Identification of Myoglobin in Human Smooth Muscle*

Myoglobin (Mb) has been believed to be absent generally from mammalian smooth muscle tissue. Examination of human rectal, uterine, bladder, colon, small intestine, arterial, and venous smooth muscle by immunohistochemical techniques shows that each of these tissues is immunopositive for both smooth muscle myosin and human Mb. Mb-specific primers were used for the polymerase chain reaction to generate cDNA from smooth muscle tissues. Southern hybridization with a Mb-specific probe gave a very strong signal with the cDNA from rectum, weaker signals from small intestine and uterus, a faint signal from colon, and no signal from bladder tissue. High performance liquid chromatography analysis coupled with sequence determination has shown that contaminating heme-binding serum albumin as well as hemoglobin in extracts of smooth muscle seriously compromise any heme-based or spectrophotometric assay of Mb. Combined affinity and size exclusion chromatography, however, provide the necessary resolution. The cDNA-derived amino acid sequence of human smooth muscle Mb was found to be identical to that of Mb from striated muscle.

metric Assay-The procedure of Schuder et al. (13) was used in an attempt to quantify Mb in muscle tissue after removal of Hb by affinity chromatography. The affinity column was made by coupling the ␣␤ dimer of human Hb to CNBr-activated Sepharose 4B. The Hb in the extract binds to the ␣␤ dimer on the column, and Mb passes through unretarded (13). A column (0.9 ϫ 7 cm) of CNBr-activated Sepharose 4B (2 g, Pharmacia Fine Chemicals) was used as described (13). Human Hb (8 mol) 2 was used for the coupling reaction. The capacity of the column was determined by applying pure human Hb and then measuring the absorbance of the eluate at 420 nm. The column retained all of 0.3 mol of Hb applied but 0.7 mol of Hb exceeded the capacity and was detected in the eluate. Therefore, quantities of extract were applied that contained less than 0.3 mol of Hb. The column was regenerated with 2.0 M NaCl, which removes bound Hb. The extraction of rectal skeletal muscle, rectal smooth muscle, and uterine tissue generally followed Schuder et al. (13). Frozen tissue (0.4 -1.0 g) was cut into small pieces and ground under liquid nitrogen. The resulting tissue powder was weighed and extracted with 10 ml of 100 mM potassium phosphate buffer, pH 7.0, 5 mM Na 2 EGTA, incubated on ice for 5 min, and then centrifuged at 10,000 -20,000 ϫ g for 10 min. The clear supernatant Extract I was used for affinity chromatography. The remaining pellet was extracted with 100 mM NaCl, 10 mM Tris-Cl, pH 7.6, 1 mM EDTA, 2% SDS, and 100 g/ml phenylmethylsulfonyl fluoride. This extract was centrifuged as above to give a second supernatant, Extract II. Extract I (4 ml) of rectal smooth muscle was applied to the column at 4°C and developed with 10 mM potassium phosphate buffer, pH 7.0. The flow was downward under gravity, and the flow rate was adjusted to be under 15 ml/h. The fractions containing apparent Mb were collected and pooled. Extract I of rectal skeletal muscle was similarly processed. The spectra of both deoxy-Mb and CO-Mb were measured as described (13). The millimolar extinction coefficients were taken to be ⑀ mM ϭ 207 mM Ϫ1 cm Ϫ1 at 424 nm (MbCO) and ⑀ mM ϭ 121 mM Ϫ1 cm Ϫ1 at 435 nm (deoxyMb), the values given for horse Mb (14). We measured the absorbance ratio, A 424 /A 435 ϭ 0.820 for horse deoxy-Mb and then calculated the ⑀ mM value at 424 nm: 0.820 ϫ 121 ϭ 99.2 mM Ϫ1 cm Ϫ1 . The apparent concentration of Mb was calculated from the following equation.
where the ⑀ mM values are for 424nm. All HPLC was performed on a SynChropak RP-P C 18 reverse phase column (250 ϫ 4.6 mm, Syn-Chrom, Inc., Lafayette, IN) driven by a Beckman model 332 gradient chromatography system. Extract I of rectal tissue and the eluate from the affinity column (see above) were applied separately to the HPLC column with the following gradient program: Buffer A 0.1% trifluoroacetic acid in water, Buffer B, 0.1% trifluoroacetic acid in acetonitrile, 0 -5 min, 0% B; 5-15 min, 0 -30% B; 15-115 min, 30 -55% B; 115-125 min, 55-0% B. Absorbance of the eluate was monitored at 220 nm with an Hitachi model 100 -10 spectrophotometer. Absorbing fractions were collected and lyophilized. The protein was redissolved in water and subjected to SDS-polyacrylamide gel electrophoresis (PAGE) (10% resolving gel and 3% stacking gel) (15). Size Exclusion Chromatography-The separation of Mb from serum albumin in Extract I (after affinity chromatography) was examined with a Bio-Gel SEC 40XL size exclusion column (300 ϫ 7.8 mm; Bio-Rad catalog number 125-0604). The buffer used was 100 mM KPO 4 , 1 mM Na 2 EDTA at pH 7.0 running at 25°C. The column eluate was monitored at either 280 or 415 nm with an Hitachi spectrophotometer (model 100 -10) with an HPLC system previously described (16). A reference mixture of bovine serum albumin (Sigma catalog number A-8531) and horse Mb (Sigma catalog number M-0630) was also examined in this way.
Western Blot-Proteins in both Extract I and Extract II of rectal smooth muscle tissue were run on SDS-PAGE. The gel was electroblotted to Protran TM Nitrocellulose membrane (BA79, 0.1-m pore size, Schleicher & Schuell). The transferring buffer contained 10 mM CAPS at pH 11 and 10% methanol. The chemiluminescent ECL TM Western blotting System (Amersham Pharmacia Biotech) with the manufacturer's protocol was used for the detection of immobilized specific antigens conjugated with horseradish peroxidase-labeled antibodies. The primary antibody was rabbit anti-human Mb (Sigma catalog number M-8648, diluted 1:500). The secondary antibody was horseradish per-oxidase-labeled donkey anti-rabbit whole antibody (Amersham Pharmacia Biotech catalog number NA934, diluted 1:1500).
Immunohistochemical Assay-Biopsy samples of human tissues were trimmed of connective tissue and pinned to balsa wood sticks between two muscles from the hind limb of an adult rat (soleus and extensor digitorum longus). These tissues were then frozen in isopentane cooled to Ϫ60°C with liquid nitrogen. Cryostat sections (10 m) were collected on gelatin-subbed slides and stored dessicated at Ϫ20°C. For immunolabeling, slides were incubated for 30 min at room temperature in phosphate-buffered saline (PBS, 137 mM NaCl, 2.7 mM KCl, 10 mM Na 2 HPO 4 , 1.8 mM KH 2 PO 4 , pH 7.4) containing 0.2% BSA, incubated in primary antibody diluted in PBS ϩ BSA overnight at 4°C, washed three times, 5 min each with PBS ϩ BSA, incubated in secondary antibody diluted in PBS ϩ BSA for 1 h at room temperature, washed three times, 5 min each with PBS, and coverslipped using FITC Guard (Testog, Chicago, IL). Primary antibodies were rabbit anti-human Mb (Sigma catalog number M-8648, diluted 1:500); monoclonal mouse antismooth muscle myosin (Sigma catalog number M-7786, diluted 1:250); monoclonal anti-human sarcomeric myosin heavy chain (clone A4.1025, Developmental Studies Hybridoma Bank, Johns Hopkins University, used as undiluted supernatant). Secondary antibodies were FITC-conjugated sheep anti-mouse Ig-G, F(abЈ) 2 fragment (Sigma catalog number F-2266, diluted 1:200) and rhodamine-conjugated goat anti-rabbit IgG, F(abЈ) 2 fragment (Cappel, Durham, NC, catalog number 55671, diluted 1:200). In some cases double immunolabeling was performed by applying a mixture of mouse anti-smooth muscle myosin and rabbit anti-Mb antibodies followed by a mixture of FITC-conjugated antimouse and rhodamine-conjugated anti-rabbit secondary antibodies. Slides were examined with a Nikon Optiphot epifluorescence microscope equipped with rhodamine and fluorescein filters. Images were acquired with an integrating CCD camera connected to a Macintosh computer containing a frame grabber and NIH image software. Images were cropped and labeled by use of Adobe Photoshop and Canvas software and printed with a Sony color printer.

Occurrence of Mb mRNA in Different Smooth Muscle
Tissues-cDNA encoding human Mb was amplified by PCR from the total cDNA prepared from different tissues. Specificity was achieved by using primers constructed on the basis of the gene-derived NH 2 -and COOH-terminal sequences (12). The same mass of tissue (ϳ0.4 g) was used from each source for Southern hybridization (Fig. 1) so that the signal should reflect approximately the relative quantity of Mb mRNA in each tissue. The strongest signal was clearly from rectal smooth muscle tissue with much less from tissue of the small intestine and uterus. Only a very weak signal was obtained from colon tissue, and none was detected from bladder tissue.
Northern hybridization of mRNA from diverse tissues probed with cDNA for human Mb gave signals only for skeletal muscle and heart mRNA ( Fig. 2). Although no signal was detected in the other tissues, the Northern hybridization is much less sensitive than the Southern hybridization prepared with PCRamplified cDNA, so the negative finding could mean only that an insufficient quantity of mRNA was present in the experiment.
Immunohistochemistry of Smooth Muscle Tissue-We examined tissues with smooth muscle from rectum, uterus, colon, small intestine, bladder, arteries, and veins with immunohistochemical techniques. Differently labeled fluorescent antibodies to both human smooth muscle myosin and Mb were used. The results reveal that human rectal muscle is immunopositive for both smooth muscle myosin and for Mb (Fig. 3a). A cryostat section passing longitudinally (panels A and B) and two sections passing orthogonally (panels C and D and panels E and F) through a bundle of smooth muscle fibers were double-immunolabeled; immunoreactivity to smooth muscle myosin was detected by using an FITC-labeled second antibody and to Mb by using a rhodamine-conjugated second antibody. Views through the fluorescein filter set are shown in panels A, C, and E, and views through the rhodamine filter set are shown in panels B, D, and F (Fig. 3a). The tissue sections in panels A, B, C, and D ( Fig. 3a) received a mixture of anti-smooth muscle myosin and anti-Mb antibodies, followed by a mixture of the two secondary antibodies. The sections in panels E and F received no primary antibody but were incubated with the mixture of the two secondary antibodies. Thus, the sections in panels E and F serve as negative controls. Although some labeling of connective tissue is obvious in the absence of primary antibody, the muscle fibers themselves are unlabeled. Additional controls (also not shown) showed that these smooth muscle fibers were unreactive with an antibody to sarcomeric myosin and that rat skeletal muscles were unreactive with the anti-smooth muscle myosin. Very similar results were obtained for artery and vein tissue (Fig. 3b), for uterine tissue (Fig. 3c), and for colon, small intestine, and bladder (data not shown).
Amino Acid Sequence of Smooth Muscle Mb-The PCR-amplified cDNA for Mb from human rectal smooth muscle was cloned into pUC19 and sequenced. The deduced amino acid sequence was found to be identical to that of Mb from striated muscle (12). The same tissue preparation also yielded PCRamplified cDNA encoding part of the heavy chain of smooth muscle myosin, thus confirming the identification of the tissue.
Spectrophotometric Quantification of Mb after Hb Removal-The principle of the procedure (see "Materials and Methods") is to use an affinity column that selectively removes Hb but not Mb. Extract I (4 ml) of rectal smooth muscle containing less than 0.05 mol of Hb was applied to the column. The amount of Hb in the aliquot of Extract I was much less than the capacity of the affinity column, which can bind at least 0.3 mol of Hb. Fig. 4 shows the HPLC pattern obtained for Extract I before ( Fig. 4a) and after (Fig. 4b) passage through the column. The ␣ and ␤ chains of Hb (about half of the total protein) have clearly been removed by the column.
Quantitative correspondence of the HPLC absorbance with the spectrophotometric determination can be determined as follows with the use of hemin and protein extinction coefficients previously determined (17). The quantity of human Hb in the original solution, determined from the ␣ and ␤ chain absorbances in Fig. 4a, was found to be 41.8 nmol. The total heme (measured as hemin) in Fig. 4a corresponds to 73.4 nmol in the original solution. The difference (73.4 Ϫ 41.8) should give the heme attributed to Mb, 31.6 nmol, which corresponds to 174 mol of Mb/kg tissue. Although this value is within 8% of the quantity of Mb, 188 mol, determined by difference spectrophotometry of Extract I after affinity chromatography, SDS electrophoresis of fraction 4 in Fig. 4b showed that it is largely composed of a 66-kDa polypeptide, and only traces of 17-kDa protein are present (data not shown). Fraction 4 was electroblotted onto a polyvinylidene difluoride membrane for sequencing by the Microanalysis Facility of the University of Texas with a model 477A Applied Biosystems Sequencer. The 20residue sequence obtained showed that the 66-kDa protein is serum albumin. Therefore, most of the apparent Mb in the spectrophotometric assay must be attributed to heme-binding serum albumin and not Mb.
Heme-binding by serum albumin was tested with the Schuder et al. (13) spectrophotometric assay for Mb content as follows. Hemin (Fluka, 1 l of 12.4 mM stock in 0.1 N NaOH) was added to 4 ml of 10 mM phosphate buffer, pH 7, containing 8 nmol of BSA (Sigma) to give a 1.5 molar ratio of hemin to albumin. An apparent Mb content, 0.66 M, was determined spectrophotometrically as described by Schuder et al. (13) (see "Materials and Methods").
In contrast to the smooth muscle analysis, Extract I of rectal skeletal muscle did not show a large quantity of serum albumin. Analysis of fraction 7 in Fig. 4c showed that it consists of Mb and that serum albumin was absent (data not shown).
Analysis of Extracts I and II of rectal smooth muscle tissue after SDS-PAGE showed that the two extracts have different protein band patterns (Fig. 5a). Additional cellular proteins were obtained in Extract II. When Extract I and Extract II were probed with anti-Mb antibody on a Western blot, a single band of about 17 kDa was detected in both Extracts I and II (Fig. 5b). We conclude that a low concentration of Mb is present in both Extracts I and II and that Extract I did not extract all the protein.
Size Exclusion Chromatography-Application of a mixture of BSA (66 kDa), horse Mb (17 kDa), and human Hb (65 kDa) to a Bio-Rad size exclusion column failed to isolate the Mb because of the partial dissociation of the human Hb tetramers to dimers, which caused the Mb and Hb peaks to overlap (data not shown). However, separation of horse Mb from BSA alone did occur (Fig. 6a). When Extract I, after removal of Hb with the affinity column (see above), was applied to the SEC column (Fig. 6b), a small shoulder on the human serum albumin peak corresponded in position to Mb. The amount of protein in this shoulder is roughly estimated to be about 10% of the serum albumin. Comparison of the quantity of serum albumin shown in Fig. 6 allows us to estimate the quantity of Mb to be 0.16 mg/g wet tissue, a rough estimate at best because we are at the limit of detectability for the Mb. This difficulty does not compromise the estimation of Mb in skeletal muscle tissue because the latter contains a much larger quantity of Mb. DISCUSSION Our results confirm the early report by Lankester (1870) that Mb occurs in human rectal smooth muscle (2,3), but only at very low concentrations. We show by immunohistological techniques that Mb is also present in all other smooth muscle tissues examined: colon, small intestine, uterus, bladder, ar-teries, and veins. Table I summarizes the presence and content of Mb in various tissues. However, attempts to quantify the Mb in the rectal tissue with existing protocols failed because they depended on the assumption that the only interfering hemecontaining protein is Hb (13). The procedure of de Duve (20), which relies entirely on determination of hemin as the pyridine hemochromogen derivative, would also fail to measure Mb accurately. The reason for these failures is that substantial quantities of hemin-binding serum albumin can be present. The protocols do not distinguish between heme or hemin from Mb and from serum albumin. The procedure of Schuder et al. (13) adequately removes contaminating Hb by affinity chromatography but does not remove serum albumin. Therefore affinity separation followed by size exclusion chromatography (Fig. 6) should be the method of choice to separate both of these contaminants from the much smaller Mb. Size exclusion chromatography alone is less satisfactory because of the tetramer to dimer dissociation of Hb. FIG. 4. Comparison of reverse phase HPLC patterns. a, rectal smooth muscle Extract I before an affinity column shows ␣ and ␤ chains of Hb. Analysis of fraction 1 shows it to be largely serum albumin (see "Spectrophotometric Quantification of Mb after Hb removal"). b, rectal smooth muscle Extract I after passage through an affinity column reveals that ␣ and ␤ chains of Hb have been removed. c, rectal skeletal muscle Extract I after passage through an affinity column. Southern hybridization with cDNA for human Mb gave a small but clear signal with tissue of the small intestine and uterus, a very weak signal from the colon, and none for the bladder, whereas the rectal smooth muscle gave a very strong signal. Several possible explanations for the large differences in mRNA may be suggested. Mb might be expressed only in special, as yet unspecified, metabolic circumstances. Alternatively, a small subset of cells in the rectal smooth muscle tissue might have a substantial content of Mb, but the analysis of a relatively large mass of tissue would provide only a diluted estimate of the Mb. The latter possibility is enhanced by studies of another tissue, the esophageal sphincter of the opossum, which has been extensively used as a model for the esophageal sphincter in man (21)(22)(23)(24). This sphincter may be similar to the rectal smooth muscle sphincter of man. The esophageal sphincter tissue of the opossum has several properties that distinguish it from the smooth muscle of the nearby esophageal body: 1) The rate of oxygen consumption of the sphincter tissue is higher than the esophageal body (25). 2) The tonic contraction of the sphincter is entirely aerobic and cannot be maintained anaerobically. In contrast, the esophageal body contractions can be partially maintained anaerobically (22). 3) The apparent mitochondrial profile area is larger in the sphincter cells than in the esophageal body (26). 4) Lactic dehydrogenase type I isozyme is present in the sphincter but not in the esophageal body (27). These observations show that the aerobic metabolism of the sphincter tissue is unique. Under aerobic conditions the muscle is tonically contracted and is only briefly relaxed to allow swallowing. The aerobic demand would make the presence of Mb advantageous. No similar metabolic studies of rectal tissue appear to have been made. The presence of mRNA for Mb in small intestine but not colon may be correlated with the different physiological functions of the tissues. Perhaps even a small amount of Mb may be advantageous in small intestine tissue but not colon. Peristaltic waves in the small intestine move at 1-2 cm/s over long periods of time, and the villi, present at a density of 10 -40/mm 2 , have smooth muscle and contract independently almost continuously every 10 s (28). In contrast, the colon is devoid of villi, and muscular activity is infrequent.
Our finding of some Mb in smooth muscle of arteries and veins suggests that it may play a role in limiting the time of vasodilation induced by nitric oxide delivered by S-nitrosohemoglobin (11). NO reacts rapidly with MbO 2 to form NO Ϫ and MetMb (29,30), so that Mb should shorten the lifetime of NO action in vasodilation. The lifetime of NO in Mb-rich heart muscle is known to be short (31). The vasodilation produced by NO depends on activation via cGMP of calcium-dependent potassium channels that cause hyperpolarization (32). The presence of Mb should shorten the time of hyperpolarization. Lancaster (33) has suggested that the action of NO may be restricted by Mb to localized areas of heart and skeletal muscle, a newly identified function for Mb. Similar localization may also occur in smooth muscle tissue. Mb should protect against the toxic effects of high NO production associated with tissue damage by accelerating its destruction. The proposed role of Mb in NO metabolism requires that the MetMb product be recycled. Ascorbate-mediated redox cycling of Mb has been shown to be capable of serving this function (34). Although the ascorbate content of human smooth muscle is not known, it is accumulated in many tissues and can reach concentrations of 300 -800 M in heart muscle and ϳ200 M in skeletal muscle (35).  ples and Wesley Thompson for valuable discussions. We thank James Christensen for valuable discussions regarding the opossum. We thank Thomas L. Vandergon for DNA from Cerebratulus lacteus. We thank Claire Riggs for critical reading.