Brain Research, 553 (1991) 300-304 (~) 1991 Elsevier Science Publishers B.V. 0006-8993/91/$03.50 ADONIS 000689939124728.1

300

BRES 24728

Short Communications

Brain-derived cells contain a specific binding site for Gpl the CD4 antigen M.R. Kozlowski 1, P. Sandier 2, P.-E

which is not

Lin 2 a n d A . W a t s o n 3

Departments of l Screening and Biochemical Research and 2Virology, Bristol-Myers Squibb Research Institute, Wallingford, CT 06492-7660 (U.S.A.) and 30ncogen, 3005 First Avenue, Seattle, WA 98121 (U.S.A)

(Accepted 2 April 1991) Key words: Human immunodeficiencyvirus (HIV-1); gpl20; CD4; Neuroblastoma; Glioblastoma

Infection with the human immunodeficiency virus (HIV-1) often produces a set of neuropsychiatric dysfunctions which have been termed the AIDS dementia complex. This complex appears due to the infection of brain cells by HIV-1. If so, brain cells might be expected to contain

a binding site for the same viral envelope glycoprotein that enables HIV-1 to bind to other cells (e.g. CD4 + T-cells), gpl20. The present study shows that the cells of the brain-derived U-138MG, U-373MG, SK-N-MC and SK-N-SH cell lines bind gpl20 in an inhibitable fashion. Binding of gpl20 to these cells is inhibited by the dyes Aurintricarboxylic acid (ATA) and Evans blue (EB), which are known to inhibit specific gpl20 and HIV-1 binding, and block HIV-1 infection, in CD4-expressing cells. Binding is not inhibited by Aurin, a dye related to ATA but lacking its anti-HIV effects. As expected, anti-CD4 antibodies are ineffective in blocking gpl20 binding to brain-derived cells. These results suggest that human brain-derived cells possess a specific binding site for gpl20 that is not the CD4 antigen. Infection by the human immunodeficiency virus (HIV1) produces, in addition to immune system dysfunction (e.g. acquired immune deficiency syndrome; AIDS), a set of neuropsychiatric disorders which are called the AIDS dementia complex (ADC) 8'19. The symptoms of A D C include cognitive impairment, apathy, and motor dysfunctions 19. A D C may affect as many as 90% of the victims of A I D S or AIDS-related complex x9. Although the underlying cause of A D C appears to be the death of brain cells, the reason for the cell loss has not been convincingly demonstrated. HIV-1 can be isolated from the brains of infected individuals 8'19, but has generally been thought to be present in monocytes, macrophages, or epithelial cells rather than in brain cells (i.e. neurons and glia) 8'19. The abnormal functioning of the infected cells would then lead to damage to brain cells 8'19. More recently, the demonstration of susceptibility to HIV-1 infection in several brain-derived cell lines and primary brain cell cultures in vitro has offered a more direct mechanism for the production of ADC 3-7'12"22. It suggests that the virus infects brain cells directly, leading to the loss of these cells. In a well-studied system of HIV-1 infection, infection of human CD4 ÷ T lymphocytes, one of the initial steps is binding of the viral envelope glycoprotein, gp120, to the CD4 cell surface antigen 8' 19.20. If a similar binding process occurs in the infection

of brain cells, the attachment site on the virus may also be gp120 since HIV-1 infection can be blocked by anti-gp120 antibodies 5'12. An early study suggested that the cellular attachment site in brain might also be the CD4 antigen TM, but an attempt to replicate these findings was not successful 9. It now appears unlikely that the CD4 antigen is involved in the infection of brain-derived cells by H I V since susceptibility of brain-derived cells to infection does not appear to be correlated with the level of expression of the CD4 antigen 3'4'6'7'22, and infection of brain-derived cells by HIV-1 is not blocked by anti-CD4 antibodies 5'12. In the present study, a radioligand gp120 binding assay is used to search for a gp120 binding site in brain-derived cells. Brain-derived and immune system-derived cells were obtained from ATCC, and cultured in roller bottles in either Eagle's minimum essential medium (MEM) supplemented with 90% Earle's salts and 10% fetal bovine serum (U-138MG, U-373MG), Delbecco's modified Eagles medium (DME) supplemented with 10% fetal bovine serum (SK-N-MC, SK-N-SH), or RPMI 1640 supplemented with 10% fetal bovine serum (U-937, CEM). To prepare the radioligand, BSC 40 cells were infected with a recombinant vaccinia virus construct containing the sequence for HIV-1 (V-gp120). With this construct the bulk of the gp120 is secreted and can be r e c o v e r e d

Correspondence: M.R. Kozlowski, Department of Screening and Biochemical Research, Bristol-Myers Squibb Research Institute, P.O. Box

1500, Wallingford, CT 06492-7660, U.S.A.

301 from the medium by binding to a lentil-lectin column. No detergents were used in the purification process. The eluted glycoprotein (3.5/zg) was labeled with 300/zCi di-iodinated BoRon-Hunter reagent (NEN, 4400 Ci/ mmol.). A molar ratio of 3.4:1 (Bolton-Hunter reagent to glycoprotein) was used. Labeling was conducted for 2 h at room temperature in PBS at pH 8.0. Unincorporated Bolton-Hunter reagent was removed by separation on a PD10 column (Pharmacia) equilibrated with PBS, 1 mg/ml gelatin, and 0.02% sodium azide. The labeled gpl20 was >40% bindable. The labeled ligand was stored at 4 °C and used within one week. On the day of the assay the cells were harvested by low speed centrifugation (1000 rpm) and resuspended in the binding buffer: phosphate-buffered saline (PBS; Sigma), 1 g/1 bovine serum albumin (BSA; Sigma), 1 g/1 glucose (Sigma) (PBG), at a concentration of 2.5 x 107 cells/ml. Antibodies or dyes were added to the cells and the mixture preincubated for 30 rain at room temperature. An equal volume of binding buffer containing the radioligand was added for a final ligand concentration of 0.2 nM. The mixture was incubated for 2 h at 37 °C, a period sufficient to achieve maximum binding. The cells were then pelleted by centrifugation (2000 rpm). In some cases the supernatant was taken for examination of unbound ligand. The cells were then washed twice with ice-cold PBG by briefly resuspending them in fresh buffer and then re-pelleting them. In some cases the pellet was left in the second wash for 3 h at 37 °C in order to allow the bound ligand to dissociate so that it could be analyzed. The final pellet was filtered through a glass fiber filtermat pretreated with BSA, and the filtermat was washed twice with PBS containing I g/1 BSA. The interval between the termination of the incubation and the filtration of the cells was less than 10 min, except as noted above. Finally, the filtermat was dried and counted using a Betaplate (LKB) scintillation counter. Total binding (DPM/106 cell, mean + S.E.M., n = 5) was 1915 + 194 for CEM cells, 886 + 208 for U-937 cells, 775 + 91 for U-373MG cells, 948 + 111 for U-138MG cells, 988 + 42 for SK-N-SH cells, and 1271 _+ 218 for SK-N-MC cells. OKT4 and OKT4A were purchased from Ortho Diagnostic Systems Inc., and G17-2 was supplied by J. Ledbetter, Oncogen Corp., Seattle, WA. The dyes were purchased from the Aldrich Chemical Company, Inc. Radio-iodinated recombinant gpl20 (125I-rgp120) binding to human glioblastoma cells (U-138MG and U-373MG) 3-7'12'22, human neuroblastoma cells (SK-NMC and SK-N-SH 3-7,12'22, and human immune systemderived cells (CEM lymphoblastoid cells and U-937 monocyte-like cells) 1~-16 was compared. The anti-CD4 antibodies OKT4a and G17-2, which bind to the same epitope of CD4 as gpl201°,16, effectively inhibited gp120

TABLE I Inhibition oflZSl-rgpl20 binding by anti-CD4 antibodies

Values are mean and standard error of the inhibition of total binding. Number of replicate experiments is shownin parentheses. Cell line

% Inhibition

U-138MG U-373MG SK-N-MC SK-N-SH CEM U-937

G17-2

OKT4A

OKT4

8 _+7 (6) 13__.10(5) 3 + 4 (3) 10 + 16 (3) 73+3 (3) 56+5 (3)

13 _+8 (5) 8 _+15 (6) -7_+18(5) -11+9 (4) -6 + 7 (3) -7 + 3 (3) -17 + 11 (3) 9 + 17 (3) 79+1 (3) - 1 + 8 (3) 46_+4 (3) -14+16(3)

binding to CEM and U-937 cells as previously described la-16 but failed to inhibit binding to SK-N-MC, SK-N-SH, U-138MG, or U-373MG cells (Table I). The failure of anti-CD4 antibodies to inhibit gp120 binding to the brain-derived cells parallels the previously described inability of these antibodies to block HIV-1 infection in U-373MG and SK-N-MC cells5'12, and further demonstrates that these brain-derived cells do not express the CD4 antigen in a form capable of binding gp120. Besides anti-CD4a antigens, other agents known to inhibit gp120 of HIV-1 binding at the CD4a antigen and prevent infection of cells include the dyes ATA and EB 1'2'21. In CEM cells, these dyes have a direct effect on the cellular gp120 binding site 1°'13. If a receptor for gp120 is present in brain it might be expected to have a gpl20 binding site similar to that of the CD4 antigen and, therefore, also to be inhibited by these dyes. To test this hypothesis, the effects of ATA and EB on gp120 binding to the brain-derived cells were examined. Surprisingly, a large, reproducible inhibition of 125I-rgp120 binding to brain-derived cells was produced by both EB and ATA (Table II). Based on the plateaus of the inhibition curves for the dyes (Fig. 1), the active dyes inhibited binding to the same extent within a cell line. This is most dearly TABLE II Inhibition of ZZal-rgpl20 binding by dyes (IO01~Mconcentrations)

Values are as in Table I. n indicates number of replicate experiments. Cell line

U-138MG U-373MG SK-N-MC SK-N-SH CEM U-937

n

4 3 3 3 4 4

% Inhibition ATA

Aurin

664-7 37_+6 31_4 36_+5 62+8 60_+9

11_+12 50_+9 23_+9 -4 4-6 35_+4 20-+7 - 6 _ 7 37__2 20_4 -2+6

Evans blue

42+4

15+5 62+5 18_+3 57+3

Acid red 97

22+10

18+5 19+3

Acid alizarin violet N

1__.1 5_+3 11 4-3 6+16

-5+7 -18+8

302 seen in the U-373MG cells, in which inhibition plateaus were obtained for each active dye (Fig. 1, top). Furthermore, in the lymphocytic cells the portion of dyeinhibitable gpl20 binding was the same as that inhibited by the anti-CD4a antibodies. The relative and absolute potencies with which dyes inhibited gpl20 binding in brain-derived cells and immune system-derived cells was similar. Thus, ATA and EB had similar low IC5o value in both cell types, Acid red 97 had similar higher IC5o values, and Acid alizarin violet N failed to inhibit gpl20 binding to either type of cell (Fig. 1, Tables II and III). There may, however, have been some subtle differences between this site and the CD4 antigen since the order of potency of ATA and EB was reversed in U-373MG and SK-N-MC cells relative to CEM cells, and Acid red 97 appeared more potent in the brain-derived cells than in the lymphocyte cells. The 120 110 100 90 80 70 60

1.0 E - 6

1.0 E - 5 Concentration

1.0 E-4 (M)

1.0 E-3

105 95 85 75 65 55 45 35

!

1.0 E- 6

I

1.0 E-5 Concentration

binding site was not a non-selective peptide recognition site since bradykinin, epidermal growth factor, n e r v e growth factor, and vasoactive intestinal peptide did not inhibit binding at concentrations of 1/~M. In order to rule out artifactual effects of these dyes that might resemble gp120 binding inhibition, two additional experiments were performed. The first experiment was done to show that the dyes did not reduce binding by lowering the effective concentration of the ligand, t25Irgp120. This might occur, for example, if the dyes bound to the ligand in such a way as to make it unable to associate with the cells. Such a mechanism can be distinguished from the inhibition of binding at a specific site because it would decrease not only inhibitable gp120 binding, but also non-inhibitable (non-specific) binding, which is also concentration-dependent. Although this mechanism seems unlikely because the dyes never produced a complete inhibition of binding (Fig. 1), the effect of the dyes on non-specific gp120 binding was examined directly. The maximum reduction in inhibitable gp120 binding of CEM cells (68%) was determined using anti-CD4a antibody, G17-2. Addition of either ATA (32 ~M) or EB (32 ArM) to the G17-2 (1 experiment) produced amounts of inhibition (70% and 67%, respectively) essentially equal to that produced by G17-2 alone. Therefore, since these dyes do not reduce non-specific gp120 binding, they are not acting by reducing the effective ligand concentration. A second possible mechanism of action for the dyes unrelated to binding site inhibition could be reduction of the viability of the cells. No toxicity to the cells was seen during the course of the assay using Trypan blue dye exclusion 17 as a measure of viability (1 experiment). To confirm that the ligand remained intact throughout the experiment, and that 12SI-rgp120 was the species binding to the cells, both unbound and bound ligand (obtained as described above) were examined by SDS-

I

1.0 E - 4 M)

10E-3

TABLE III 110

1(7.5ovalues for the inhibition of J2Sl-rgpl20 binding by dyes

100

Number of experiments used to generate the values is shown in parentheses. ICso values for U-373MG and SK-N-MC cells were defined as one half of the maximal inhibition produced by the dye, determined by extrapolating the inhibition curves. These values were 40% of total binding for U-373MG cells and 50% for SK-N-MC cells. For CEM cells maximal inhibition was taken as that produced by G17-2 (73%).

90 80 70

5O 3C

20'

I 1.0E-6

1 1.0E-5

1.0

~_ -4

10E-3

Fig. 1. Inhibition of 1251-rgp120 bindin 8 to U-373MG (top panel), SN-N-MC (center panel), and CEM (bottom panel) cells by ATA (filled triangle), EB (filled square), and Acid red 97 (filled circle). Results are the mean of 3-4 experiments. Standard errors are represented by the error bars.

Dye

ATA Evansblue Acid red 97

lCso (~M) U-373MG

SK-N-MC

CEM

20 -+8 (3) 16+8 (3) 93 -+22 (3)

34 + 4 (3) 23-+5(3) 133 + 7 (3) 1

28 + 7 (4) 46-+4 (4) 200 + 60 (3)

303

Fig. 2. SDS-PAGE analysis of 12SI-rgp120. Aliquots of l~I-rgpl20 either released after binding to CEM (first lane) and U-138MG (second lane), or incubated with U-138MG (fourth lane) and CEM (fifth lane) cells but not bound are compared to unincubated ligand (sixth lane). 14C-labeledprotein molecular weight markers (Amersham) are shown in lane 3 with bands (from top) corresponding to molecular weights of 200000, 97400, 69000, 46000 and 30000. Protein bands on the 8% SDS-PAGE gel were visualized autoradiographieally using X-OMAT RP film (Kodak) with Cronex intensifying screens (Dupont). The figure is a composite of two exposures of the gel, one for two weeks (lanes 1-3) and one for 1 day (lanes 4--6).

P A G E analysis at the end of the incubation period (Fig. 2). Prior to use in the assay, the majority of the radioactivity in the ligand migrated as a single broad band with a molecular weight of approximately 120000, as expected for authentic rgp120 ~°. A much smaller amount of the radiolabel was present in a band at 50000 MW, and very faint bands at 83000, 64000 and 35000 MW could also be detected. The profile of the unbound ligand at the end of the incubation period was identical to that of the unincubated ligand. This showed that the ligand was not degraded during the incubation. Likewise, the profile of the bound ligand following release was essentially the same as that of the unincubated ligand (Fig. 2). A minor difference between unincubated ligand and that released from binding to the CEM cells was that the latter contained additional, low molecular weight bands. Conversely, ligand released from binding to U-138MG cells lacked the band seen at 35000 MW with unincubated ligand. In both cases, however, the majority of the radiolabeled material migrated at 120000 MW. This demonstrated that the ligand was not significantly altered

by the binding, and that the bulk of the bound ligand was rgp120. Several features of the inhibition of 125I-rgp120 binding to brain-derived cells by dyes suggest that the inhibitable binding occurs at a set number of distinct sites with particular binding requirements; hallmarks of specific binding. First, and most importantly, the binding is inhibitable, which is not the case with non-specific associations. Second, the amount of inhibitable binding is the same regardless of the inhibitor, suggesting a fixed number of sites. Third, dyes structurally related to those that inhibit binding, as well as other peptides, are inactive. Thus, there are specific structural requirements for binding. Finally, the inhibition of gpl20 binding by dyes in brain-derived cells parallels that in the immune system-derived cells, which occurs at an identified set of specific binding sites, the CD4 antigen. The characteristics of gpl20 binding inhibition in the brain-derived cells examined are consistent with it being involved in HIV-1 infection. Thus, binding is inhibited by dyes known to prevent HIV-1 infection in lymphocytic cells (ATA and EB), but not by a related dye lacking this effect (Aurin), and the concentrations of the dyes neeessary to inhibit binding are similar to those needed to prevent HIV-1 infection 1"z'21. Furthermore, the binding is not inhibited by anti-CD4a antibodies, which do not block HIV-1 infection in brain-derived cells5,12. If this binding site is involved in HIV infection, however, it is surprising that SK-N-SH cells, the only line used in this study that has not been shown to be infectable by HIV-13-7'12"22, also possesses gpl20 binding sites. There are at least two possible explanations for this apparent anomaly. First, since the non-infectable status of these cells has not been established using a highly sensitive system such as co-cultivation with susceptible lymphocytic cells 3'4'6'7"22, it is possible that these cells are merely less easily infected. Second, these cells might be able to bind gp120 but might still be uninfectable because they do not allow viral fusion or replication. Since non-CD4 gp120 binding sites were present in all of the human brain-derived cell lines examined, it seems reasonable to suppose that similar sites exist in normal brain tissue. The presence of such sites has important implications for the development of therapies for treating HIV infection by inhibiting viral binding. It implies, for example, that therapies aimed solely at blocking HIV binding to the CD4 antigen will not be successful at preventing viral infection in all tissues. In particular, although CD4 blockade might prevent the development of AIDS by blocking the infection or destruction of lymphocytes, it may not halt ADC since it does not protect brain cells. A more fruitful strategy would be the development of gp120 binding inhibitors that are not

304

tissue-selective. The dye molecules active in inhibiting gpl20 binding in both lymphocytic and brain-derived cells may provide valuable templates for the development of drugs with this activity.

The authors acknowledge the assistance of M. Rosser and A. Longden in culturing the lymphocytic and neuroblastoma cells used in this study, and of Dr. S.P. Manly in the running of gels. The helpful comments of Drs. S. O'Connor, R. Datema, M. Gorman and D. Taylor were also appreciated.

1 Balzarini, J., Mitsuya, H., DeClerq, E. and Broder, S., Aurintricarboxylic acid and Evans blue represent two different classes of anionic compounds which selectively inhibit the cytopathogenicity of human T-cell lymphotropic virus type III/lymphadenopathy-associated virus, Biochem. Biophys. Res. Commun., 136 (1986) 64-71. 2 Balzarini, J., Mitsuya, H., DeClerq, E. and Broder, S., Comparative inhibitory effects of suramin and other selected compound on the infectivity and replication of human T-cell lymphotropic virus (HTLV-III)/lymphadenopathy-associated virus (LAV), Int. J. Cancer, 37 (1986) 451-457.

12 Li, X.L., Moudgil, T., Vinters, H.V. and Ho, D.D., CD4independent, productive infection of a neuronal cell line by human immunodeficiency virus type 1, J. Virol., 64 (1990) 1383-1387. 13 Linsley, ES., Ledbetter, J.A., Kinney-Thomas, E. and Hu, S.-L., Effects of anti-gp120 monoclonal antibodies on CD4 receptor binding by the env protein of human immunodeficiency virus type 1, J. Virol., 62 (1988) 3695-3702. 14 Ludlin, K., Nygren, A., Arthor, L.O., Robey, W.G., Morein, B., Ramstedt, U., Gidland, M. and Wigzell, H., A specific assay measuring binding of 125I-gp120from HIV-1 to T4÷/CD4 + cells, J. lmmunol. Methods, 97 (1987) 93-100. 15 Matthews, T.J., Weinhold, K.J., Lyerly, H.K., Langlois, A.J., Wigzell, H. and Bolognesi, D.P., Interaction between the human T-cell lymphotropic type IIIB envelope glycoprotein gpl20 and the surface antigen CD4: role of the carbohydrate in binding and cell fusion, Proc. Natl. Acad. Sci. U.S.A., 84 (1987) 5424-5428. 16 McDougal, J.S., Kenedy, M.S., Sligh, J.M., Port, S.P., Mawle, A. and Nicholson, J.K.A., Binding of HTLV-III/LAV to T4÷ T cells by a complex of the ll0K viral protein and the T4 molecule, Science, 23 (1986) 382-385. 17 Patterson Jr., M.K. In S.P. Colowick and N.O. Kaplan (Eds.), Methods in Enzymology, Vol. 63, Academic Press, New York, 1979, pp. 141-152. 18 Pert, C.B., Hill, J.M., Ruff, M.R., Berman, R.M., Robey, W.G., Arthur, L.O., Ruscetti, EW. and Farrar, W.L., Octapeptides deduced from the neuropeptide receptor-like pattern of antigen T4 in brain potently inhibit human immunodeficiency virus receptor binding and T-cell infectivity, Proc. Natl. Acad. Sci. U.S.A., 83 (1986) 9254-9258. 19 Price, R.W., Brew, B., Sidtis, J., Rosenblum, M., Scheck, A.C. and Cleary, P., The brain in AIDS: central nervous system HIV-I-1 infection and AIDS dementia complex, Science, 239 (1988) 586-592. 20 Sattentau, Q.S. and Weiss, R.A., The CD4 antigen: physiological ligand and HIV-1 receptor, Cell, 52 (1988) 631-633. 21 Schols, D., Baba, M., Pauwels, R., Desmyer, J. and DeClerq, E., Specific interaction of aurintricarboxylic acid with the human immunodeficiency virus/CD4 cell receptor, Proc. Natl. Acad. Sci. U.S.A., 86 (1989) 3322-3326. 22 Srinivasan, A., Dorsett, D., York, D., Bohan, C. and Anand, R., Human immunodeficiency virus replication in human brain cells, Arch. Virol., 98 (1988) 135-141.

3 Chang-Mayer,C., Rutka, J.T., Rosenblum, M.L., McHugh, T., Stites, D.P. and Levy, J.A., Human immunodeficiency virus can productively infect cultured human glial cells, Proc. Natl. Acad. Sci. U.S.A., 84 (1987) 3526-3530. 4 Chiodi, F., Fuerstenberg, S., Midlund, M., Asjo, B. and Fenyo, E.M., Infection of brain-derived cells with the human immunodeficiency virus, J. Virol., 6 (1987) 1244-1247. 5 Clapham, P.R., Weber, J.N., Whitby, D., Mclntosh, K., Dalgleish, A.G., Maddon, P.J., Deen, K.C., Sweet, R.W. and Weiss, R.A., Soluble CD4 blocks the infectivity of diverse strains of HIV-1 and SIV for T cells and monocytes but not for brain and muscle oells, Nature, 337 (1989) 368-370. 6 Dewhurst, S., Bresser, J., Stevenson, M., Sakai, K., EvingerHodges, J. and Volsky, D.J., Susceptibility of human glial cells to infection with human immunodeficiency virus (HIV-1), FEBS Lett., 213 (1987) 138-143. 7 Dewhurst, S., Stevenson, M. and Volsky, D.J., Expression of the T4 molecule (AIDS virus receptor) by human brain-derived cells, FEBS Lett., 213 (1987) 133-137. 8 Ho, D.D., Pomerantz, R.J. and Kaplan, J.C., Pathogenesis of infection with human immunodeficiency virus, N. Engl. J. Med., 317 (1987) 278-286. 9 Kozlowski, M.R., Hall, E. and Watson, A., Characterization of gpl20 binding in CEM cells and hippocampns, Neurosci. Abstr., 15 (1989) 671. 10 Kozlowski, M.R. and Watson, A., Characterization of gpl20 binding to the CD4 antigen and detection of specific inhibitors, Antiviral Chem. Chemother., 1 (1990) 175-182. 11 Landau, N.R., Warton, M. and Littman, D.R., The envelope glycoprotein of the human immunodeficiency virus binds to the immunogiobulin-like domain of CD4, Nature, 334 (1988) 159162.