1

Accelerated Publication Journal of Biological Chemistry

Heme Oxygenase–2 Activated by Calcium-Calmodulin

Darren Boehning‡, Leela Sedaghat‡, Thomas W. Sedlak‡¶, and Solomon H. Snyder‡§¶║

‡Departments of Neuroscience, §Pharmacology and Molecular Sciences, and

¶Psychiatry and Behavioral Sciences, Johns Hopkins University School of Medicine,

725 N. Wolfe Street, Baltimore, MD 21205, USA

║To whom correspondence should be addressed: Department of Neuroscience, Johns

Hopkins University School of Medicine, 725 N. Wolfe Street, 813 WBSB, Baltimore, MD

21205. Tel.: 410-955-3024; Fax: 410-955-3623; E-mail: ssnyder@jhmi.edu.

Character count: 27,868 Running title: Calcium-calmodulin activates HO2

by guest on February 16, 2018http://w

ww

.jbc.org/D

ownloaded from

2

Summary

The heme oxygenase family of enzymes catalyzes the metabolism of heme to

biliverdin, ferrous iron, and carbon monoxide (CO). At least two isoforms exist, heme

oxygenase-1 (HO1) and heme oxygenase-2 (HO2), which are encoded by separate

genes. Heme oxygenase-2 is selectively enriched in neurons, and substantial evidence

suggests that HO2-derived CO functions as a neurotransmitter/neuromodulator.

However, a molecular mechanism for the rapid activation of HO2 during neuronal

activity has not been described. Through a yeast-2-hybrid screen we identified

calmodulin as a potential regulator of HO2 activity. Calmodulin binds with nanomolar

affinity to HO2 in a calcium-dependent manner via a canonical 1-10 motif, resulting in a

three fold increase in catalytic activity. Mutations within this motif block calmodulin

binding and calcium-dependent stimulation of enzyme activity in vitro and in intact cells.

The calcium mobilizing agents ionomycin and glutamate stimulate endogenous HO2

activity in primary cortical cultures, establishing in vivo relevance. Calcium-calmodulin

provides a mechanism for rapid and transient activation of HO2 during neuronal activity.

Introduction

Nitric oxide (NO)1 and carbon monoxide (CO) are increasingly appreciated as

neurotransmitters in the central and peripheral nervous system(1). Most

neurotransmitters are stored in synaptic vesicles with only a small percentage released

following each neuronal depolarization and large storage pools remaining for

succeeding neural volleys. Neither NO or CO are stored in vesicles so that their release

following neuronal depolarization is presumably coupled to activation of their respective

by guest on February 16, 2018http://w

ww

.jbc.org/D

ownloaded from

3

biosynthetic enzymes, neuronal NO synthase (nNOS) and heme oxygenase-2 (HO2).

For NO, calcium influx into depolarized neurons binds to calmodulin, activating

nNOS(2). No mechanism for rapid activation of HO2 has yet been reported. Recently

we showed that CK2 (casein kinase-2) phosphorylates and activates HO2(3). In brain

cultures protein kinase C phosphorylates and activates CK2. Since protein kinase C

can be activated by calcium, neuronal depolarization and associated calcium entry

might activate protein kinase C to phosphorylate and activate CK2 which in turn

phosphorylates and activates HO2(3). However, it is not likely that this mechanism

would be rapid enough to provide CO for most instances of neurotransmission.

In the present study we report the binding of calmodulin to HO2 in a calcium-

dependent fashion. In response to calcium mobilization by ionomycin and the excitatory

neurotransmitter glutamate in neurons, we observe physiologic activation of HO2 and

formation of the heme metabolite, bilirubin. Thus, as with NO, CO formation in

neurotransmission reflects calcium-calmodulin activation of the biosynthetic enzyme.

Experimental Procedures

Yeast-2-hybrid Yeast 2-hybrid analysis was performed using the MATCHMAKER two-

hybrid system 3 (Clontech, Palo Alto, CA). Amino acids 1-292 of rat HO2 were cloned

by PCR into the EcoR1/Pst1 sites of pGBKT7 to express HO2 as a fusion with the

GAL4 DNA binding domain. Expression of the fusion protein was confirmed by western

blotting, and a rat brain cDNA library fused to the activating domain was transformed

into PJ69-4A yeast. Yeast transformants were grown on -Leu/-Trp/-Ade/-His plates. All

by guest on February 16, 2018http://w

ww

.jbc.org/D

ownloaded from

4

colonies were re-streaked onto the same media containing X-α-gal, and colonies which

turned blue were sequenced.

Expression constructs- Rat HO2 amino acids 1-292 was cloned into the EcoRI/NotI

sites of glutathione-S-transferase (GST) expression vector pGEX 4T-2 (Amersham

Biosciences Corp, Piscataway, NJ) by PCR. Forward primer rHO2F was 5’-

TTGAATTCATGTCTTCAGAGGTGGAG, and reverse primer was 5’-

TTTTGCGGCCGCCTTCCTCAGCACAGCCAT. Rat HO2 1-90 and 1-55 were cloned

using rHO2F forward primer and reverse primers rHO90R (5-

TTTTGCGGCCGCGTGGTCCTTGTTGCGATC) and rHO55R (5-

AAGCGGCCGCTCATTTGACAAACTGGGTATT), respectively. Site-directed mutations

F56A, F56E, L57A, L57Q, F66A, and F66E in rat HO2 were introduced using the

QuickChange Site-directed Mutagenesis Kit (Stratagene, La Jolla, CA). Sequences of

the mutagenic primers are available as supplementary Table I. All mutations and were

confirmed by sequencing. Rat HO2 in pcDNA3.1 (Invitrogen, Carlsbad, CA) has been

described previously(3).

In vitro binding assays- Binding of GST fusion proteins to calmodulin was assayed by

rotating 20 µg of fusion protein with 20 µl of calmodulin-agarose (1.8 mg/ml; Sigma, St.

Louis, MO) for 1 hour at 4oC in buffer A (150 mM NaCl, 50 mM Tris [pH 7.8], 1% Triton,

1 mM EDTA +/- 5 mM CaCl2). The agarose beads were washed X 3 with 500 µls buffer

A +/- CaCl2. The washed beads were quenched directly with SDS sample buffer and

analyzed by SDS-PAGE. Western blotting was used to determine HO2 binding using

by guest on February 16, 2018http://w

ww

.jbc.org/D

ownloaded from

5

either an anti-GST antibody (Sigma, St. Louis, MO) or in some cases an anti-HO2

antibody(4). Analysis of the site directed mutants was performed by incubating 2 µg of

GST fusion protein with 10 µg of purified calmodulin (Sigma) in 500 µl buffer A +/-

calcium. Interacting complexes were immobilized on GST-agarose, washed, and

analyzed by western blotting. Input of HO2 protein was analyzed by Ponceau S

staining, and pull-down of calmodulin was visualized with anti-calmodulin antibody

(Upstate Biotechnology, Charlottesville, VA).

Co-immunoprecipitation- Rat cortex was cut sagitally into left and right hemispheres,

and one hemisphere was homogenized in buffer A with calcium, and the other without

calcium in buffer A. Lysates were centrifuged at 10,000 xg, and the 100 µg of the

supernatant was pre-cleared of rat immunoglobulin with Pansorbin (EMD Biosciences,

San Diego, CA). Cleared lysates were immunoprecipitated with anti-HO2 antibody

overnight in the presence or absence of calcium. Where indicated, 10 µM of the peptide

calmodulin antagonist comprised of amino acids 290-309 of calmodulin-dependent

protein kinase II was included in the immunoprecipitation. Immunocomplexes were

washed and analyzed by SDS-PAGE, followed by immunoblotting with anti-HO2 and

anti-calmodulin.

Enzyme assays- Heme oxygenase activity was measured essentially as described

previously(3) with a modified assay buffer which consisted of 150 mM KCl, 50 mM

HEPES 7.5, 500 µM BAPTA, and 100 µM deferoxamine. Free calcium concentrations

were calibrated using a calcium-selective mini-electrode and confirmed using

by guest on February 16, 2018http://w

ww

.jbc.org/D

ownloaded from

6

fluorescent dyes. Recombinant cytochrome P450 reductase (Sigma, St. Louis, MO),

1mM reduced β-nicotinamide adenine dinucleotide 2'-phosphate, and 15 µM

[55Fe]hemin (RI Consultants LLC, Hudson, NH) were added just prior to the start of each

assay. GST-HO2 was added at a concentration determined for each preparation to be

within the linear range of the assay. In some experiments, GST-HO2 was

phosphorylated for 1 hour at 30oC with recombinant CK2 as described elsewhere(3).

Bilirubin measurement- Approximately 107 cells were assayed for bilirubin formation in

triplicate per experiment, with each experiment repeated three times. Where indicated,

cells were pre-incubated with 20 µM calmidazolium chloride (CzCl; Biomol, Plymouth

Meeting, PA) prior to agonist addition. Agonists were applied for 10 min, after which an

equal volume of ethyl acetate was added and the cells were vortexed vigorously for 5

min to extract cellular bilirubin. Control plates without agonist treatment were quenched

at time zero (just before agonist addition) and after the 10 min time course. Samples

were centrifuged for 10 min at RT, and the organic phase was removed into a fresh tube

and evaporated under a nitrogen gas stream. The residue was resuspended in 50 mM

NaOH and the absorbance was measured at 440 nm. Bilirubin standards (0.05, 0.5,

5.0, and 25 nmoles) were freshly prepared in 50 mM NaOH and processed exactly as

the biological samples to generate a standard curve for quantitation of bilirubin

formation. A small aliquot of cells was saved from each condition for protein

measurement and western blotting. Bilirubin formation in HEK 293 cells was nominal

unless HO2 was transfected, demonstrating the validity of this assay for measuring HO2

activity. All manipulations were performed in dim light.

by guest on February 16, 2018http://w

ww

.jbc.org/D

ownloaded from

7

Data Analysis- Non-linear curve fitting to the four parameter logistic equation

y=min+(max-min)/(1+(x/EC50)^Hillslope) was used to determine EC50 values using

SigmaPlot software (SPSS Science, Chicago, IL). Statistical significance was evaluated

using unpaired two-tailed t tests on data points from at least 3 separate experiments.

Data comparisons were deemed significant if p < 0.02.

Results and Discussion

In yeast 2-hybrid analysis utilizing HO2 without the C-terminal transmembrane

domain as bait, we identify calmodulin as a binding partner (Fig. 1). Analysis of the

primary sequence of HO2 using an algorithm to detect calmodulin binding sites (5)

indicates that the region encompassing amino acids 56-76 on the N-terminal region of

the protein was likely to bind calmodulin (Figure 1A). Manual inspection of this

sequence reveals three separate 1-10 consensus calmodulin binding motifs

encompassing amino acids 56-65, 57-66, and 66-75. The 1-10 motif is characterized by

bulky aromatic/hydrophobic residues at the 1st and 10th position on an alpha helix(6).

Heme oxygenase-1 (HO1) is a closely related inducible enzyme whose crystal structure

has been elucidated(7). Sequence alignment and mapping of the HO2 sequence onto

the HO1 structure reveals that only the 1-10 motif encompassing amino acids 66-75

resides on an alpha helix (Fig. 1A; Supplementary Figure 1). Thus, we predict that 1-

10 (66-75) is the primary binding site. To test this hypothesis, we examined the binding

of HO2 expressed as a glutathione-S-transferase (GST) fusion protein to calmodulin

immobilized on agarose beads (Fig. 1B). The full-length protein lacking the

by guest on February 16, 2018http://w

ww

.jbc.org/D

ownloaded from

8

transmembrane domain, HO2(1-292), binds robustly in a calcium-dependent fashion to

calmodulin, as does HO2(1-90). By contrast, HO2(1-55), the N-terminal portion of the

enzyme lacking the predicted calmodulin binding consensus sequence, fails to bind

calmodulin-agarose. To elucidate which consensus 1-10 motif is responsible for binding

of HO2 to calmodulin, we performed site-directed mutagenesis (Fig. 1C). Mutation of

phenylalanine 56 to alanine or glutamic acid fails to alter calcium-dependent binding of

HO2 to calmodulin, suggesting that 1-10 (56-65) is not responsible for calmodulin

binding. Similarly, mutation of leucine 65 to alanine does not alter binding (data not

shown). Surprisingly, conversion of leucine 57 to alanine results in constitutive high-

affinity binding. Mutation of this residue to glutamine, the corresponding residue in

HO1, does not affect binding to calmodulin. In contrast, mutation of phenylalanine 66 to

either alanine or glutamic acid virtually abolishes binding. Combined with the lack of

effect of mutation L57Q and the presence of the 1-10 (66-75) motif on an alpha helix

(Supplementary Fig. 1), these data strongly suggest that calmodulin binds to 1-10 (66-

75).

One parallel between NOS and HO is the existence of discrete constitutive and

inducible forms of the enzymes. Inducible NOS (iNOS) is not normally activated by

stimuli that increase intracellular calcium, but instead new iNOS protein synthesis

occurs following stimuli such as endotoxin exposure(8-10). iNOS binds calmodulin at

low, physiologic calcium levels so that stimuli that increase intracellular calcium do not

alter enzyme activity(11). The inducible HO isoform, HO1, has 50% homology in the

sequence encompassing the 1-10 motif of HO2, but lacks one of the motif’s key amino

acids, phenylalanine 75 (Fig.1A). As predicted by the absence of this amino acid, we

by guest on February 16, 2018http://w

ww

.jbc.org/D

ownloaded from

9

detect negligible binding of HO1 to calmodulin in the presence or absence of calcium

(Fig.1D).

To ascertain whether endogenous HO2 and calmodulin interact in mammalian

tissues, we conducted immunoprecipitation experiments from rat brain extracts (Fig.

1e). We observe robust, calcium-dependent co-immunoprecipitation of calmodulin with

HO2, which is reversed by calmodulin antagonist (Fig. 1e).

We demonstrate a direct requirement of calcium-calmodulin for HO2 catalytic

activity (Figure 2). Calcium stimulates HO2 in the presence of 100 nM calmodulin with

an EC50 of ~300 nM (Fig. 2a). This concentration is consistent with physiologic levels

which would occur during neuronal depolarization. Calmodulin potently stimulates

enzyme activity with an EC50 of 2 nM, similar to the potency of calmodulin in activating

nNOS(2) (Fig. 2b). Mutation of phenylalanine-66, a key amino acid in the 1-10 motif for

calmodulin binding (Fig. 1C), abolishes activation of the enzyme. The importance of

calmodulin is further substantiated by inhibition of HO2 activity with the calmodulin

antagonist calmidazolium with an IC50 of 8 µM, similar to its potency in blocking other

calmodulin-dependent processes(12-14). Another calmodulin antagonist, W7, also

inhibits the calcium-calmodulin dependent activation of HO2 (data not shown).

We wondered whether the phosphorylation of HO2 by CK2 synergizes with or

antagonizes calmodulin activation, and so examined the influence of calcium-calmodulin

on the unphosphorylated and the phosphorylated enzyme (Fig. 2d). The effect of CK2

pre-phosphorylation on calmodulin activation was examined at 10 nM and 10 µM

calcium. As reported previously(3), phosphorylation by CK2 triples enzyme activity at

by guest on February 16, 2018http://w

ww

.jbc.org/D

ownloaded from

10

low calcium levels. No further stimulation occurs in the presence of calmodulin with

either 10 nM or 10 µM calcium.

To evaluate the role of calcium-calmodulin in regulating HO2 activity in intact

cells, we employed HEK 293 cells transfected with HO2 and primary cortical neuronal

cultures (Figure 3). In transfected HEK 293 cells, the calcium ionophore ionomycin

markedly activates the recombinant enzyme as evidenced by the increased production

of the HO2 metabolite bilirubin, with stimulation prevented by the calmodulin antagonist

calmidazolium (Fig. 3A). Mock transfected cells are not responsive to ionomycin, due to

low endogenous levels of HO2 (Fig. 3A, inset). HO2 with phenylalanine-66 mutated to

glutamate is not activated by ionomycin. In cortical cultures ionomycin increases

bilirubin formation 4.5 fold with stimulation blocked by calmidazolium (Fig. 3B). The

excitatory neurotransmitter glutamate also stimulates bilirubin formation, with stimulation

blocked by calmidazolium. Serotonin (10 µM) fails to stimulate bilirubin formation in

cortical cultures. Levels of bilirubin in serotonin-treated cultures, the same as in basal

cultures, are decreased by calmidazolium, indicating that calcium-calmodulin activation

of HO2 determines basal levels of bilirubin.

Our findings indicate a rapid activation of HO2 in intact cells by calcium-

calmodulin. How does this mechanism interface with HO2 activation by CK2? We did

not observe any influence of calcium-calmodulin upon CK2 activation of the enzyme.

CK2 phosphorylates HO2 at serine-78 (rat sequence), only 3 amino acids away from

the calmodulin binding site on the same alpha helix (Supplementary Fig. 1). We

hypothesize that conformational changes induced by calmodulin binding and by

phosphorylation of serine 78 are similar, and thus are not additive.

by guest on February 16, 2018http://w

ww

.jbc.org/D

ownloaded from

11

Acute exposure of cortical neurons to the excitatory neurotransmitter glutamate

increased HO2 activity in a calcium-calmodulin dependent manner (Fig. 3B). Other

groups have shown that stimulation of metabotropic(15,16) and ionotropic(17-21)

glutamate receptors leads to increased CO production. Leffler et. al.(17) found in

cerebral microvessels that tyrosine kinase inhibitors blocked CO production in response

to glutamate, whereas manipulations of intracellular calcium had no effect. Notably,

ionomycin treatment in calcium-replete medium had no effect on CO production in the

microvessels, and depletion of internal calcium stores in calcium-free medium did not

affect CO production in response to glutamate. In that study, the microvessels were

exposed to 1 µM ionomycin and/or 100 µM glutamate for a minimum of 30 min while CO

was allowed to accumulate within the headspace. These manipulations would be

expected to cause secondary effects on cellular physiology, including induction of the

ER stress response and apoptosis. Therefore it is unclear if the actions of glutamate in

cerebral microvessels are truly calcium-independent. This discrepancy can also be

explained by glutamate activating a separate, calcium-independent pathway in

endothelial cells, requiring the activation of tyrosine kinases.

Activating HO2 by calcium-calmodulin in stimulated neurons presumably occurs

on a millisecond timescale, and thus is consistent with requirements for synaptic

transmission. Other physiologic processes regulated by CO are more long-term. For

instance, CO regulates long-term adaptation to odorant stimulation in olfactory

neurons(22,23), and mediates slow relaxation of the internal anal sphincter(3,24).

Conceivably CK2 phosphorylation mediates these longer-term activations of HO2.

Analogous differential roles for calcium-calmodulin-dependent and phosphorylation-

by guest on February 16, 2018http://w

ww

.jbc.org/D

ownloaded from

12

dependent stimulations occur for endothelial NOS (eNOS). Thus, acetylcholine, by

releasing intracellular calcium, rapidly activates eNOS to initiate blood vessel relaxation,

whereas long-term relaxation mediated by shear stress and vascular endothelial growth

factor maintain vasodilation by activating Akt kinase to phosphorylate eNOS, increasing

enzyme activity in a calcium-independent fashion(25-27). Similarly, penile erection is

initiated by depolarization-induced calcium-calmodulin activation of nNOS (28,29) and

sustained by subsequent augmentation of phosphatidylinositol-3 kinase, which activates

Akt to phosphorylate and activate eNOS (29,30).

Acknowledgments

This work was supported by USPHS grant MH-18501 and DA-000266 and Research

Scientist Award DA-00074 (SHS), and National Research Service Award NS-043850

(DB). DB wishes to thank Julia and Alexandra Boehning for inspiration.

References

1. Boehning, D., and Snyder, S. H. (2003) Annu Rev Neurosci 26, 105-131

2. Bredt, D. S., and Snyder, S. H. (1990) Proc Natl Acad Sci U S A 87, 682-685

3. Boehning, D., Moon, C., Sharma, S., Hurt, K. J., Hester, L. D., Ronnett, G. V.,

Shugar, D., and Snyder, S. H. (2003) Neuron 40, 129-137

4. Baranano, D. E., Wolosker, H., Bae, B. I., Barrow, R. K., Snyder, S. H., and

Ferris, C. D. (2000) J Biol Chem 275, 15166-15173

5. Yap, K. L., Kim, J., Truong, K., Sherman, M., Yuan, T., and Ikura, M. (2000) J

Struct Funct Genomics 1, 8-14

by guest on February 16, 2018http://w

ww

.jbc.org/D

ownloaded from

13

6. Hoeflich, K. P., and Ikura, M. (2002) Cell 108, 739-742

7. Schuller, D. J., Wilks, A., Ortiz de Montellano, P. R., and Poulos, T. L. (1999) Nat

Struct Biol 6, 860-867

8. Lowenstein, C. J., Glatt, C. S., Bredt, D. S., and Snyder, S. H. (1992) Proc Natl

Acad Sci U S A 89, 6711-6715

9. Lyons, C. R., Orloff, G. J., and Cunningham, J. M. (1992) J Biol Chem 267, 6370-

6374

10. Xie, Q. W., Cho, H. J., Calaycay, J., Mumford, R. A., Swiderek, K. M., Lee, T. D.,

Ding, A., Troso, T., and Nathan, C. (1992) Science 256, 225-228

11. Stevens-Truss, R., and Marletta, M. A. (1995) Biochemistry 34, 15638-15645

12. Asano, M. (1989) J Pharmacol Exp Ther 251, 764-773

13. Sampaio-Maia, B., and Soares-Da-Silva, P. (2001) Cell Biol Int 25, 245-252

14. Agullo, L., and Garcia, A. (1997) Eur J Pharmacol 323, 119-125

15. Lin, C. H., Lo, W. C., Hsiao, M., Tung, C. S., and Tseng, C. J. (2004) J

Pharmacol Exp Ther 308, 1213-1218

16. Lo, W. C., Chan, J. Y., Tung, C. S., and Tseng, C. J. (2002) Eur J Pharmacol

454, 39-45

17. Leffler, C. W., Balabanova, L., Fedinec, A. L., Waters, C. M., and Parfenova, H.

(2003) Am J Physiol Heart Circ Physiol 285, H74-80

18. Parfenova, H., Fedinec, A., and Leffler, C. W. (2003) J Cereb Blood Flow Metab

23, 190-197

19. Fiumana, E., Parfenova, H., Jaggar, J. H., and Leffler, C. W. (2003) Am J Physiol

Heart Circ Physiol 284, H1073-1079

by guest on February 16, 2018http://w

ww

.jbc.org/D

ownloaded from

14

20. Robinson, J. S., Fedinec, A. L., and Leffler, C. W. (2002) Am J Physiol Heart Circ

Physiol 282, H2371-2376

21. Li, X., and Clark, J. D. (2002) Eur J Pharmacol 450, 43-48

22. Zufall, F., and Leinders-Zufall, T. (1997) J Neurosci 17, 2703-2712

23. Leinders-Zufall, T., Shepherd, G. M., and Zufall, F. (1995) J Neurophysiol 74,

1498-1508

24. Watkins, C. C., Boehning, D., Kaplin, A. I., Rao, M., Ferris, C. D., and Snyder, S.

H. (2004) Proc Natl Acad Sci U S A 101, 2631-2635

25. Fulton, D., Gratton, J. P., McCabe, T. J., Fontana, J., Fujio, Y., Walsh, K.,

Franke, T. F., Papapetropoulos, A., and Sessa, W. C. (1999) Nature 399, 597-

601

26. Dimmeler, S., Fleming, I., Fisslthaler, B., Hermann, C., Busse, R., and Zeiher, A.

M. (1999) Nature 399, 601-605

27. Michell, B. J., Griffiths, J. E., Mitchelhill, K. I., Rodriguez-Crespo, I., Tiganis, T.,

Bozinovski, S., de Montellano, P. R., Kemp, B. E., and Pearson, R. B. (1999)

Curr Biol 9, 845-848

28. Burnett, A. L., Lowenstein, C. J., Bredt, D. S., Chang, T. S., and Snyder, S. H.

(1992) Science 257, 401-403

29. Hurt, K. J., Musicki, B., Palese, M. A., Crone, J. K., Becker, R. E., Moriarity, J. L.,

Snyder, S. H., and Burnett, A. L. (2002) Proc Natl Acad Sci U S A 99, 4061-4066

30. Burnett, A. L., Johns, D. G., Kriegsfeld, L. J., Klein, S. L., Calvin, D. C., Demas,

G. E., Schramm, L. P., Tonegawa, S., Nelson, R. J., Snyder, S. H., and Poss, K.

D. (1998) Nat Med 4, 84-87.

by guest on February 16, 2018http://w

ww

.jbc.org/D

ownloaded from

15

Footnotes

1The abbreviations used are: CK2, casein kinase-2; CO, carbon monoxide; CzCl,

calmidazolium chloride; eNOS, endothelial nitric oxide synthase; GST, glutathione-S-

transferase; HEK, human embryonic kidney; HO2, heme oxygenase-2; HO1, heme

oxygenase-1; iNOS, inducible nitric oxide synthase; NO, nitric oxide; nNOS, neuronal

nitric oxide synthase.

Figure Legends

Figure 1 Calmodulin binds in a calcium-dependent manner to HO2 (A) Schematic

representation of rat HO2 illustrating the amino acids which encompass the CK2

phosphorylation site (S78), the catalytic core (Cat), the transmembrane helix (TM) and

the predicted calmodulin binding site (CaM). Sequence analysis of the calmodulin

binding site reveals three potential 1-10 calmodulin binding motifs (56-65, 57-66, and

66-75). Comparison of HO2 sequence to HO1 reveals only motif 66-75 resides on an

alpha helix (see supplementary figure 1). The CK2 phosphorylation site is indicated by

an asterisk in the alignment. (B) GST-HO2 constructs encoding amino acids 1-292, 1-

90, and 1-55 were tested for the ability to bind calmodulin-agarose in the presence or

absence of calcium. Load is 1/10 the input of fusion protein. (C) Pull-down of purified

calmodulin with GST-HO2 containing point mutations in the 1-10 binding motifs. Amino

acids are indicated by single letter notation, the number of the residue (rat HO2

sequence), and the amino acid it was replaced by. GST-HO2 was analyzed by

Ponceau S staining, and pull-down of calmodulin (CaM) was visualized by western

blotting of the same nitrocellulose sheet. (D) GST-HO2 or GST-HO1 were tested for

by guest on February 16, 2018http://w

ww

.jbc.org/D

ownloaded from

16

the ability to bind calmodulin-agarose +/- calcium. Load is 1/10 the input of fusion

protein. (E) HO2 from rat brain was immunoprecipitated +/- calcium. Where indicated,

20 µM of the pseudo-substrate calmodulin antagonist CaM kinase II (290-309) was

added (CaM Inh.). Immunoprecipitates were analyzed by SDS-PAGE and western

blotting for HO2 and calmodulin. Lysate is 1/10 the amount offered for

immunoprecipitation.

Figure 2 Calmodulin stimulates HO2 activity in vitro (A) Wild-type GST-HO2 was

incubated with 100 nM calmodulin at various concentrations of calcium, and the

enzymatic activity of HO2 was examined. Enzyme activity was increased more than

three-fold in the presence of calmodulin, with an EC50 of 287 nM calcium (solid circles).

In the absence of calmodulin, calcium had no effect (open circles). The F66E mutant

retained basal catalytic activity, but was not activated by calcium-calmodulin (closed

triangles). (B) Wild-type GST-HO2 was incubated with 1.0 µM calcium and varying

concentrations of calmodulin (closed circles). The EC50 for activation by calmodulin was

2 nM. Calmodulin had no effect on the F66E mutant (closed triangles). (C) Effect of

the calmodulin antagonist calmidazolium chloride on the activity of HO2 assayed in the

presence of 100 nM calmodulin and 1.0 µM free calcium. Under these conditions,

calmidazolium chloride inhibited HO2 activity with an IC50 of 8.0 µM. (D) Effect of pre-

phosphorylation of HO2 with CK2 on the calcium sensitivity of enzyme activity in the

presence of 100 nM calmodulin. Enzyme activity was activated similarly at 10 nM and

10 µM calcium by CK2. Calcium-calmodulin activation at 10 µM calcium was not

additive to CK2 activation. Confirmation of phosphorylation efficiency was determined

by guest on February 16, 2018http://w

ww

.jbc.org/D

ownloaded from

17

by western blotting of a small portion of the kinase reaction with phosphoserine

antibodies (inset; Zymed laboratories, San Francisco, CA). All data are pooled from at

least three separate determinations in duplicate. *p < 0.02

Figure 3 Calmodulin mediates rapid activation of HO2 in response to calcium

mobilizing agonists (A) HEK 293 cells were transfected with pcDNA3.1 (mock), rat

HO2, or F66E HO2. Bilirubin formation was measured in response to a 10 minute

stimulation with DMSO (none) or 1 µM ionomycin (iono) +/- 20 µM calmidazolium

chloride (CzCl). Mock transfected cells had no increase in bilirubin formation in

response to ionomycin, as expected from the low amounts of endogenous HO2 (inset).

Transfection of HO2 caused a ~4.5 fold increase in the rate of bilirubin formation in

response to ionomycin, which was blocked by CzCl. In contrast, cells transfected with

HO2 F66E were insensitive to ionomycin treatment. (B) Primary rat cortical neuronal

cultures were treated with DMSO (none), 1 µM ionomycin (iono), 100 µM glutamate

(glu), or 10 µM serotonin (5-HT) in the presence or absence of CzCl for 10 min and

bilirubin formation was determined as in panel A. Data are pooled from 6 separate

determinations performed in duplicate *p < 0.01

Supplementary Figure 1 Structure properties of HO2 amino acids 56 to 76

mapped onto the HO1 crystal structure (A) Location of HO2 amino acids F56 to F76

mapped onto the full HO1 crystal structure, and (B) residues F56 to the end of HO1

helix 3. The putative calmodulin binding region F56 to F76 is highlighted in yellow. The

CK2 phosphorylation site at Ser 78 (rat HO2 sequence) is also indicated (CK2-P,

highlighted in yellow). Mapping of the equivalent residues of HO2 onto the HO1

by guest on February 16, 2018http://w

ww

.jbc.org/D

ownloaded from

18

structure was done by alignment with ClustalW (Higgins D., Thompson J., Gibson

T.Thompson J.D., Higgins D.G., Gibson T.J. (1994) Nucleic Acids Res. 22:4673-4680).

HO2 was confirmed to have a similar predicted secondary structure to HO1 within the

region of the calmodulin binding site using the PredictProtein server (B Rost and J Liu

(2003) Nucleic Acids Research 31(13): 3300-3304). The crystal structure of HO1

(MMDB accession # 13250) was manipulated with Cn3D version 4.0

(http://www.ncbi.nlm.nih.gov/Structure/CN3D/cn3d.shtml).

by guest on February 16, 2018http://w

ww

.jbc.org/D

ownloaded from

Figure 1 Boehning et. al.

62

49

38

28

Ca :2+ - + - + - + Load

1-292GST-HO2: 1-901-90 1-55

AS78

N CTM

CatCaM

29256 76 145 177

B

C

Ca :2+

- - + + - +

CaM

Blot

Lysate

HO2 i.p.

17

HO2

Blot

- + - + CaM Inh.:

D

E

38

Ca :2+ - + - + - + - + - + - + - +

GST-HO2

WT F56A F56E L57A L57Q

1-55

F66A F66E

CaM

Ca :2+ - + - + Load

HO2 HO1

HO

2HO

1

62

49

1-29

2

HO2 (54-78) :

HO1 (35-59) :K FLKGN I KKELFKR FQKGQVTRDGFK

*DN

LATTALYFTYSLVMASLYH I YV

1-10 Motifs

α-helixcoil

Structure (HO1) :

by guest on February 16, 2018http://w

ww

.jbc.org/D

ownloaded from

Figure 2 Boehning et. al.

B

A C

Calmodulin (nM)

0.001 0.01 0.1 1 10 100 1000 10000

50

100

150

200

250

EC50

= 2 nM

HO

2 a

ctivity

(pm

ole

s/m

g/m

in)

F66E

WT

[Ca 2+ ] (µM)

0.001 0.01 0.1 1.0 10 100 10000

100

200

300

400

500WT +CaM (100nM)

WT No CaM

EC50 = 287 nM

F66E +CaM (100nM)

HO

2 a

ctivity

(pm

ole

s/m

g/m

in)

Calmidazolium Chloride (µM)

0.001 0.01 0.1 1 10 100 1000

0

100

200

300

400

500

600

IC50

= 8.0µM

HO

2 a

ctivity

(pm

ole

s/m

g/m

in)

D

0

50

100

150

200

250

300

350

HO

2 a

ctivity

(pm

ole

s/m

g/m

in)

CK2 - - + + - - + + CaM - + - + - + - +

10 nM 10 µM

P-Ser - +

CK2

**

*

* *

Ca 2+

:

by guest on February 16, 2018http://w

ww

.jbc.org/D

ownloaded from

Figure 3 Boehning et. al.

A

0

200

400

600

800

1000

Bili

rub

in f

orm

atio

n

(p

mo

les/m

g/m

in)

B

Agonist: none iono iono glu glu 5-HT 5-HT

+CzCl +CzCl +CzCl

*

*

Cortical Neurons

*

Treatment: none iono iono none iono iono +CzCl +CzCl

Mock HO2

none iono iono +CzCl

F66E

noneHO2F6

6E

Bili

rub

in f

orm

atio

n

(p

mo

les/m

g/m

in)

HEK 293

800

0

200

400

600

by guest on February 16, 2018http://w

ww

.jbc.org/D

ownloaded from

Supplementary Table I Boehning et. al.

MUTAGENIC PRIMERS

Construct Forward primer sequence (5’-3’) F56A TTTGTCAAAGACGCCTTGAAAGGAAAC F56E TTTGTCAAAGACGAGTTGAAAGGAAAC L57A GTCAAAGACTTCGCGAAAGGAAACATT L57Q GTCAAAGACTTCCAGAAAGGAAACATT F66A AAGAAGGACCTAGCTAAGCTGGCCACC F66E AAGAAGGACCTAGAGAAGCTGGCCACC Mutations are indicated in bold. Reverse primers were the exact complement of the forward primer

by guest on February 16, 2018http://w

ww

.jbc.org/D

ownloaded from

Supplementary Figure 1 Boehning et. al.

CK2-P

F56

K63

helix 3

helix 3

CK2-P

F56

K63

F76

A

B

by guest on February 16, 2018http://w

ww

.jbc.org/D

ownloaded from

Darren Boehning, Leela Sedaghat, Thomas W. Sedlak and Solomon H. SnyderHeme oxygenase-2 activated by calcium-calmodulin

published online June 2, 2004J. Biol. Chem. 

  10.1074/jbc.C400222200Access the most updated version of this article at doi:

 Alerts:

  When a correction for this article is posted• 

When this article is cited• 

to choose from all of JBC's e-mail alertsClick here

Supplemental material:

  http://www.jbc.org/content/suppl/2004/06/10/C400222200.DC1

by guest on February 16, 2018http://w

ww

.jbc.org/D

ownloaded from