by ELISA showed that the starting al-IFN-acontains both IFN types, and that IFN-y molecules segregatewith glycosylated molecules that are excluded by the antibodycolumn (Table 5). We set up a model of cooperation betweenIFN-a and IFN-y using experimental mixtures of recombinantpreparations. Results of such experiments, indeed, confirm thatthe combination of recombinant IFN-y and IFN-a at ratios ofbetween 1 : 0.5 and 1 : 12 respective units yields a potentiationof antiviral activity ranging from 2.3- to 3.2-fold above thetiters expected by simple addition of IFN units (refs. 19 and 20,and data not shown). When such mixtures were preincubated at

pH 2, no potentiation was observed.In the next set of experiments, ammonium sulfate precipi-

tated al-IFN-a preparations either before and after treatment atpH 2 were separated by affinity chromatography on IFN-y-specific antibody.As seen in Fig. 1, the elution profiles clearly demonstrate that

acid treatment specifically affects the antiviral activity of theIFN with affinity for IFN-y antibody, whereas the antiviralactivity excluded by this column remains unchanged after acidtreatment. In fact, the excluded peak contained the same

amount of antiviral units, regardless of whether or not theapplied sample had been pretreated at pH 2. These units ac-

counted for about 20% of the starting antiviral activity, andcontained >90% of the starting IFN-a, as determined byELISA. These results support those obtained with the antiIFN-a column, demonstrating that effective separation of theIFN-a and IFN-y molecules causes a loss of antiviral activity,and that the only acid-labile IFN component present in theal-IFN-a copurifies with IFN-y.To support coinduction of IFN-y further, its mRNA was

analyzed in cellular extracts from PBMC induced with H9-HIV, using dot blot analysis as described under Materials andMethods. As shown in Fig. 2, while the IFN-y message was not

IFN titer (U/ml) IFN tltet (U/ml)

1 3 5 7 9 11 13 15 17 19 21 23 25 27

fraction number1 3 5 7 9 11 13 15 17 19 21 23 25 27

fraction numberFIG. 1. Affinity chromatography of al-IFN-a before (A) and after (B) acid treatment on anti-IFN-y antibody column. The twoarrows indicate the addition of the eluant (ethylene glycol) and of PBS, respectively.

436 CAPOBIANCHI ET AL.

FIG. 2. Dot blot hybridization of PBMC RNA with an IFN-7-specific cDNA probe. A. PBMC induced with SEB (0.5p-g/ml). B. Uninduced PBMC. C. PBMC induced with H9-HIV.

Table 7. Effect of Anti-IFN-u AffinityChromatography on Acid Lability of Antiviral

Activity from Pooled AIDS Sera3

Antiviral activity (units)

Fraction Total Acid stable

AppliedExcludedRetained

450<10200

150<10200

"After ammonium sulfate precipitation.

DISCUSSION

detected in uninduced PBMC, it was clearly present in PBMCthat were induced with either H9-HIV or the known IFN-7inducer SEB.

Purification of al-IFN-a present in the serum ofAIDSpatients

We then investigated whether the results obtained with al-IFN-a produced in vitro also apply to the al-IFN-a present inthe serum of HIV-infected patients. We employed anti-IFN-aaffinity chromatography to investigate a pool of sera from pa-tients that were selected for the presence of high titers of al-IFN-a. The antiviral activity of these sera was reduced to vari-ous extents (13-40%) after exposure to acid, and was largelyneutralized by treatment with antibody to IFN-a, as shown inTable 6.

Comparable amounts (about 2 ml) of these sera were pooledand subjected to ammonium sulfate precipitation between 30and 65% saturation. The precipitate was dissolved in PBS andapplied to an anti IFN-a affinity column. As shown in Table 7,the retained and subsequently eluted IFN was completely stableto acid treatment. No antiviral activity was observed in the unre-

tained fractions, and the recovery was about 45% of the startingantiviral activity. Substantially the same results were obtained,when pooled sera were applied to the antibody column withoutprior precipitation by ammonium sulfate (not shown).

Circulating al-IFN-a appears to be a characteristic feature ofautoimmune diseases and AIDS. It is also produced in vitro byPBMC stimulated with HIV-infected cells, by monocyte-mac-rophage cultures infected with HIV, and by B-cell lines infectedwith human T-cell leukemia virus I.'8'2 ',29> Although the possi-bility that the abnormal acid sensitivity of this IFN-a is theconsequence of a unique structural difference from other knownacid-stable IFN-a subtypes is attractive, no direct evidence forthis assumption has been reported and the true reason for acid-lability has remained obscure. The suggestion that al-IFN-abelongs to the IFN-a II subfamily'23' does not apply to our

system, as antiviral activity of al-IFN-a is not neutralized bymonoclonal antibodies to IFN-to,, which was provided by Dr.G.R. Adolf of the Ernst Boehringer Institute, Vienna, Austria(unpublished results).In this study we have characterized an in v/fro-produced

al-IFN-a, and have shown that the IFN-a molecules, afterpurification with antibody chromatography, are intrinsicallyacid-stable. We report the same evidence for the IFN-a presentin the serum of AIDS patients, although in this case the scarcityof the available material precluded characterization of the po-tentiating antiviral activity not retained by the anti-IFN-a anti-body column. Our results substantially agree with previous datashowing that al-IFN-a found in the serum of patients withsystemic lupus erythematosus (SLE) is not intrinsically acid-labile,'23' although in the latter case the mechanism for instabil-ity to acid treatment seems to be different. Indeed, potentiationof antiviral activity due to cooperation with an acid-labile

Table 6. Properties of Antiviral Activity in the Serum of HIV-Infected Patients

Serum samplefrom patient12345

Total

8641281632

Antiviral activity (units/ml)

Afteranti-IFN-a

<2<2<2<4<4

Afteranti-IFN-y

321281232

AfterpH 2

22416412

HIV-INDUCED IFN-a IS NOT INTRINSICALLY ACID-LABILE 437

glycoprotein, i.e., IFN-y, appears to be the determinant in our

case and argues against an acid-activated protease, as suggestedin SLE.'24' Also, we can exclude a contribution of tumor necro-sis factor-a (TNF-a) in potentiating the antiviral activity ofIFN-a in our experimental system, since antibody to this lym-phokine does not neutralize the antiviral activity either of start-ing al-IFN-a or of the fraction separated by antibody chroma-tography (results not shown). We should like to point out thatpurification of al-IFN-a from SLE patients on affinity columnswith anti-IFN-a yielded only 35% of the starting antiviral activ-ity.'25' Also, the acid-lability of the purified IFN-a was notdetermined. Furthermore, the anti-IFN-a-excluded materialwas not considered in these experiments. Previous results byGreen showed that al-IFN induced by herpes simplex virus inhuman PBMC is a synergistic mixture of varying concentra-tions of IFN-a and IFN-y.'26' Furthermore, in PBMC cultures.Dengue virus-infected cells induced antiviral activity that wasneutralized by antibody to IFN-a and not by antibody to INF-y,but the presence of small amounts of IFN-y was shown byradioimmunoassay (RIA).'27' Similar to our results, in this sys-tem the antiviral activity measured in the bioassay was higherthan that expected on the basis of the IFN molecules evaluatedby the RIA, but, unfortunately, acid-lability was not tested.The IFN present in HIV-infected patients as well as in au-

toimmune diseases is referred to as either IFN-a or IFN-y whenimmunoreactive molecules are evaluated'6'28' or as al-IFN-awhen tested by antiviral assay.'3-5'7'21'23,29' Also, several re-ports show that the antiviral activity present in the serum, whilenot being affected by antibody to IFN-y, is not completelyneutralized by antibody to IFN-a. The addition of combinedantibodies to IFN-a and to IFN-y completely suppresses theresidual antiviral activity,'29' as also shown in our experimentalsystem. Based on these considerations, the increase of acidlability of the IFN present in the serum from HIV-infectedpatients concomitant with disease progression'3'" could be sec-

ondary to the increase of the relative amount of IFN-y(6'2s'causing a parallel increase of acid-labile potentiation of the totalantiviral activity.

One aspect of our study is intriguing and is presently underinvestigation. Although antiviral activity of al-IFN-a is a resultof potentiation by IFN-y, and during purification acid-labilitysegregates with the small amounts of IFN-y present in thepreparation, treatment with antibody to IFN-y alone does notaffect the activity of unfractionated al-IFN-a preparations, asone would expect. Thus, at the present time we hypothesize thatthe potentiating effect of IFN-y, in contrast to its antiviralactivity, is not neutralized by its interaction with IFN-y specificantibody.

ACKNOWLEDGMENTS

This work was supported by grants from CNR, N. PO.00032PF70. Ministère délia Sanità (fondi AIDS), and FondazioneCenci Bolognetti to F.D., and MPI (fondi 60%) to Maria Rosa-lia Capobianchi. H.A. was in the Institute of Virology of Romeas a recipient of Senior International Fellowship from the Fog-arty Center, National Institutes of Health.

We thank Robert Friedman for helpful suggestions and dis-cussion, leanne Wietzerbin for providing us polyclonal anti-

body to recombinant IFN-y, and Carl Dieffenbach for giving us

the IFN-y-specific cDNA clone.The technical assistance of M.C. Privitera and the secretarial

help of S. Tamburrini are gratefully acknowledged.

REFERENCES

1. BARON, S., COPPENHAVER, D.H., DIANZANI, F., FLEISH-MAN, W.R., Jr., HUGHES. T.K., Jr., KLIMPEL, GR., NIE-SEL, D.W.,STANTON,G.J.,andGYZlNG, S.K. eds. (1992).Interferon: Principles and Medical Applications. UTMB,Galveston.TX.

2. SKURKOVICH, S., SKURKOVICH, B.. and BELLANTI, JA.(1987). A unifying model of the immunoregulatory role of theinterferon system: can interferon produce disease in humans'?Clin. Immuol. Immunopathol. 43, 362-373.

3. EYSTER, M.E., GOEDERT, J.J., POON, M.C, and PREBLE,O.T. (1983). Acid labile a-interferon. A possible preclinicalmarker for the acquired immunodeficiency syndrome in hemo-filia. N. Engl. J. Med. 309, 583-585.

4. PREBLE. O.T., HOOK, A.H.. QUINNAN, G.V., et al. (1984).Role of interferon in AIDS. Ann. N.Y. Acad. Sei. 437, 64-75.

5. DE STEFANO, E., FRIEDMAN, R.M., FRIEDMAN-KIEN,A.F., GOEDERT. J.J., HENRIKSEN. D., PREBLE, O.T.,SONNABEND, J.A., and VILCEK, J. (1982). Acid-labile hu-man leucocyte interferon in homosexual man with Kaposi's sar-coma and Lymphadenopathy. J. Infect. Dis. 146, 451-455.

6. ROSSOL. S., VOTH, R., LAUBENSTEIN, HP., et al. (1989).Interferon production in patients infected with HIV-1. J. Infect.Dis. 159,815-821.

7. HOOKS. J.J., MOUTSOPOULOS, H.M., and NOTKINS, A.L.(1981-1982). Circulating interferon in human autoimmune dis-eases. Tex. Rep. Biol. Med. 41, 164-168.

8. CAPOBIANCHI. MR., DE MARCO, F., DI MARCO, P., andDIANZANI, F. (1988). Acid labile human interferon alpha pro-duction by peripheral blood mononuclear cells stimulated by HIV-infected cells. Arch. Virol. 99, 9-19.

9. CAPOBIANCHI. MR., MALAVASI, F., MATTANA, P.,MERCURI. F., and DIANZANI, F. (1990). Membrane interac-tions involved in the induction of acid labile interferon alpha. J.Biol. Reg. Homeost. Agents 4, 19-24.

10. CAPOBIANCHI, M.R., MATTANA, P., GENTILE, M., CON-CIATORI, G.C., ANKEL, H., and DIANZANI, F. (1991). Roleof glycosylation in the susceptibility of "acid labile" interferonalpha to acid treatment. J. Biol. Reg. Homeost. Agents 5, 103-109.

11. POPOVIC, M.. SARNGADHARAN, M.G., READ. E., andGALLO, R.C. (1984). Detection, isolation and continuous pro-duction of cytopathic retroviruses (HTLV-III) from patients withAIDS and pre-AIDS. Science 224, 497-500.

12. CANTELL, K., HIRVONEN, S.. KAUPPINEN, H.L., andMYLLYLA, G. (1981). Production of interferon inhuman leuko-cytes from normal donors with the use of Sendai virus. MethEnzymol. 78, 29-38.

13. CAPOBIANCHI. M.R., FACCHINI, J., Di MARCO, P.. AN-TONELLI, G., and DIANZANI, F. (1985). Induction of alphainterferon by membrane interaction between viral surface andperipheral blood mononuclear cells. Proc. Soc. Exp. Biol. Med.178,551-556.

14. STANTON, G.J., LANGFORD, M.J., and DIANZANI, F.(1981). Virus yield reduction assay for interferon by titration ofSindbis virus hemagglutinin. Meth. Enzymol. 78, 351-357.

15. ZOON, K.. ZUR NEDDEN, D.L., ENTERLINE, J.C., MANIS-CHEVITZ, J.F., DYER, D.R., BOY-KINS, R.A.. BEKISZ, J.,

438 CAPOBIANCHI ET AL.

and GERRARD, T.L. (1987). Chemical and biological character-ization of natural lymphoblastoid interferon alphas, in: The Biol-ogy of the Interferon System, 1986. K. Cantell and H. Schellekens(eds.). Amsterdam: Martinius Nijhoff. pp. 567-569.

16. CHOMOCZYNSKI, P., and SACCHI, N. (1987). Single stepmethod of RNA isolation by acid guanidinium thiocynate-phenol-chloroform extraction. Anal. Biochem. 162, 156-161.

17. DEVOS, R., CHERDUTRE, H., TAYA, Y., DEGRAVE, W.,VAN HEUVERSWYN, H., and FIERS, W. (1982). Molecularcloning of a human immune interferon cDNA and its expression ineukaryotic cells. Nucleic Acids Res. 10, 2487-2501.

18. LLOYD, K.O. (1970). The preparation of two insoluble forms ofthe phytohemagglutinin, concanavalin A, and their interactionswith polysaccharides and glycoproteins. Arch. Biochem. Bio-phys. 137,460-467.

19. FLEISCHMANN, W.R. JR., FLEISCHMANN, CM., and FI-ERS, W. (1984). Potentiation of interferon action by mixtures ofrecombinant DNA-derived human interferons. Antivir. Res. 4,357-360.

20. OLESZAK, E, and STEWART, WE. (1985). Potentiation of theantiviral and anticellular activities of interferons by mixtures ofHuIFN-7 and HuIFN-a or HuIFN-ß. J. Interferon Res. 5, 361-371.

21. SZEBENI, J, DIEFFENBACH, C, WAHL, S.M., et al. (1991).Induction of interferon by human immunodeficiency virus type 1in human monocyte-macrophage cultures. J. Virol. 65, 6362-6364.

22. BOUMPAS, D.T., HOOKS, J.J., POPOVIC, M., TSOKOS,G.C., and MANN, D.L. (1985). Human T-cell leukemia-lym-phoma virus I and/or Epstein-Barr virus-infected B-cell linesspontaneously produce acid-labile a-interferon. J. Clin. Immunol.5, 340-344.

23. KONTSEK, P., BORECKY, L., and NOVAK, M. (1991). Arethe acid-labile Inteferon-ot and Interferon w-l Identical? Virology181,416-418.

24. YEE, A.M.F., BUYON, J.P.. and YIP, Y.K. (1989). Interferon

alpha associated with systemic lupus erythematosus is not intrinsi-cally acid labile. J. Exp. Med. 169, 987-993.

25. PREBLE, O.T., BLACK, R.J., FRIEDMAN, R.M., KLIPPEL,J.H., and VILCEK, J. (1982). Systemic lupus erythematosus:presence in human serum of an unusual acid-labile leukocyteinterferon. Science 216, 429-431.

26. GREEN, J.A. (1989). Virus induced immune interferon containsboth interferon-a and -7. J. Infect. Dis. 160, 543.

27. KURANE, I., MEAGER, A., and ENNIS, FA. (1986). Induc-tion of Interferon alpha and gamma from human lymphocytes byDengue virus-infected cells J. Gen. Virol. 67, 1653-1661.

28. CORDIALI FE1, P., MASSA, A., PRIGNANO, G., PI-ETRAVALLE, M., ALEMANNO, L., VITELLI, G., PALA-MARA, G., GIGLIO, A., GANDOLFO, G.M., GENTILI, G.,CAPRILL1, F., and AMEGLIO, F. (1992). Behavior of severalprogression markers during the HIV-Ab seroconversation period.Comparison with later stages. J. Biol. Reg. Homeost. Agents 6,57-64.

29. LAU, AS., DER, S.D., READ, SE., and WILLIAMS, B.R.G.( 1991 ). Regulation of tumor necrosis factor receptor expression byacid-labile interferon-a from AIDS sera. AIDS Res. Human Ret-rovir. 7, 545-552.

30. BUIMOVICI-KLEIN, E., LANGE, M., KLEIN, R J, GRIECO.M.H., and COOPER, L.Z. (1986). Long-term follow-up of se-rum-interferon and its acid-stability in a group of homosexualmen. AIDS Res. 2, 99-103.

Address reprint requests to:Dr. Maria R. Capobianchi

The Institute of VirologyViale di Porta Tiburtima, 28

00185 Rome, Italy

Received 7 February 1992/Accepted 3 August 1992