Homodimerization of Soluble Guanylyl Cyclase Subunits

Soluble guanylyl cyclase (sGC) is an α/β-heterodimeric hemoprotein that, upon interaction with the intercellular messenger molecule NO, generates cGMP. Although the related family of particulate guanylyl cyclases (pGCs) forms active homodimeric complexes, it is not known whether homodimerization of sGC subunits occurs. We report here the expression in Sf9 cells of glutathione S-transferase-tagged recombinant human sGCα1 and β1 subunits, applying a novel and rapid purification method based on GSH-Sepharose affinity chromatography. Surprisingly, in intact Sf9 cells, both homodimeric GSTα/α and GSTβ/β complexes were formed that were catalytically inactive. Upon coexpression of the respective complementary subunits, GSTα/β or GSTβ/α heterodimers were preferentially formed, whereas homodimers were still detectable. When subunits were mixed after expression, e.g.GSTβ and β or GSTα and β, no dimerization was observed. In conclusion, our data suggest the previously unrecognized possibility of a physiological equilibrium between homo- and heterodimeric sGC complexes.

Nitric oxide plays an important role as an intercellular messenger in a great variety of physiological processes (1,2). To mediate these effects, NO binds to and regulates several proteinaceous and nonproteinaceous cellular targets. The presently best characterized signal-transducing receptor for NO is the heme-containing enzyme soluble guanylyl cyclase (sGC) 1 (3)(4)(5). Upon binding of NO to the prosthetic heme group, sGC catalyzes the conversion of GTP to cGMP, which in turn regulates various effector proteins, such as protein kinases, phosphodiesterases, and ion channels (6,7).
The N-terminal domains of both subunits are essential for the stimulation of the enzyme by NO (9,10), whereas heme-binding occurs solely in the ␤ subunit (11)(12)(13). Both the ␣ and ␤ subunit contain a C-terminal cyclase homology domain (CHD), which, in analogy to adenylyl cyclases and pGCs, constitutes a bipartite catalytic center by the association of the ␣ and ␤ C-terminal domains (4,14).
Whereas active pGCs are formed by homodimerization, i.e. association of identical CHDs (14), sGC activity depends on heterodimerization. Only the coexpression of ␣ and ␤ subunit cDNAs in heterologous expression systems constitutes active sGC (15)(16)(17), whereas separate expression of ␣ or ␤ subunits yields neither NO-sensitive nor basally active enzyme. Moreover, sGC activity could not be restored by mixing of the expressed subunits (15).
The central parts of sGC␣ and ␤ (9) share extensive homologies with each other and with a 43-amino acid sequence in pGC that is essential for homodimerization (18). This prompted us to speculate whether homodimer formation of sGC␣ or sGC␤ may also occur. It is, however, unknown whether ␣/␣ and/or ␤/␤ homodimers can assemble intracellularly or whether separately expressed sGC subunits stay monomeric in the absence of a complementary subunit. This would prevent the formation of a "two-CHD" catalytic center and thereby explain the lack of cGMP formation.
Using glutathione S-transferase (GST)-tagged recombinant human sGC subunits expressed in a baculovirus expression system, we here present a single-step purification method for recombinant human sGC (rhsGC) and its subunits, which enabled us to analyze the oligomerization behavior of the sGC subunits. Here we demonstrate for the first time the formation of homodimeric yet inactive sGC complexes. The possible physiological implications for regulation of sGC activity in intact cells are discussed.

EXPERIMENTAL PROCEDURES
Materials-GSH was purchased from Roche Molecular Biochemicals; cell culture materials were from Life Technologies, Inc. (Eggenstein, Germany). All other chemicals were of the highest purity grade available and obtained from either Sigma Chemicals (Deisenhofen, Germany) or Merck AG (Darmstadt, Germany). Water was deionized to 18 M⍀ cm (Milli-Q; Millipore, Eschborn, Germany).
GSH-Sepharose Affinity Chromatography-Glutathione-Sepharose 4B (Amersham Pharmacia Biotech, Freiburg, Germany) was equilibrated with lysis buffer (see above) containing 75 mM NaCl, incubated with crude supernatant fractions of rhsGC-containing Sf9 cells for 1 h at 25°C in a rotation mixer, and washed two or three times with lysis buffer containing 75 mM NaCl. For GSH elution, GSH-Sepharose was incubated with 5 mM GSH in 50 mM Tris-HCl, pH 8.0, for 5 min at 25°C. Fractions were brought to a final concentration of 10% (v/v) glycerol and kept at Ϫ20°C.
Size Exclusion Chromatography-Crude rhsGC-containing Sf9 supernatant fractions were subjected to fast protein liquid chromatography on a Superose 6 column (Amersham Pharmacia Biotech) at a flow rate of 0.2 ml min Ϫ1 in 50 mM Tris-HCl, pH 6.7, 300 mM NaCl, 1 mM EDTA and 1 mM dithiothreitol. Aliquots of each fraction were assayed for rhsGC-immunoreactive protein by Western blot. Signals were quantitated by flatbed scanning and densitometry using the NIH Image software (Division of Computer Research and Technology, National Institutes of Health, Bethesda, MD). The column was calibrated with standard proteins (Sigma) of known Stoke's radii: thyroglobulin (8.5 nm), apoferritin (6.1 nm), alcohol dehydrogenase (4.55 nm), bovine serum albumin (3.55 nm), and carbonic anhydrase (2.01 nm).
sGC Activity Assay-sGC activity was measured as the formation of cGMP at 37°C for 10 min in a total volume of 100 l, containing 50 mM triethanolamine HCl, pH 7.4, 3 mM GSH, 1 mM 3-isobutyl-1-methylxanthine, 5 mM creatine phosphate, 0.25 mg ml Ϫ1 creatine kinase, 500 M GTP, and either 3 mM MgCl 2 or 3 mM MnCl 2 . Reactions were started by adding the enzyme-containing fraction and immediately thereafter the sGC activator sodium nitroprusside (100 M). The cGMP content was determined by an enzyme-linked immunoassay (Biotrend, Cologne, Germany). Results are expressed as the means Ϯ S.E. of at least three experiments.
Western Blot-Immunodetection of nontagged and GST-tagged rh-sGC subunits was performed as described previously (17), using polyclonal antibodies raised against peptide sequences that correspond to hsGC␣1 (amino acids 634 -647) and hsGC␤1 (amino acids 593-614), which were affinity-purified against the respective peptides. Blots were developed using the ECL detection system (Amersham Pharmacia Biotech) according to the manufacturer's protocol.

RESULTS AND DISCUSSION
To examine sGC subunit association, we established a rapid and efficient method to purify rhsGC and its subunits. Fusion proteins composed of a GST affinity tag (25 kDa), and the recombinant human sGC␣1 and ␤1 subunits (rhsGC␣1, 79.5 kDa, and rhsGC␤1, 68, 5 kDa (17)) were constructed and expressed in the baculovirus/Sf9 cell system.
As shown in Fig. 1B, the GST-tagged sGC subunits retained their ability to heterodimerize with the respective complementary nontagged subunit. When coexpressed in Sf9 cells, nontagged ␤ co-eluted with GST␣ and vice versa (lanes 1-4), indicating a direct physical interaction and demonstrating that addition of the GST tag did apparently not interfere with the dimerization function. Whereas coexpression of GST␣ and ␤ subunits yielded active and NO-sensitive sGC in crude Sf9 cell supernatant fractions, only basal sGC activity was obtained upon coexpression of GST␤ and ␣ (Table I). Apparently, GST tagging of the sGC␤ N terminus interfered with NO stimulation, probably because of the close proximity of the GST to the heme-binding site (11)(12)(13). GST␣/␤ activity in the presence of Mg 2ϩ was very similar to that of nontagged rhsGC expressed in the same system (17). In the presence of Mn 2ϩ , basal sGC activity was increased, whereas NO-stimulated activity was decreased (Table I), as reported for native (21) and recombinant sGC (9). Specific basal sGC activity was dramatically increased in the GSH eluate fraction (Table I), similar to sGC␣/␤ prepared by multi-step purification procedures (8,(22)(23)(24). Moreover, NO sensitivity of purified GST␣/␤ was preserved, because the enzyme was activated 30-fold by 100 M sodium nitroprusside (Table I).
Heterodimerization of both GST␣/␤ and GST␤/␣ complexes was dependent on coexpression of the respective subunits (Fig.  1B, lanes 5-8). Accordingly, no basal or NO-stimulated sGC activity was detected in the load or eluate fractions derived from separately expressed and mixed GST␣ and ␤ (or GST␤ and ␣) subunits (data not shown), which was in agreement with previous findings on nontagged ␣ and ␤ subunits (15) When GST␣/␣ and GST␤/␤ were coexpressed in Sf9 cells, none of the crude cell lysates contained any detectable sGC activity (i.e. Ͻ30 pmol cGMP mg Ϫ1 min Ϫ1 ; Table I), neither with Mg 2ϩ nor Mn 2ϩ . To investigate whether sGC subunits homo-oligomerized, fractions were analyzed by Western blot. Interestingly, nontagged ␤ efficiently copurified with GST␤, demonstrating a direct physical interaction between sGC␤ subunits ( Fig. 2A, lanes 3 and 4). This ␤/␤ homodimerization was dependent on coexpression of both subunits (lanes 7 and 8).
Similarly, the nontagged ␣ subunit copurified with GST␣. However, binding of GST␣ to the GSH-Sepharose was weakened when nontagged ␣ was coexpressed ( Fig. 2A, lanes 1 and 2), requiring longer exposure times during Western blot analysis (Fig. 2B, upper panel). In contrast, nontagged ␣, which was expressed separately and mixed with GST␣, had no effect on binding of GST␣ to the beads ( Fig. 2A, lanes 5 and 6). This demonstrated that nontagged ␣ interacted with GST␣ upon coexpression, leading to an interference with binding to GSH-Sepharose, possibly because of steric hindrance. Coelution of ␣ and ␤ with the respective GST-tagged subunit did not result from unspecific binding to the GSH-Sepharose, because the wash steps prior to elution did not contain any nontagged sGC (Fig. 2B). Likewise, ␣ and ␤ did not coelute with GST alone (Fig. 2C). Therefore, coelution of ␣ with GST␣ and of ␤ with GST␤ represent specific protein-protein interactions of identical rhsGC subunits.
To analyze whether single rhsGC subunits form dimeric (as does ␣/␤) or multimeric complexes, crude Sf9 supernatant fractions containing rhsGC were subjected to size exclusion chromatography (Fig. 3). To exclude any potential artifacts caused by the GST tag, nontagged subunits were applied in this set of experiments. As shown in Fig. 3A, coexpressed rhsGC␣ and ␤ subunits perfectly co-eluted upon size exclusion chromatography, with a Stoke's radius of 5.3 nm (Fig. 3A, inset). A similar Stoke's radius (4.8 nm) was reported for heterodimeric sGC purified from rat lung (3). Importantly, rhsGC␣ and rhsGC␤, when expressed separately, showed an elution pattern very similar to heterodimeric sGC (Fig. 3B), suggesting that both subunits indeed exist as homodimers. However, a fraction of rhsGC␤ (Ͻ20%) formed aggregates under these conditions (Fig.  3B, fractions 2-6). The peaks of rhsGC␣ and rhsGC␤ were slightly separated, reflecting apparent differences in the Stoke's radii between the ␣/␣ and ␤/␤ homodimers (Fig. 3B).
To further investigate whether sGC homodimer formation can occur also in the presence of complementary subunits, Sf9 cells were cotransfected with three viruses coding for GST␣, ␣, and ␤ (or GST␤, ␣ and ␤). Viruses coding for ␣ and ␤ were applied at a high m.o.i. to ensure that nearly all cells (Ͼ99%) expressing GST␣ (or GST␤) were simultaneously infected with both ␣ and ␤ viruses. As shown in Fig. 2D, heterodimer (i.e. GST␤/␣ or GST␣/␤), formation is preferred under these conditions in insect cells. However, about 10% of the recombinant protein formed GST␣/␣ or GST␤/␤ complexes even in the presence of the respective complementary subunits. This demonstrated the existence of an equilibrium between homo-and heterodimeric sGC.
Based on these data, sGC␣ and ␤ subunits are in fact capable of forming homodimeric complexes in intact Sf9 cells. It has to be clarified whether classical purification methods efficiently remove homodimeric sGC from crude preparations of recombi-FIG. 2. Homodimerization of rhsGC␣1 and ␤1 subunits. Crude supernatant fractions of Sf9 cells coexpressing GST-tagged and nontagged rhsGC subunits as indicated were subjected to GSH-Sepharose affinity chromatography, and fractions were analyzed as described in the legend to Fig. 1. A, homodimerization of rhsGC subunits is dependent on coexpression. B, GST␣ and ␣ and GST␤ and ␤ specifically co-elute from GSH-Sepharose. C, rhsGC␣1 and ␤1 do not interact with GST. Crude supernatant fractions of Sf9 cells coexpressing GST and rhsGC␣1 (upper panel) or GST and rhsGC␤1 (middle panel) were affinity-purified and analyzed as described in the legend to Fig. 1. rh-sGC␣1 and rhsGC␤1 did not bind to and co-elute with GST (upper and middle panel). GST was specifically eluted with GSH, as shown by Coomassie Blue protein staining (lower panel). D, formation of heterodimeric rhsGC is preferred in Sf9 cells. GST␣, ␣, and ␤ (lanes 1 and 2) or GST␤, ␣, and ␤ (lanes 3 and 4) were coexpressed (see "Experimental Procedures"), and crude supernatant fractions were affinity-purified and analyzed as described in the legend to Fig. 1  nant sGC or whether apparently homogenous sGC preparations contain "silent" homodimeric sGC. With the novel purification method presented here, only GST␣/␤ heterodimers will be purified, because ␤/␤ does not bind to the column. There are different potential mechanisms that might underlie the observed preferential heterodimer formation in Sf9 cells. Homodimer formation may be suppressed in intact cells because of a much higher affinity between complementary subunits. So far, no data on sGC dimerization kinetics and apparent K D values are available. However, the high stability of homodimeric sGC upon mixing of separately expressed sub-units, which seems to be similar to that of heterodimeric sGC, argues against substantial affinity differences. On the other hand, sGC dimerization might be a regulated process in living cells. The existence of at least two different isoforms of each subunit (␣1, ␣2, ␤1, and ␤2) led to the concept that sGC activity is regulated in vivo by alternative heterodimerization (4,5). It is an intriguing possibility that regulation of sGC activity in vivo might involve not only alternative heterodimerization but also changes in the extent of homodimerization. Based on our data, it cannot be excluded that a further protein is associated with homodimeric sGC complexes, which may be involved in complex formation and therefore be a functional inhibitor of sGC activity. An endogenous inhibitor of sGC has been described in Ref. 25. Interestingly, both homo-and heterodimerization depend on co-or post-translational processes. It has been suggested that chaperone-mediated formation of heterooligomeric protein complexes is involved in the regulation of signaling pathways (26). A related process may regulate sGC protein-protein interactions and thus NO/cGMP signaling.