Inhibition of Human HL-60 Cell Responses to Chemotactic Factors by Antisense Messenger RNA Depletion of G Proteins*

Chemotactic factors bound to receptors of the seven- transmembrane domain family signal leukocytes through associated guanine nucleotide-binding (G) pro- teins. Human leukocytes of the HL-60 line, which ex- press G protein-coupled receptors for leukotriene (LTB4) and (fMLP) after differentiation with vitamin Ds and trans- forming growth factor-p, were transfected with expression plasmids containing antisense-oriented cDNAs encoding the a-chains of Go, Gil, GB, and Gis. Antisense mRNA for Go and GB a-chains suppressed by over 80% the level of the respective G protein. Go-deficient HL-60 cells had depressed functional and intracellular calcium responses to LTB4 and fMLP, but no alterations in the responses of cyclic 3’,5’-monophosphate (CAMP). contrast, HL-60 cells deficient lost only responses of the intracellular concentration of CAMP. mRNA suppression of distinct G proteins thus may delineate some transductional requirements for cellular responses. 1:1,000 dilution of the carboxyl-terminal peptides 345-354 of the Gil and Gi2 a-subunits or the Gi3 and Go a-sub- units or with a 1:500 dilution of rabbit antibody to bovine brain Go a-subunit Bound antibodies were detected with a lumines-cence reagent system (Immun-Lite, Bio-Rad). (Amersham transfected HG60 quantified described (16), using GF/C glass fiber filters (Whatman) to resolve bound from unbound radioactivity. Chemotactic responses 2 x lo6 differentiated transfectantdml of HBSS with 0.05 dl00 ml ovalbumin and 25 m~ HEPES (pH = 7.3) in each chamber (Adaps, Dedham, MA) were quantified after 2 h at 37 "C. The 8-pm diameter pore filters were fixed and stained, and the number of HG60 cells per high power field above the background in filters from chambers without a stimulus were enumerated microscopically, as described (16,18, 19). To examine lyso- somal degranulation, suspensions of 4 x lo6 differentiated transfectantdml were incubated in HBSS with 0.1 dl00 ml ovalbumin, 5 pdml cytochalasin B (Sigma), and 10 m~ HEPES (pH = 7.4) for 30 min at 37 "C without and with LTB4 (Biomol), fMLP (Sigma), or ionophore A23187 (Calbiochem).After

general topological property of seven membrane-spanning domains and substantial homology of amino acid sequence (4). Current evidence suggests that more than one type of G protein associates with some receptors and may transmit distinct signals to different effectors, as well as additive or opposing signals to the same effector. The physical coupling of two or more different G proteins to one receptor was shown by immunoprecipitation of solubilized receptors with antibodies specific for G proteins (5,6) and by anti-G protein antibody-mediated uncoupling of receptor-G protein complexes (7,8). The functional interactions of one receptor with multiple G proteins have been demonstrated by introduction into intact cells of anti-G protein antibodies by patch-clamp pipettes (9) and of antisense oligonucleotides (10, 11), application of anti-G protein antibodies in cell membrane preparations (12,131, and specific competition by a novel peptide antagonist for G protein binding to recombinant receptors in phospholipid vesicles (14). Interpretation of the results of many past studies, however, have been complicated by the limitations of specificity and intracellular access of some immunochemical probes, only partial effects of antisense oligonucleotides and lack of concurrent assessment of functional and biochemical effects of alterations in G proteins. Transfection of human HL-60 cells with expression plasmids encoding full-length antisense (AS) messages, sufficient to suppress each G protein a-chain selectively, now is shown to identify their receptor-specific signals.
EXPERIMENTAL PROCEDURES Construction of Antisense cDNA Plasmids-Four full-length cDNAs encoding different G proteins were provided by Dr. R. R. Reed (Johns Hopkins University School of Medicine) (15). EcoRI-EcoRI fragments containing the full-length coding region and additional 5'-and 3'-untranslated sequences for the a-chains of Go (2.1 kb), Gil (2.0 kb), Gi2 (1.8 kb), and Gi3 (3.1 kb) were separately inserted into the EcoRI site of pcDNAI (Invitrogen) in reverse orientation to the direction of transcription of the cytomegalovirus promoter. The structure of each construct was verified by direct sequence analysis.
Culture, lhnsfection, and Differentiation of HL-60 Cells-HL-60 cells were cultured in complete RPMI medium, as described (161, washed and suspended at 1 x 107/ml in 5-10 ml of serum-free Opti-MEM (Life Technologies, Inc.), and transfected with antisense cDNA by incubation with 150-200 pg of plasmid DNA and 120 pl of DOTAP (Boehringer Mannheim) for 6 h at 37 "C in 5% C02:95% air. An equal volume of complete medium was added and the incubation continued for 18 h. The HL-60 cell transfectants then were induced to differentiate by washing and resuspension at 1 x 106/ml in complete medium containing 100 n M 1,25-dihydroxyvitamin D, and 1 ng/ml human purified recombinant transforming growth factor-p (TGF-P; Austral Biologicals), and further culture for 3-14 days. The density of HL-60 cells in culture increased at 4 days to means of 2.3, 2.5, and 2.7 x 106/ml, respectively, for sham (vector only), Go-antisense (AS), and Giz-AS transfectants, and at 8 days to 4.0, 4.0, and 4.3 x 106/ml (n = 3). The mean extent of differentiation, assessed by a-naphthyl esterase staining (16), was 16,17, and 24% at 4 days for sham, Go-AS, and Giz-AS transfectants, respectively, and 67, 64, and 71% at 8 days (n = 3). Thus neither AS construct altered cell growth or differentiation significantly.

Antisense Messenger RNA Depletion of G Proteins
The proteins resolved were transferred to nitrocellulose and developed with a 1:1,000 dilution of rabbit antibodies to the carboxyl-terminal peptides 345-354 of the Gil and Gi2 a-subunits or the Gi3 and Go a-subunits (Calbiochem, Inc., La Jolla, CA), or with a 1:500 dilution of rabbit antibody to bovine brain Go a-subunit (Dr. Su-Chen Tsai, National Institutes of Health). Bound antibodies were detected with a luminescence reagent system (Immun-Lite, Bio-Rad). Assessment of HL-60 Cell Binding and Effects of Leukotriene B, (LTB3 and N-Formyl-methionyl-leucyl-phenylalanine (fiULP)-The binding of [3H]LTB4 (Amersham Corp.) and C3H1fMLP (DuPont NEN) to differentiated control and transfected HG60 cells was quantified as described (16), using GF/C glass fiber filters (Whatman) to resolve bound from unbound radioactivity. Chemotactic responses of 2 x lo6 differentiated transfectantdml of HBSS with 0.05 dl00 ml ovalbumin and 25 m~ HEPES (pH = 7.3) in each chamber (Adaps, Dedham, MA) were quantified after 2 h a t 37 "C. The 8-pm diameter pore filters were fixed and stained, and the number of HG60 cells per high power field above the background in filters from chambers without a stimulus were enumerated microscopically, as described (16,18,19). To examine lysosomal degranulation, suspensions of 4 x lo6 differentiated transfectantdml were incubated in HBSS with 0.1 d l 0 0 ml ovalbumin, 5 pdml cytochalasin B (Sigma), and 10 m~ HEPES (pH = 7.4) for 30 min at 37 "C without and with LTB4 (Biomol), fMLP (Sigma), or ionophore A23187 (Calbiochem).After centrifugation at 2,000 x g for 5 min at 4 "C, supernatants were transferred to clean test tubes, buffer was added to the pellets, pellet suspensions were sonified at 200 watts for 2 min a t 4 "C, and P-glucuronidase was quantified in supernatants and pellet lysates (Sigma) as described (19) 3), and 100 p~ isobutyl methylxanthine and Ro 1724 (Calbiochem) were preincubated for 10 min at 37 "C, stimulated as indicated, and were incubated for 2 min a t 37 "C. The responses were stopped by addition of cold ethanol to 70%, and cyclic AMP in the 3,000 x g supernatants was quantified by radioimmunoassay after acetylation of cyclic AMP according to the manufacturer's protocol (DuPont NEN).

RESULTS AND DISCUSSION
Lipotransfection of human HL-60 cells with the high level expression vector pcDNA I, carrying cDNA that encodes fulllength AS messages for the a-chains of Gil, Gi2, Gi3, and Go, alone and in combination, suppressed the cellular level of the respective G proteins. The extent of suppression of G protein a-chains was quantified by SDS-PAGE and Western blot analysis of extracts of an equal number of each transfectant and of control HL-60 cells sham-transfected with vector alone (Fig. 1). The limit of detection of G protein a-chains by Western blot was approximately 5% of the total present in undiluted extracts of sham-transfected HL-60 cells. The two anti-carboxyl-terminal peptide G protein antibodies used detect common antigens in Gi2 and Gil and in Go and Gi3, respectively. Maximal suppression of the cellular content of Gi2 + Gil by Gi2 AS and Go + Gi3 by Go AS, respectively, was 80-95% and 80% at 6-7 days (Fig.   1). Western blot analyses of the course of suppression of Gi2 + Gil and Go + Gi3 protein a-chains showed less than half of the maximum at 3 4 days, no greater suppression at 10 days, and loss of more than half of the maximum effect a h r 14 days (n = 2). The time for attainment of the greatest suppression presumably reflects a requirement for catabolism of existing G proteins, after inhibition of synthesis by antisense mRNA.
Gil is not reflected in analyses of Gi2 (Fig. 1, panel A), as expected from the lack of detection of Gil protein (20) and of Gil assess the level of significance of differences in responses compared to mRNA in previous Northern blot analyses of HL-60 cells. In contrast, residual a-chain immunoreactivity apparently attributable to unsuppressed Go (Fig. 1, panel B ) may reflect the unaffected content of cross-reactive Gi3. Immunoquantification thus may underestimate the suppression of Go, which like Giz may be reduced to less than 20% of normal. Go has not been identified consistently in HL-60 cells differentiated without TGF-P. The HL-60 cell content of Go a-chain, as assessed with a monospecific antibody, was maximal after 6 days with TGF-P and dihydroxyvitamin D3 in sham transfectants, and was suppressed by a mean of 90% in Go-AS transfectants (n = 2). The selectivity of the AS effect was demonstrated by the lack of suppression of Go or Gi3 in Gi2-AS transfectants and of Gi2 in Go-AS transfectants.
Chemotactic responses of Go-deficient transfectants to LTE& and fMLP, each of which is recognized by a distinct set of high affinity G protein-coupled receptors (18, 19, 21, 221, were sig-  1 p~ LTB,, 0.1 and 1 p~ fMLP, and 1 pA23187, respectively, were 9.9   f 2.1,10.4 2 2.6,8.8 f 3.8, 13.6 2 3.3 and 10.7 * 2.6, above a background of release of 5.8-10.2% in buffer alone. A two-sample t test was used to assess the significance of the decreases in p-glucuronidase release, compared to that by sham transfectants. *, p < 0.01; +, p < 0.05; AS, antisense. nificantly less than those of the sham-and Giz-deficient transfectants (Fig. 2). The response of Go-deficient transfectants to the highest concentration of LTB4 was indistinguishable from that of sham transfectants. Pertussis toxin alone had never completely inhibited the chemotactic and other responses of blood PMN leukocytes and HL-60 cells to high concentrations of LTB4 and other chemotactic factors (23,24). The incomplete suppression achieved by pertussis toxin was explained by contributions from the pertussin toxin-insensitive Gq class of G proteins (25). The partial inhibitory effectiveness of Go-specific AS mRNA and the apparently lesser dependence of chemotaxis on Go protein at high concentrations of LTB, could be attributable to involvement of residual Go, as well as unaffected G proteins. The lysosomal degranulation of HL-60 cells through G protein-coupled receptors for LTBI and fMLP, as reflected in release of P-glucuronidase, was significantly less for Go-deficient, but not Gi2-deficient, transfectants than for sham transfectants (Fig. 3). The possibility that the reduced functional responses of Go-deficient transfectants were attributable to abnormalities in binding of the chemoattractants was excluded by the results of three studies of binding of r3H1LTB4 and [3H]fMLP. The respective high affinity site Kd values of sham, Go-AS, and Gi2-AS transfectants were 10, 7.6, and 8.2 n~ for LTB4, and 6.7, 5.9, and 4.8 nM for fMLP (mean, n = 3), which reflect no statistically significant differences. Similarly, AS transfections did not alter significantly the total number of receptors or the number of high affinity receptors for LTB4 and fMLP detected in the same studies of binding.
A disturbance of one or more signal transduction pathways in Go-deficient transfectants was suggested by depressed func- tional responses, despite normal binding of LTBl and fMLP. The deficiency in Go a-chain prevented the expected inositol trisphosphate-dependent increase in [Ca2+li (23,241 elicited by optimal concentrations of either LTB4 or fMLP (Fig. 4). In contrast, normal increases in [Ca2+li were observed after LTB4 or fMLP was added to combined transfectants deficient in a-chains of Gil, Giz, and Gi3. The inhibition of calcium responses in Go-deficient HL-60 cell transfectants was not in- creased substantially by the co-transfection of antisense messages directed to Gil-3 (Fig. 4). In additional studies, only Go deficiency suppressed significantly the [Ca2+Ii responses of HL-60 cell transfectants to LTB4 and fMLP ( Table I). The biochemical consequences of Gi2 deficiency in HL-60 cell transfectants was examined in relation to the known inhibition of adenylyl cyclase activity mediated by Gi proteins (1-3). LTB4 inhibited by 75% or more the forskolin-induced increase in concentration of CAMP in sham and Go-deficient transfectants, but only inhibited by a mean of 40% the increase observed in Gi2-deficient HL-60 cells (Table 11).
The application of antisense mRNA approaches to suppress selectively the levels of individual G proteins in HL-60 cells, bearing receptors for LTB4 and fMLP after differentiation, allows assignment of the specific roles of some G proteins in signal transduction. High affinity binding of chemotactic factors was less dependent on G proteins than simal transduction.
both chemotactic and degranulation responses, whereas the inhibition of adenylyl cyclase attributable to Gi2 had no detectable involvement in either functional response.