Disintegration of Retroviruses by Chelating Agents

By V. Wunderlich and G. Sydow
Central Institute for Cancer Research, Robert-Rossle-Institute,
Academy of Sciences of the German Democratic, Berlin
German Democratic Republic
With 2 figures Accepted April 2, 1982

Summary

Exposure in vitro of various mammalian retroviruses to the chelating agents EDTA or EGTA in millimolar concentrations resulted in partial disintegration of viral membranes as measured by accessibility or even release of reverse transcriptase, an internal viral protein, without any other treatment usually required. Among the viruses responding to chelators were mammalian type C viruses, primate type D viruses and Bovine leukemia virus. The effect was dose-dependant. The avain type C virus AMV, howeveer, was found to be not susceptible to the agents. Rauscher mouse leukemia virus treated in vitro with EDTA or EGTA showed reducedinfectivity in mice. The results are considered as evidence for some association of divalent cations with membranes of mammalian retroviruses. The disintegrating activity of EGTA suggests that Ca2+ is an integral constituent of viruses but Mg2+ may also be involved. These cations seem to be responsible for maintaining the integrity of retroviral membranes which, after chelation of ions, are either disrupted or become permeable for the exogenous template of reverse transcriptase. In addition, the disintegrating activity of trifluoperazine may indicate that a calmodulin-like protein occurs in retroviral membranes.

Introduction

Retroviruses contain a single stranded RNA genome and the enzyme reverse transcriptase (RT), both of which are located with the virion core. Viral particles are released by budding from the surface of infected cells and the cores are thought to acquire in this process an outer unit membrane of host cell origin. However, little is known about the composition, architecture and topography of the viral membrane witht he possible exception of spikes and knobs protruding from the exterior of the envelope. The porphological elements are composed of predominantly glycosylated proteins encoded in the viral genome and determine many of the biological properties of the virus. (see 23 for a review). Therefore, previous investigations on the retroviral envelope have mainly been focused on those components.

The demonstration of RT activity requires the disruption of virus particles as usually performed with the aid of detergents. Our observation that the preparation of type D retroviruses in buffers containing the chelating agent ethylene-diaminetetraacetic acid (EDTA) results in considerable accessibility or even release of RT activity without the use of any detergent (32) led us to investigate more systematically the action of chelating agents on different types of retroviruses. The results presented in this study demonstrate that susceptibility to such agents is a characteristic property of various retroviruses. Thus, the structure of retroviral membranes obviously stabilized by divalent cations may be more complex than hitherto known. Appreciation of this complexity may allow a better understanding of certain aspects of virus-cell interaction and may facilitate the future design of antiretroviral compounds.

Materials and Methods

Chemicals

EDTA (Chelaplex III, p.a.) was purchased from VEB Berlin-Chemie, Berlin-Adlershof, GDR. EGTA [ethylene glycol bis 92-aminoethyl ether)-N,N,N',N'-tetraacetic acid, Dr. Th. Schuchardt GmbH., Munchen, FRG) was kindly provided by Dr. H. Will, Berlin. Propranol was obtained from Isis-Chemie KG< Zwickau, GDR, either as solution 9 1 mg/ml) ready for injection or ina solid form. Crystalline samples of congeners of propranol were generously furnished by the Imperial Chemical Industries Ltd., Pharmaceutical Division, Macclesfield, CHeshire U.K. (practolol, atenolol) and the Mead Johnson Pharmaceutical Division, Evansville, Indiana, U.S.A., through courtesy of Dr. K. W. Wheeler (sotalol-HCL). Trifluoperazine was purchased from Smith Kline & French laboratories, Hertfordshire, U.K., as a solution. The same company kindly supplied TFP in a crystalline form.

Viruses

Avian myeloblastis virus (AMV) was isolated from the plasma of leukemic chickens. Raucher leukemia virus (RLV) was isolated from the plasma of RLV- infected leukemic NMRI mice or from supernatants of the persistently infected continuous mouse cell line Y 8e. The other viruses were propagated in tissue culture. Simian sarcoma virus (SSV) containing an excess of simian sarcoma associated virus (38) was grown in the human lymphocyte cell line NC 37. Bovine leukemia virus (BLV) was isolated from chronically infected fetal lamb kidney cells (34) provided by Dr. A. Burny, Brussels, Belgium. Mason-Pfizer monkey virus (MPMV), originally isolated from rhesus monkeys (8), was grown either on the human ovarial carcinoma cell line Tu 197 (14) or the human rhabdomyosarcoma call line A 204. Squirrel monkey retrovirus (SMRV) (15) was also propagated in A 204 cells. PMF virus (PMFV), a retrovirus first isolated from malignant permanent human cells (14) and shown to be closely related to but different from MPMV (33) was grown in Tu 197 or A 204 cells. SSV, MPMV, and SMRV as well as uninfected NC 37 and A 204 cells were originally provided by the Viral Oncology Program, Office of Program resources and Logistics, NCI Bethesda, MD, U.S.A., through the courtesy of Dr. J. Gruber. Virus producing cells were maintained in suspension (A204, NC 37) or monolayer cultures (all others) in glass flasks in either RPMI 1640 or Eagle MEM supplemented with 5-10 per cent heat-inactivated fetal or normal calf serum and antibiotics. Culture medium was daily replaced and cells were grown to near confluency.

Viruses were concentrated from clarified (14,000 x g, 15 minutes) plasma or tissue culture fluid by centrifugation (50,000 x g, 120 minutes) through a cushion of 30 percent sucrose (w/v) in TN buffer (0.01 mol/l Tris-HCl, 0/1 mol/l NaCl, ph 8.3). Sometimes viruses were further purified using both velocity sedimentation and isopycnic centrifugation in sucrose density gradients (33). To keep the level of spontaneous release of RT activity as low as possible, freshly harvested virus and a short preparation procedure were usually preferred but there was, in principle, no difference in the response of semi- or highly-purified virus to the agents under study. If necessary, virus preparations were stored at -20C until used.

In Vitro Protocol

As described elsewhere (41), two approaches were used to evaluate the effect of agents on retroviruses in vitro.

In intial experiments, chelator was used in place of detergent in the preincubation step before RT assay, 20 ul of a solution of either EDTA, EGTA (2 mmol/l) or Nonidet P40 (NP 40, 0.1 pet cent) in 0.01 mol/l Tris-HCl buffer, pH 8.3, were added to 20 ul virus suspension (about 4 ug of protein) containing 0.1 mol/l dithiothreitol. Incubation was performed at 0C for 30 minutes. The RT reaction mixture (10 ul) was then completed and the assay was done as described (40) using poly rA.oligo dT (Boehringer) as a primer-temlate. However, the concentration of divalent cations was raised in equivalence to the amount of previously added chelator.

In the majority of experiments, however, every virus preparation was subdivided into four aliquots assayed each for RT activity after treatment as follows:

(a) virus not treated with agent and incubated without detergent

(b) virus not treated with agent and incubated with NP 40.

(d) virus treated with agent as indicated and incubated with NP 40.

Treatment of virus with agents in aliquots (c) and (d) was carried out by resuspending high sped viral pellets thoroughly (approximately 4 ug of protein/ml) in 5 ml of 0.01 mol/l Tris-HCl buffer, pH 8.3, containing the agents as indicated, followed immediately by pelleting at 50,000 x g for 60 minutes at 4 C. Control aliquots (a) and (b) were handled in the same way under omission of agents. The final pellets were resuspended in 100 ul Tris buffer, incubated at 0 C for 30 minutes with or without 0.05 per cent NP 40, and assayed for RT activity as described (40) using either Mn++ (0.3 mmol/l for RLV and SSV) or Mg++ (9.6 mmol/l, all other viruses) as a divalent cation. Reaction mixtures were incubated at 37 C except those of BLV and RLV having their optima at 25 and 30 C, respectively. The lytic activity (LA) of a given agent was defined as the difference of the amount of undisrupted virus present before and after treatment with agent referred to the initial amount of undisrupted virus. LA was calculated by the means of the formula:

LA= 1- (d-c) x 100

b-a

where a,b,c and d refer to the activities of RT (dis/min incorporated into acid insoluble material) in the above mentioned aliquots. Incorporation of radioactivity in controls (b) was in the range of 50,000 to 200,000 dis/min. Spontaneous release of RT activity in controls (a) was variable with different virus preparations but did not exceed a level of 10 - 15 per cent of controls (b), otherwise the virus was eliminated from consideration.

Proteins were determined by a micromethod based on staining with bromophenol blue (31).

Infectivity assays with RLV

2.0 ml of a 1:50 in 0.15 mol/l NaCl (w/v) diluted cell free extract of spleens from RLV infected leukemic NMRI mice were incubated with 2.0 ml of a solution of EDTA or EGTA (2 mmol/l) in 0.15 mol/l NaCl or, as a control, slaine alone at 0 C for one hour. Male NMRI mice of our inbred colony (2 - 3 months old, body weight of about 20 g) were then innoculated i.p. with each 0.1 ml of extract treated or not with one of the chelating agents. Animals were kept under conventional conditions. On day 13, half of mice were killed and mean spleen weights in the treated and control groups, consisting each of 10 animals, were analyzed. At different times after injection activity of particle associated RT was determined in pooled bloods obtained by orbital bleeding from each 5 mice of treated or control groups, respectively. The remaining mice were kept until death to ascertain mean survival time.

Results

Disintegration of Retroviruses by EDTA and EGTA as measured in the RT release Assay

Table 1 shows the result of a representative experiment performed to directly explore the action of chelating agents on retroviral membranes. When adding during the preincubation preceding the RT assay, EDTA and EGTA obviously disassociated some membrane components as indicated by the appearance of RT activity in otherwise undestructed virus samples. This effect however, was sometimes hard to reproduce possible owing to some irreversible chelator induced activation of RT, a zinc metallozyme not reactivable by Mg++ or Mn++ (25), even although either ion was additionally added to the RT reaction mixture in amounts equimolar to the chelator.

Efficacy of EDTA and EGTA in Disintegrating Various Retroviruses

To circumvent the problems just described, most experiments were performed in a fashion allowing a more favorable ratio of chelator to virus during exposure as well as the subsequent elimination of major amounts of chelating agents before assaying for RT activity, i.e., sedimentation of viruses through a solution containing the agent under study. Results are expressed as lytic activity LA as defined in Materials and Methods. This index reflects the ability of a given agent to bring about a partial or complete disintegration of virus particles. Partially disintegrated particles, although to a variable extent damaged by exposure to chelator, remain sedimentable but exhibit RT activity as seein in aliquots (c) treated with agent but not with detergent. On the other hand, the proportion of virus particles completely disassembled during exposure may be obtained by the difference of RT activities appearing in aliquots (b) and (d) treated without or with agent followed by disruption of the remaining virus with detergent. To give an example with RLV exposed in aliquots (c) and (d) to 1 mmol/l ETA: the four viral aliquots displayed RT activities of (a) 1.1, (b) 38.0, (c) 10.4, (d) 32.7 dis/min (each x 10 to the third), respectively, yielding an LA value of 40. It appeared, however, not meaningful to discriminate between partial and complete disintegration and the LA value included by definition both events was therefore used.

Fig. 1 (not shown) depicts the action of EDTA and EGTA on various retroviruses. These include mammalian type C viruses (RLV, SSV), primate type D viruses (MPMV, SMRV, PMFV) and BLV. Both chelating agents exert in millimolar concentrations a pronounced lytic activity upon most of the viruses tested so far. Mammalian type C viruses seem to be somewhat less susceptible than the other viruses. The avian type C virus AMV, however, was exceptional in that it did respond to neither EDTA or EGTA. The reason for that was not investigated but may be related to some peculiarities of the envelopes of the avian type C viruses as compared to their mammalian counterparts (23). Since RLV just as AMV was obtained from the plasma of leukemic animals and there was no difference in the susceptibility of RLV irrespective of originating from infected animals or tissue cultures, it appears to be unlikely that unresponsiveness of AMV is due to certain conditions of extracellular viral maturation.

Concentration Dependence of Disintegrating Activity of EDTA and EGTA

In initial experiments it was noted that retroviruses respond to chelating agents in a dose dependant manner. Fig. 2 (not shown) illustrates that PMFV, studied in greater detail, is susceptible to chelators over a wide range of concentrations. Complete disintegration, however, was reached with neither EDTA nor EGTA even in high concentrations up to 50 mmol/l. This may indicate that only a certain fraction of particles present in a given virus population is sensitive to these agent possible owing to variations in age, stage of maturity, or other factors. Such variables are known to influence the response of retroviruses to detergents (43).

Which Ion Is Involved in Disintegration

A major question arising from the experiments described so far is the nature of the cation complexed by the chelating agents and probably somehow responsible for the integrity of viral particles. Effectiveness of EGTA with its high binding affinity for Ca++ (stability constant log K=10.9) and relatively low affinity for Mg++ (log K=5.9) (3) clearly supports a role of calcium in virus integrity. However, the ability of EGTA to produce disintegration even in low concentrations may suggest that magnesium is also involved in maintaining the integrity of retroviruses, since the EDTA has binding affinities to both Ca++ (log K = 10.7) and Mg++ (log K = 8.9). The low amount of virus available for analysis did not yet allow an identification of the respective cation(s) by means of chemical or physicochemical methods.

The possibility that EDTA and EGTA chelate different cations in retroviral membranes led us to examine a possible synergistic action of both agents on retroviruses. Thus far, however, no increase in LA values has been observed after exposure of viruses to equimolar mixtures of both chelators.

To further substantiate the involvement of one of the ions under consideration, experiments were performed to ascertain whether the addition of divalent cations could prevent the action of EDTA or EGTA (not shown). As already reported for PMFV (32) and now confirmed with other viruses, both MgCl2 and CaCl2 prevented disintegration of retroviruses when simultaneously added with the chelating agent. However, addition of these cations at a later time did not cancel the effect produced by EDTA or EGTA. These results corroborate the assumption that both Ca++ and Mg++ ions are associated with retroviruses. Moreover, they exclude the possibility of involvement of such heavy metal ions binding more strongly to EGTA than Ca++, because in that case addition of CaCL2 or MgCl2 would not have prevented disintegration.

Effect of EDTA and EGTA on Infectivity of RLV

That chelating agents indeed adversely affect retroviral membranes was independently demonstrated in another set of experiments. RLV, contained in cell-free spleen homogenates, was incubated in vitro with EDTA or EGTA and then injected into mice for analysis of leukemogenic capacity. As compared to saline incubated virus in the controls, this treatment led to a marked inhibition of splenomegaly as well as to a doubling of mean survival time of infected animals (Table 2 not shown). Therefore, disintegration of virus particles as biochemically measured is paralleled by an equivalent loss of infectivity. The remaining virus, however, was still able to cause leukemia being diagnosed (courtesy of Prof. F. Fey) at the time of death of animals. Nevertheless, antiviral activity of chelating agents was also reflected in the pronounced depression ofparticle bound RT activity in the plasma of mice inoculated with treated virus. During the course of development of leukemias, there was a considerable delay in reaching comparable levels of RT activity in blood of treated as against control mice.

Susceptibility of BLV to Different Beta Blockers

The hypothesis that Ca++ may be associated with retroviruses was further tested in animals with beta receptor blocking drugs. Propranolol, a nonselective beta blocker used in medical practice, has been shown to be able to interact with membranes under concomitant Ca++ displacemant (2). It also disrupts membranes of type C and type D retroviruses (41). Contrary to propranolol, some of its congeners cardioselective beta blockers like practolol, sotalol and atenolol lacking the hydrophobicity of propranolol, do not essentially influence membrane phospholipids and membrane boud Ca++ (2,17,24). For that reason it appeared of interest to examine the susceptibility of a retroviruses to these drugs. BLV was chosen because it was not included in our earlier studies. Table 3 (not shown) shows that exposure to propranolol in vitro clearly disintegrates BLV, as previously demonstrated with other retroviruses. Exposure to one of the propranolol congeners however produces only a small, if any, disintegration of BLV. These results may thus provide additional evidence for calcium as the ion to take into consideration.

Lytic Action of Trifluoperazine on Retroviruses

The effect of trifluoperazine (TFP) on retroviruses ws investigated because (i) other phenothiazines have been found to exert a lytic effect on retroviruses in vitro (41), (ii) drugs of this type are known to produce, along with a variety of other effects on biological membranes, a displacement of Ca++ from membrane components (29) and (iii) TFP is able to bind in a Ca++ dependant manner (18,19) highly effectively to calmodulin, a widespread regulatory protein (16), and is therefore widely considered as a specific probe for calmodulin. In view of these properties disintegrating activity of TFP on retroviruses could serve as an indicator both for the presence of Ca++ and for the identification of a putative Ca++ binding molecule in retroviral membranes. In fact, following exposure to TFP retroviruses of type C or type D as well as BLV displayed a release of RTactivity (table 4) [not shown]. The effect was again dose dependant. To favor a Ca++ specific binding of TFP to viral components, exposure was performed at pH 7.2 and not at pH 8.3 as usually, because pH values above 7.5 diminish the specificity of binding (36) . Despite some variations with different viruses, TFP showed a similar disintegrating activity as various phenothiazines (41). Although AMV failed to respond to EDTA and EGTA, this virus was found to be susceptible to TFP, similarly as to other phenothiazines (41), too.

From the results of this set of experiments it is tempting to conclude that retroviral membranes donot only contain Ca++ but possible also a Ca++ binding protein (7) which may be calmodulin like in sharing with this protein the property of responding to TFP.

Discussion

In the present work designed to evaluate the action of the action of chelation agents on retroviruses in vitro we have mainly used a procedure sonsisting of centrifugation of intact virus particles through a solution of each respective agent and subsequent assay for RT to detect disintegration of the viral membrane. The results show that these agents are capable of disintegrating all of the mammalian retroviruses tested. This finding was supported by experiments showing decreased infectivity of RLV after the virus had been treated with chelating agents.

Principally, disintegration of retroviruses in vitro may occur either spontaneously or by treatment with membrane active agents. Spontaneous disintegration thought to be due simply to aging of particles is variable but usually low. On the other hand, a variety of agents including detergents (21) , lipid solvents (21), and neurotropic drugs (41) as well as proteins such as human (37) or nonhuman primate complement (30) and melittin (12) cause a complete or nearly complete lysis of retroviruses. Complement mediated virolysis affects the P15(E) protein known to be embedded directly into the retroviral membrane (1). Other interesting agents are the polyene antibiotics and membrane channel formers filipin (22) and nystatin (12) which induce some alterations in but nor disintegration of the membranes of retroviruses while retaining their infectivity. This study revealed still another type of response with some dissociaton of membrane components that affects the infectivity of virus but does not necessarily result in complete destruction of viral particles because a variable proportion of them remains sedimentable even after treatment with chelators. Thus, retroviruses may exhibit a broad range of response to exogenous factors. However, among such factors, chelators like EGTA are rather exceptional in that they do not represent true membrane active agents, although several observations point to a role of Ca++ in membrane stability. (6).

There is increasing evidence for some association of Ca++ with viruses and a role of this ion in maintaining viral structure. The observation of calcium binding sites is nearly 20 plant viruses including tobacco mosaic virus, and also bioenergetic considerations, led DURHAM (10,11) to propose that Ca++ might generally control disassembly of such viruses. Other authors succeeded in disassociated of polyoma virus by chelation of Ca++ (4) and reassembly of infectious viral particles by subsequent addition of Ca++ (5). Exposure of rotavirus particles to calcium chelators resulted in an unmasking of internal RNA polymerase activity (9). On the basis of these findings it is tempting to assume that association with Ca++ is a widespread property among viruses of different families.

The viral components associated with Ca++ have not been identified. One possibility is that the attachment of Ca++ to phosphatidylserine (13) known to occur in the so far analyzed retroviral membranes (28). Distinctive features of model membranes have been attributed to interactions of Ca++ with membrane phospholipids and there is evidence that the viral membranes behave in this respect similarly as model membranes, e.g., in virus induced cell fusion processes (27). On the other hand, certain proteins could serve as receptors for Ca++ owing to their ability to bind Ca++ in a selective and reversible fashion. The prototype of such Ca++ binding proteins is calmodulin, which, upon binding of Ca++, undergoes a confirmational change needed for its biological activity (16). In the presence of Ca++, calmodulin avidly binds phenothiazines (with TFP being the most effective one) and becomes thereafter biologically inactive (36). Propranolol is another antagonist of calmodulin (35). Consequently the strong retrovirus disintegrating activity of TFP, other phenothiazines, and propranolol (41) may be considered as preliminary evidence for the occurrence of calmodulin like proteins in retroviral membranes, though it remains to be identified. During complex formation of Ca++ both with phospholipids (26) and proteins (39) there is some synergism with Mg++ and it is, therefore, well conceivable that both cations simultaneously occur in rteroviral membranes.

Whatever the mode of Ca++ binding to viral components is, and irrespective of whether Ca++ is accidentally or even specifically complexed to them, the occurance of Ca++ in retroviral membranes may have biological implication with regard to the assembly and disassemble of viral particles in and their budding from infected cells. Generally Ca++ has been found to influence a wide variety of functional properties of biological membranes.

Finally, identification of Ca++ binding viral components may eventually prove useful in the search for new and effective retroviral agents. The presently limited success of virus chemotherapy with chelating agents (20) might be generally augmented by considering such components as drug targets, too. Our recent demonstration that haloperidol, a colmodulin binding butyrophenone (19), exerts an antiviral effect on Raucher murine leukemia virus in vivo (42), may support the feasibility of this approach.