Autoinhibition of Endothelial Nitric-oxide Synthase

The primary sequences of the three mammalian nitric- oxide synthase (NOS) isoforms differ by the insertion of a 52–55-amino acid loop into the reductase domains of the endothelial (eNOS) and neuronal (nNOS), but not inducible (iNOS). On the basis of studies of peptide derivatives as inhibitors of ⋅NO formation and calmodulin (CaM) binding (Salerno, J. C., Harris, D. E., Irizarry, K., Patel, B., Morales, A. J., Smith, S. M., Martasek, P., Roman, L. J., Masters, B. S., Jones, C. L., Weissman, B. A., Lane, P., Liu, Q., and Gross, S. S. (1997) J. Biol. Chem. 272, 29769–29777), the insert has been proposed to be an autoinhibitory element. We have examined the role of the insert in its native protein context by deleting the insert from both wild-type eNOS and from chimeras obtained by swapping the reductase domains of the three NOS isoforms. The Ca2+concentrations required to activate the enzymes decrease significantly when the insert is deleted, consistent with suppression of autoinhibition. Furthermore, removal of the insert greatly enhances the maximal activity of wild-type eNOS, the least active of the three isoforms. Despite the correlation between reductase and overall enzymatic activity for the wild-type and chimeric NOS proteins, the loop-free eNOS still requires CaM to synthesize ⋅NO. However, the reductive activity of the CaM-free, loop-deleted eNOS is enhanced significantly over that of CaM-free wild-type eNOS and approaches the same level as that of CaM-bound wild-type eNOS. Thus, the inhibitory effect of the loop on both the eNOS reductase and⋅NO-synthesizing activities may have an origin distinct from the loop’s inhibitory effects on the binding of CaM and the concomitant activation of the reductase and ⋅NO-synthesizing activities. The eNOS insert not only inhibits activation of the enzyme by CaM but also contributes to the relatively low overall activity of this NOS isoform.

The enzymatic activities of the three NOS 1 isoforms (1-7) differ in their Ca 2ϩ -dependence: nNOS (NOS-I) and eNOS (NOS-III) are Ca 2ϩ -dependent constitutive isoforms, whereas iNOS (NOS-II), as typified by the inducible macrophage and hepatocyte form, is essentially Ca 2ϩ -independent. This difference in the Ca 2ϩ dependence of the NOS isoforms is the result of a Ca 2ϩ requirement for the reversible binding of CaM to the constitutive isoforms (8,9), in contrast to the almost Ca 2ϩindependent, high affinity binding of CaM to the inducible isoform (10).
The NOS isoforms are also differentiated by their maximum enzymatic activity, nNOS (8,(11)(12)(13)(14) and iNOS (15)(16)(17)(18) exhibiting much higher overall activities than eNOS (19 -24). We have shown that the lower activity of eNOS is caused by a lower ability of its flavoprotein reductase domain to transfer electrons to the catalytic heme domain (25). However, the structural features that impair the reductase activity in eNOS, and hence lower the overall catalytic activity, remain unknown.
Recent evidence (26) suggests that the Ca 2ϩ dependence of nNOS and eNOS is caused by the presence of an autoinhibitory loop, absent in iNOS and P450 reductase, which interferes with the binding of CaM (Fig. 1). The presence of such an autoinhibitory moiety has precedence among CaM-binding enzymes. Indeed, the skeletal and smooth muscle myosin light chain kinases,multifunctionalCaM-dependentproteinkinaseII,CaMdependent protein phosphatase 2B, calcineurin, and phosphorylase kinase are among the CaM-dependent enzymes that possess autoinhibitory structural elements (27). In these enzymes, CaM acts by displacing an autoinhibitory element that interferes with substrate access to the catalytic domain. In calcineurin, the autoinhibitory element is composed of two ␣ helices of 23 amino acids located 54 amino acids after the CaM binding sequence (28,29). In nNOS and eNOS, the proposed autoinhibitory domain is composed of a 52-55-amino acid insert within the FMN binding domain located approximately 80 amino acid residues after the CaM binding sequence (Fig. 1). Location of the putative autoinhibitory element near the junction between the NOS oxygenase and reductase domains is consistent with a possible role in controlling electron transfer between these two domains.
To establish definitively whether the peptide insert in its native protein context functions as an autoinhibitory loop, we have constructed mutants of wild-type and chimeric NOS enzymes lacking the insert and have studied the consequences of deleting the loop on their Ca 2ϩ dependence and catalytic activities. Our results clearly identify the eNOS insert as an autoinhibitory loop that functions not only as an effector of the Ca 2ϩ dependence but also as an electron transfer control element that lowers the catalytic activity of eNOS relative to that of nNOS and iNOS.
Alexis Biochemicals (San Diego), and HEPES buffer from Fisher. Recombinant human CaM was purified from Escherichia coli according to published procedures (31). DNA purification kits and Ni 2ϩ -nitrilotriacetic acid-agarose were purchased from QIAGEN (Chatsworth, CA). BL21(DE3)-competent cells were from Novagen (Madison, WI). All other reagents and materials were from Sigma.
DNA Manipulations-The genes encoding the N/E and I/E chimeras were prepared as described previously (25). The structures of all of the protein constructs employed in this study are shown schematically in Fig. 2. All PCR extensions were performed with Vent R polymerase, which possesses a 3Ј-5Ј proofreading exonuclease to minimize extension errors. The E portion was generated by PCR primer extension using template pBluescript/eNOS, primer 1 (5Ј-CAACC ATCCT GTACG CTAGC GAGAC CGGCC GGG), which introduced the 5Ј-NheI splice site (in bold), and primer 2 (5Ј-CAGCC CCTCT CTTCT AGAAC TGCAG TGG), which produced a 3Ј XbaI site (bold) after the stop codon. These sites were used to produce pcWori//N/E by subcloning of the eNOS reductase domain PCR fragment into pcWori/nNOSNheI761, a Thr 761 3 Ser nNOS mutant possessing the NheI splice site (25). Ser is the homologous amino acid found in eNOS. The iNOS fragment containing the heme domain and CaM binding of I/E was generated by PCR extension using template pBluescript/iNOS and primers 3 (5Ј-AGTCT CACAT ATGGC TTGCC CGTGC AAGTT TCTGT TCAA) and 4 (5Ј-CGGGC GTCGC TAGCA AAGAG GACTG TGGC) to produce, respectively, 5Ј-NdeI and 3Ј-NheI restriction sites (bold), which allowed for subcloning into pcWori//N/E. The genes encoding for chimeras E/I and N/I were prepared in a similar manner. The iNOS reductase domain was generated by PCR primer extension using primer 5 (5Ј-CACAT GCCTC TTTGC TAGCG AGACA GGGAA GTCT) and primer 6 (5Ј-TGCTC TAGAT CAGAG CCTCG TGGCT TT) to produce NheI and XbaI sites (bold), respectively. These sites were used to subclone the macrophage reductase domain PCR fragment into pcWori//E/N and pcWori//N/E, producing pcWori// E/I and pcWori//N/I, respectively.
Protein Expression and Purification-Expression and purification were performed as described previously (25).

FIG. 2.
Schematic of wild-type, chimeric, and insert-deleted NOS proteins. Sequences are shown schematically (not to scale) of the wild-type, chimeric, and deletion proteins. Sequences in the chimeras are color-coded to match those of the wild-type sequences (black, nNOS; dark and light gray, eNOS; white, iNOS). Splice sites in the chimeras (arrows) are indicated in the wild-type sequences (arrows) by a number corresponding to the last amino acid before the splice site. Deleted amino acids in eNOS⌬, N/E⌬, and I/E⌬ are indicated by diagonal shading, with a light gray shading used to indicate an eNOS reductase domain whose insert has been removed.
using an extinction coefficient of ⑀ Soret ϭ 100 mM Ϫ1 cm Ϫ1 for the NOS proteins and ⑀ 276 nm ϭ 3300 M Ϫ1 cm Ϫ1 for human CaM.
Calcium Dependence Measurements-Free calcium concentrations were reproduced as reported before (33) at an ionic strength of 100 mM KCl (or NaCl) and using the value K D (Ca 2ϩ -EGTA) ϭ 27.9 nM at pH 7.50, 37°C.

RESULTS
Expression and Purification-Expression and purification of the NOS proteins (for protein structures, see Fig. 2) was done as reported previously using Ni 2ϩ -nitrilotriacetic acid-agarose and 2Ј,5Ј-ADP-agarose affinity chromatography (25). The final purity of all of the proteins was judged to be greater than 95% by SDS-polyacrylamide gel electrophoresis, and the spectra of all the proteins were consistent with the formation of hemebound, functional proteins (not shown).
Ca 2ϩ Dependence-The insert in the eNOS and nNOS reductase domains may function as an autoinhibitory element that impedes the Ca 2ϩ -dependent binding of CaM except at high Ca 2ϩ concentrations (26). We therefore examined the ⅐ NO synthesizing activity of the enzymes as a function of the free-Ca 2ϩ concentration. The effects of the insert in the reductase domain on the Ca 2ϩ and CaM dependence of the activities of the recombinant proteins were measured in the presence of 500 nM CaM and various free Ca 2ϩ concentrations (Fig. 3). The desired free Ca 2ϩ concentrations were obtained by adding EGTA and Ca 2ϩ -EGTA as indicated by the Ca 2ϩ -EGTA equilibrium dissociation constant at the given temperature and ionic strength (33). The ionic strength in these studies was controlled by adding KCl to the buffer. The results thus obtained with wildtype nNOS (Fig. 3A), from which an apparent K D (K D(app) ) of 300 nM for Ca 2ϩ is calculated, were consistent with those reported by Ruan et al. (34). A slightly lower value of 150 nM was obtained for eNOS. The iNOS activity was independent of the free Ca 2ϩ concentration, again in accord with Ruan et al. (34). The K D(app) values for Ca 2ϩ in the N/E, E/I, and N/I chimeras were 200, 10, and 10 nM, respectively (Fig. 3, A and B). For these three chimeras, the Ca 2ϩ dependence correlates well with the presence (N/E) or absence (E/I, N/I) of the insert in the reductase domain.
Proteins with the insert-deleted eNOS reductase domain required significantly less free Ca 2ϩ for catalytic activity than their parent isoforms (Fig. 3B). eNOS⌬, which only differs from wild-type eNOS by deletion of the putative autoinhibitory loop, had a K D(app) value of 20 nM for Ca 2ϩ rather than 150 nM and thus was activated by a 7-fold lower Ca 2ϩ concentration. Likewise, activation of N/E⌬ required 4-fold less Ca 2ϩ (50 nM The I/E and I/E⌬ chimeras did not differ significantly in their Ca 2ϩ dependence; both exhibited some activity even at a 0.1 nM free Ca 2ϩ concentration, but their activity was not fully expressed at the lowest Ca 2ϩ concentrations (Fig. 3B).
⅐ NO Synthesis-NOS-dependent ⅐ NO synthesis, monitored via the oxidation of oxyhemoglobin, and cytochrome c reduction were assayed in two different buffers. The first buffer, which included flavins and other necessary cofactors, was similar in its lack of added salt to buffers conventionally employed to assay the pure proteins (35,36). The second buffer differed from the first only in that it contained 100 mM KCl to reproduce the conditions employed to assay Ca 2ϩ dependence.
In the presence of CaM, the activity of the wild-type constitutive NOS isoforms is expected to approach zero when [Ca 2ϩ ] free Ͻ Ͻ K D(app) for Ca 2ϩ (Fig. 3A). A similar Ca 2ϩ dependence might be expected for the mutants without the insert, but the maximum NOS activities for the insert-less mutants might be expected to be the same as those for the corresponding CaM-bound wild-type proteins when [Ca 2ϩ ] free Ͼ Ͼ K D(app) for Ca 2ϩ . Under these conditions, the CaM-bound insert-less eNOS mutant should have the same activity as CaM-bound, wild-type eNOS. Contrary to this expectation, the proteins without the peptide insert have enhanced catalytic activities (Fig. 4). In the presence of a high salt concentration, the activities of eNOS⌬, N/E⌬, and I/E⌬ (51.4 Ϯ 2.3, 128 Ϯ 19, and 55.4 Ϯ 1.2 min Ϫ1 , respectively) are approximately double those for the parent eNOS, N/E, and I/E proteins (23.5 Ϯ 2.3, 69.9 Ϯ 2.0, and 32.6 Ϯ 2.8 min Ϫ1 , respectively). Identical effects were seen with KCl or NaCl as the ionic strength buffer (not shown). Thus, in the presence of an approximately physiological salt concentration, the loop deletion mutants consistently expressed a higher activity than the corresponding parent proteins. The chimeras were the only proteins whose activities were altered significantly by changes in the salt concentration, with a higher activity being observed for chimeras with the eNOS reductase domain and a lower activity for those with the iNOS reductase domain. Significant salt-dependent changes in activity were not seen for wild-type eNOS or for the loop-deleted mutant eNOS⌬ in which the wild-type quaternary structure is presumably preserved better than in the chimeras. The increases in the activities caused by salt alone are therefore probably linked to changes in the quaternary structure introduced by the chimeric construct. The subtle change in the reductase/oxygenase interface which would suffice to increase electron transfer and catalytic activity is consistent with the suggestion that salt effects modulate interdomain electron transfer without causing major changes in the NOS structure (37).
Cytochrome c Reduction-Cytochrome c reduction can be used to measure the intrinsic reductase activities of both NOS (38) and P450 reductase (39 -41). For eNOS and nNOS, the binding of CaM stimulates cytochrome c reduction by up to 10-fold (38). The cytochrome c-reducing activities of iNOS (18,25) and CaM-stimulated nNOS (12,38) are comparable to that of P450 reductase (40 -42).
In the presence of CaM and 500 M Ca 2ϩ , the trends in the cytochrome c reductase activities of the proteins adhere to the pattern observed for ⅐ NO synthesis. The rates of the CaM-bound loop deletion mutants (Fig. 5, gray bars) are up to 3-fold higher than those of the corresponding parent proteins. Thus, the activities (ϩKCl) of eNOS⌬, N/E⌬, and I/E⌬ are 3,440 Ϯ 210, 5,580 Ϯ 210, and 1,990 Ϯ 130 min Ϫ1 , respectively, compared with 1,354 Ϯ 24, 1,770 Ϯ 49, and 1,373 Ϯ 90 min Ϫ1 for the corresponding parent proteins.
The enhancement of cytochrome c reduction caused by removal of the insert is even more pronounced for the CaM-free proteins (Fig. 5, white bars, and Table I): CaM-free eNOS⌬ and N/E⌬ had activities of 2,150 Ϯ 120 and 2,720 Ϯ 370 min Ϫ1 , respectively, values that are 10 -30-fold higher than those for the corresponding parent proteins (111 Ϯ 4 and 119 Ϯ 5 min Ϫ1 , respectively). In accord with this finding, the E/I and N/I chimeras with the iNOS loopless reductase have high cytochrome c-reducing activities in the absence of CaM (3,360 Ϯ 150 and 4,390 Ϯ 170 min Ϫ1 ). The CaM-free activities of I/E and I/E⌬ could not be determined because the proteins had to be coexpressed with CaM. As for overall NOS activity, the cytochrome c reduction rates of the proteins with the iNOS reductase suffered a rate decrease of 50 -80% at high salt concentrations.
CaM Dependence of Cytochrome c Reduction-A comparison of the white and gray bars for each protein in Fig. 5 shows that the stimulation ("disinhibition") by CaM is much lower for the loop-deleted mutants. The binding of CaM to eNOS and N/E, which have the wild-type eNOS reductase domain, enhances their activities 12-and 15-fold, respectively. In contrast, the binding of CaM to the corresponding loop-deleted mutants only enhances that of eNOS⌬ by a factor of 0.5 and N/E⌬ by a factor of 2. The comparison could not be made for I/E and I/E⌬ because the proteins had to be coexpressed with CaM. The greater enhancement in the cytochrome c reductase activities of the CaM-free than CaM-bound NOS proteins caused by insert deletion in each instance makes the activities of the CaM-free and CaM-bound proteins nearly equal, as indicated by a lowering in the ratio of the rates (Table I).
The high reductase activities of the CaM-free, insert-deleted proteins are similar to those of the CaM-free chimeras with the insert-less iNOS reductase domain. For example, the cytochrome c reduction rates are comparably high for E/I in the presence and absence of CaM (3,360 Ϯ 150 and 21,37 Ϯ 27 min Ϫ1 , respectively), and the corresponding rates for N/I are 4,390 Ϯ 170 and 1,654 Ϯ 27 min Ϫ1 , respectively. Similarly, the isolated iNOS reductase domain expressed in E. coli has high reductase activity (43).
Ferricyanide Reduction-Ferricyanide reduction provides an independent and useful measure of reductase activity. Electron transfer from P450 reductase to ferricyanide occurs from the FAD cofactor, in contrast to electron transfer to cytochrome c, which occurs from the FMN cofactor (44). In the case of NOS, the binding of CaM accelerates electron transfer to both cytochrome c and ferricyanide (45). We have examined ferricyanide reduction by two proteins containing the eNOS reductase domain, eNOS and N/E, and by the corresponding loop deletion mutants, eNOS⌬ and N/E⌬. In the absence of CaM, the consequences of removing the peptide loop were essentially the same for ferricyanide reduction as for cytochrome c reduction (Fig. 6,  A and B, white bars and Table II). A modest enhancement was observed in the activities of the proteins when the assays were run at a higher salt concentration. eNOS and N/E reduced ferricyanide at similar rates in the low salt medium (482 Ϯ 54 and 955 Ϯ 32 min Ϫ1 ) and gave rise to similar modest increases in activity when salt was added (939 Ϯ 105 and 2116 Ϯ 230 min Ϫ1 ). Deletion of the inserts in the reductase domains of these proteins, yielding eNOS⌬ and N/E⌬, increased ferricyanide-reducing activity 3-7-fold (to 3,160 Ϯ 120 and 3,180 Ϯ 150, respectively) in the low salt medium and to a larger extent (5,890 Ϯ 220 and 5,712 Ϯ 220 min Ϫ1 ) under higher salt conditions. The effects of deleting the loops on ferricyanide reduction were much less dramatic for the CaM-bound than CaM-free proteins (Fig. 6, A and B, gray bars). In the absence of salt, CaM-bound eNOS and N/E reduced ferricyanide at similar rates of 2,250 Ϯ 190 and 3,470 Ϯ 230 min Ϫ1 , respectively. eNOS⌬ and N/E⌬ had higher rates of 8,480 Ϯ 910 and 7,840 Ϯ 400 min Ϫ1 , respectively. In the presence of salt, CaM-bound eNOS⌬ had a higher rate than eNOS (13,100 Ϯ 820 min Ϫ1 and 6,950 Ϯ 310, respectively), and N/E⌬ had a similarly higher rate than N/E (14,430 Ϯ 500 and 9,850 Ϯ 390 min Ϫ1 ). Thus, although loop-deleted proteins showed an enhancement in ferricyanide reduction (less than 2-fold in the presence of salt), the enhancement was much less than the enhancements in the CaM-free proteins (3-6-fold in the presence of salt). This result agrees with the finding that the loop deletions caused smaller enhancements of cytochrome c reduction in the CaM-bound than CaM-free proteins (e.g. in Fig. 4, compare the difference between eNOS minus CaM and eNOS⌬ minus CaM with the difference between eNOS plus CaM and eNOS⌬ plus CaM). DISCUSSION The Ca 2ϩ dependence of ⅐ NO synthesis is a major distinguishing factor among the NOS isoforms. nNOS and eNOS, which have an insert in the FMN binding domain, have a much higher Ca 2ϩ requirement than iNOS, which does not have such an insert. As we show here, deletion of the putative autoinhibitory insert dramatically lowers the Ca 2ϩ requirement for ⅐ NO synthesis by both eNOS⌬ and the N/E⌬ (Fig. 3B). Furthermore, replacement of the reductase domains of eNOS and nNOS with the loopless iNOS reductase domain produces the loopless E/I and N/I chimeras that have a similarly shifted Ca 2ϩ requirement (Fig. 3A). Thus, all of the NOS proteins without the insert, whether based on the wild-type or a chimeric structure, have a lower Ca 2ϩ requirement for activity than the corresponding insert-bearing parent protein.
In addition to shifting the Ca 2ϩ requirement, deletion of the loop unexpectedly produces a modest to large enhancement in the maximum rate of ⅐ NO synthesis. Thus, the activities of eNOS⌬, N/E⌬, and I/E⌬ are higher than those, respectively, of the parent eNOS, N/E, and I/E proteins (Fig. 4), the increase being larger in the presence of a physiological salt concentration. This increase in ⅐ NO synthesis was caused by an increase in the reductase activity (see below), as found previously for other chimeric proteins (25).
Three key observations must be addressed regarding the effect of the loop on reductase activity. 1) Removal of the loop yields a protein with a reductase activity approaching that of the corresponding CaM-bound parent protein. 2) CaM is required for ⅐ NO synthesis despite the increased reductase activity of the loop-free proteins.
3) The loopless proteins have a higher reductase activity than the corresponding loop-possessing parent protein whether CaM is bound or not.
The binding of CaM to nNOS has been reported to stimulate electron transfer not only from the flavins to cytochrome c but also from NADPH to the flavins (45). The reduction of ferricyanide and cytochrome c by P450 reductase has been shown to occur at different sites (44). Cytochrome c accepts electrons exclusively from the FMN, whereas ferricyanide accepts electrons from the FAD and possibly also the FMN. Deletion of the loop increases the rates of cytochrome c (Fig. 5) and ferricyanide ( Fig. 6) reduction by the NOS proteins in both the CaMfree and CaM-bound states. All of the proteins with a specific loop-deleted reductase domain have similar reductase activities if they are compared either in the CaM-free or CaM-bound state (Fig. 5). Furthermore, the reductase activities of these proteins in the absence of CaM are higher than that of the CaM-bound wild-type reductase (Fig. 5).
CaM enhances cytochrome c and ferricyanide reduction by the loop deletion mutants less effectively than by the parent proteins (Tables I and II). A restatement of this is that the CaM-free, loop-deleted proteins already reduce cytochrome c and ferricyanide at rates that more closely approach those for the CaM-bound proteins. A corollary of this is that loop removal increases the reductase activities of the CaM-free proteins to a greater extent than it does the activities of the CaM-bound enzymes. These differences, which are in accord with the finding that the CaM-free proteins are the most stimulated when the insert is deleted, establish that the insert in the eNOS and nNOS reductase domains fulfills an autoinhibitory role. In principle, autoinhibition should be relieved in the CaM-bound proteins whether they retain the autoinhibitory loop or not, so the difference between, for example, CaM-bound eNOS and CaM-bound eNOS⌬ is expected to be small. An increase in the activities of the CaM-bound proteins caused by loop deletion is consistent with the fact that, even in the loop-free proteins, there is a small CaM-dependent increase in the rates of both cytochrome c and ferricyanide reduction. If this enhancement is caused by a conformational change that facilitates electron transfer to the heme domain, the conformational change must occur even in the loopless proteins to account for the CaM-de- pendent incremental stimulation of activity.
P450 reductase and the NOS proteins without an autoinhibitory loop, including the isolated iNOS reductase domain, fulllength iNOS, and the CaM-free N/I and E/I chimeras, exhibit a high electron transfer activity as measured by their ability to reduce cytochrome c, and, in the case of the full-length NOS isoforms, to produce ⅐ NO (Figs. 3 and 4). Conversely, CaM-free NOS proteins with the autoinhibitory loop possess a lower electron transfer activity, as found for CaM-free nNOS, eNOS, and the N/E chimera. CaM-bound proteins with the intact eNOS reductase domain also have a relatively low electron transfer activity, although this activity is higher than that for the CaM-free proteins. In the case of nNOS and I/N, the electron transfer rates when CaM is bound approach those for P450 reductase and iNOS. Likewise, the loopless eNOS⌬ and E/I, N/I, and N/E⌬ chimeras have enhanced reductase activities even when free of CaM. In contrast, the electron transfer activities of eNOS, I/E, and N/E (under salt-free conditions) are low even when CaM is bound. All of these observations are consistent with an autoinhibitory role for the loop with respect to the binding of CaM and the activation of electron transfer and ⅐ NO synthesis. When the strongly inhibitory eNOS loop is removed from these proteins, the enzymatic activities approach, even if they do not equal, those of the loopless iNOS isoform.
The low activity of eNOS compared with nNOS, both of which have inserts in the FMN domain, suggests that only the eNOS autoinhibitory loop is sufficiently effective that its impairment of electron transfer is only partly relieved even when CaM is fully bound. The following evidence supports this inference. 1) The electron transfer activities of CaM-free eNOS and nNOS are lower than that of P450 reductase or the CaM-free iNOS reductase domain expressed as an isolated polypeptide (43). This diminution of the electron transfer activity is completely relieved when the loop in the nNOS reductase domain is displaced by the binding of CaM but is only partially relieved in the eNOS reductase domain by either loop deletion or CaM binding. 2) CaM-free nNOS has a higher intrinsic reductase activity than CaM-free eNOS, suggesting that autoinhibition of electron transfer is less pronounced in CaM-free nNOS. 3) The eNOS reductase activity is low even when CaM is bound, suggesting that CaM-bound eNOS remains partially inhibited. This residual autoinhibition is alleviated when the insert is removed, as evidenced by the fact that the reductase activity of eNOS⌬ approaches those of nNOS and iNOS (Fig. 5). 4) Deletion of the eNOS insert, as in eNOS⌬ and N/E⌬, curtails the CaM dependence of the reductase activity. The enhanced (or disinhibited) activity of eNOS⌬ and N/E⌬, observed even in the absence of CaM, is similar to that of E/I, N/I, and the isolated iNOS reductase domains, which have the naturally loopless iNOS reductase domain (43).
Thus, the peptide insert has two salient effects. First, the insert mediates a CaM-dependent inhibition that contributes to the Ca 2ϩ dependence of eNOS and nNOS. Second, the insert lowers the intrinsic activity of the reductase domain. CaM-free eNOS⌬ and N/E⌬ therefore have higher reductase activities than even CaM-bound eNOS. These effects may be mediated by inter-or intramolecular mechanisms: either 1) the loop directly or through a conformational change sterically hinders access to the reductase by cytochrome c and ferricyanide, or 2) the loop induces a conformational change within the reductase domain which inhibits intradomain electron transfer; that is, removal of the loop increases intramolecular electron transfer, perhaps in a manner similar to that of the "hinge" domain in P450 reductase which has been suggested to improve the efficiency of electron transfer between the flavins. Another possibility is that the loop interferes with productive docking of the NOS oxygenase and reductase domains and that displacement of the loop by CaM eliminates this interference and allows proper docking. However, this possibility rationalizes the activation by CaM but does not explain the higher intrinsic activity of the loop-free reductase in the absence of CaM. The data, in accord with the observations of other groups (46), indicate that the reductase activity is controlled entirely within the reductase domain.
We provide strong evidence that the insert in the eNOS reductase domain is an autoinhibitory structural element that affects the Ca 2ϩ dependence of eNOS and, by extension, also of nNOS. The insert is largely, but not solely, responsible for the Ca 2ϩ /CaM dependence of the constitutive NOS isoforms because CaM is still required for ⅐ NO synthesis by the eNOS⌬ and N/E⌬ chimeras. Furthermore, its absence cannot be the sole determinant of the Ca 2ϩ independence of iNOS because chimeras such as I/N (25) and I/E (Fig. 3) with a loop-containing reductase synthesize ⅐ NO even in the presence of 5 mM EGTA, conditions under which the free Ca 2ϩ concentration is far below the K D(app) for Ca 2ϩ in eNOS and nNOS. The autoinhibitory loop also plays a critical role in governing the net electron transfer ability of the reductase domain. In eNOS, the insert depresses enzymatic activity by inhibiting electron transfer from the reductase to the heme even when CaM is bound. The insert in nNOS may also act in an autoinhibitory manner but is much weaker, in accord with the peptide inhibition studies of Salerno et al. (26) and with the fact that the rate of ⅐ NO synthesis by purified nNOS, albeit high, is still lower than that of iNOS (8,12,15,16,19,20). This difference in the autoinhibitory potency of the eNOS and nNOS inserts may have evolved in response to the different requirements for the synthesis of ⅐ NO satisfied by the different isoforms in their in vivo cellular context.