Human Eccrine Sweat Contains Tissue Kallikrein and Kininase II

Toshihiko Hibino, Toshiyuki Takemura, and Kenzo Sato Marshall Dermatology Research Laboratories, Department of Dermatology, Universiry of Iowa Co llege of Medicine, Iowa Ciry, Iowa, U.S.A.

We attempted to determine the level of sweat kallikrein kallikrein is the glandular type. Purified sweat and salivary (kininogenase) and to purify and characterize it using sweat kallikrein showed similar Mr and responses to inhibitors and collected over a white petrolatum barrier. Thermally induced antibodies. Using immunohistochemistry, kallikrein activity eccrine sweat obtained from 24 healthy subjects showed kal­ was localized in luminal ductal cells and in the peripheral rirn likrein activity of 24.4 ng kinins generated/1 mg of sweat of secretory coil segments, presumably the outer membrane protein when heated plasma was used as the substrate and of the myoepithelium. We also observed kininase activity in 16.1 ng kinin when purified low molecular weight bovine sweat at Mr 160,000, which was inhibited by ethylenedia­ kininogen was used as the substrate. Sweat was sequentially mine tetraacetic acid, captopril, and angiotensin converting purified by Sephacryl S-200, diethyaminoethyl Sephacel, enzyme inhibitor peptide, indicating that it is kininase II (or and fast flow liquid chromatography Mono Q chromatogra­ angiotensin converting enzyme). Sweat also contains abun­ phy. Sweat kallikrein had aMrof 40,000 and was inhibited by dant non-kallikrein hydrolases for S-2266 and S-2302. The aprotinin but not by soybean trypsin inhibitor. The peptide demonstration of glandular kallikrein, its tissue localization, generated by sweat kallikrein was identified as lys-bradykinin and the presence of kininase II in sweat provide the basis for using reverse phase high-performance liquid chromatogra­ future studies on the physiologic role of the kallikrein/kinin phy and by its amino acid sequence. Anti-human urinary system in the eccrine sweat gland. Key words: ki,tinoge1lase/ kallikrein immunoglobulin G neutralized the sweat kal­ sweat gland/protease. ] Invest Dermatol1 02:214 - 220, 1994 likrein activity completely, indicating that the sweat

he eccrine sweat gland shares a number of features the presence of a kallikrein/kinin system in the eccrine sweat gland with other exocrine glands, especially with regard to is hard to dismiss. Fox and Hilton [16] first reported in 1958 that the mechanisms involved in the secretion of fluids, eccrine sweat and peri glandular dermal fluid contained a substance electrolytes, and nonelectrolytes [1 ,2]. Both the sali­ that contracted the rat uterus when such fluids were preincubated vary gland and saliva contain high concentrations of with pseudo-globulin. They interpreted the finding as indicating Tglandular kallikrein, a serine protease [3]. Kallikreins (or kinino­ that bradykinin was produced from pseudo-globulin by kallikrein in genases) are also present in a variety of other tissues and body fluids sweat or dermal fluid. Frewin et al [17] also observed weak kinin­ [4-8]. It has been postulated that glandular kallikrein serves many generating activity in sweat samples from some, but not all, subjects functions including the regulation of salivary glandular blood flow during exercise. Fraki et al [18] showed benzoyl-L-arginine ethyl through kinin generation [9], proteolytic processing of various ester (BAEE) - hydrolyzing, aprotinin-sensitive proteinase in growth factors and polypeptide hormones [10], and the regulation human sauna sweat, which suggests that sweat may contain kal­ of membrane transport also via kinin generation [11-13]. likrein-like enzymes. During the course of the present study, May­ Thermoregulatory eccrine sweating is associated with active re­ field et al [19] also observed that human sweat contains immunore_ flex vasodilation (ARV) [14]. In patients with hereditary anhidrotic active glandular kallikrein with kinin-generating activity by using ectodermal dysplasia where sweat glands are congenitally absent, radioimmunoassays for kallikrein and for generated kinin. Never­ ARV does not occur during heat exposure despite the fact that their theless, the kallikrein activity was absent or near the detection limit sympathetic innervation to cutaneous blood vessels appears to be of the ~ssay ~ystem. in some unconcentrated sweat samples. Further­ normal [15]. Whether or not the ARV during heat-induced sweat­ more, 111 their studies, sweat samples were not collected over a White ing is due to bradykinin produced by the sweat gland and the lack of petrolatum barrier and the sodium concentration in some of their ARV in anhidrotic ectodermal patients is due to the absence oflocal sweat samples was abnormally high [19]. Thus the possibility of bradykinin production are unknown. Nevertheless, evidence for epidermal contamination and evaporative water loss in some of their sweat samples cannot be entirely ruled out. The goals of the present study have been 1) to confirm the pres­ ence of kallikrein activity in clean sweat, 2) to determine whether Manuscript received March 19, 1993; accepted for publication September hydrolysis of synthetic substrates reflects the kallikrein activity in 2,1993. Reprint requests to: Dr. Kenzo Sato, Department of Dermatology, Uni­ sweat, 3) to determine that sweat kallikrein is of the glandular (or versiry ofIowa College of Medicine, 271 Medical Laboratories, Iowa Ciry, tissue) type, but not the plasma type, 4) to study whether or not IA 52242-1181. kinin-degrading enzyme(s) (i.e., kininase II) is also present in sweat, Abbreviations: ARV, active reflex vasodilation; HMW, high molecular and 5) to study the localization of kallikrein in the sweat gland using weight; HP, heated plasma; LMW, low molecular weight. immunohistochemistry .

0022-202X/94/S07.00 Copyright © 1994 by The Sociery for Investigative Dermatology, Inc. 214 VOL. 102, NO.2 FEBRUARY 1994 KALLIKREIN IN SWEAT 215

I: Q) 60 ~ 0 (TOc) ~ A n=24 ~52.5 ~ :!i ...I~ t 45 c 40 Ea;... _!:?C5 0 0 co. .210 -Q) 20 B -Sll)u~ 0 ~ o~ ... D..e Y=0.68X + 0.68 m::::" R=0.80, P<0.001 ~E .~~ 0 .- c .~~ n=5 ~~ 0 10 20 30 40 50 60 70 Kinin production from H.P. 4 B n=14 0 5-2266 A 5-2302 .... 1: 3 0 ft:l~:i! A ~ ~ LI--.----r---r--~--_,,_--r_--,_--_. ~S 0 r I OA 15 20 0 30 40 50 min 2 AO AO 0 80 II) • 0 C GI 0 ~ .. OA .. o fromH.P. ...ra i n=24 0 • from LMW K-gon - II 60 o iii .. • .Q ~ :::l~ .. f o (1).5 uCD 40 .:'0 0 CD-- • • .co. 0 10 20 30 40 50 60 70 0 o 0 ~ AI 0 Cra Kinin production from H.P. 20 0 -->-CD OJ. 0 II)~ 4 t • all) 0 5-2266 • • ~ • C n=14 0 rn~ A 5-2302 0 200 400 600 800 1000 Ui.E .?;os 3 0 Total sweat volume (mI/50minlm2) f:?o A "tie ~s 0 Figure 1. Kininogenase and kininase activity in thermally induced sweat 2 ~ 0 A from the back. The solid line in A is the temperature protocol used through­ 0 0 out the present sauna experiments. The mean sweat rate was derived from 0 ~ ~ ~ [20]. B) Time course of kallikrein (solid symbols) and kininase (open symbols) 0 A activity in sweat, expressed as ng kinin production from LMW kininogen or .. ng bradykinin degradation per ml of sweat/h, respectively. C) Kinin pro­ ~ A duction versus total sweat output (m1j50 min/m2 on the back) in 24 men, ages 18-30. tI, number of subjects. Open circles, kinin production from HP 0 and solid triallgles from bovine LMW kininogen. Kinin production was ex­ 0 10 20 30 40 50 60 pressed on a per-sweat protein basis (which is similar to that expressed on the Kinin production from LMW K-gen sweat volume basis because sweat protein content is similar in different sweat (ng/h/mg sweat protein) samples; not shown). Approximately 24.4 ± 15.4 ng of kinin was generated by 1 mg of sweat protein for the HP (used as substrate) and 16.1 ± 13.6 ng for bovine LMW kininogen (used as substrate), which correspond to 0.18 ng Figure 2. Kininogenase activity as determined using natural subrates versus and 0.16 ng of kinins generated by 1 ml of original sweat, respectively. synthetic substrates. A) Comparison of kinin generation from bovine LMW When EDTA and 1,10 phenanthroline were not included in the assay mix­ kininogen and human HP. B) Comparison of kinin generation from HP and ture for kinin, detection of kinin was dramatically decreased (data not hydrolysis of S-2266 (widely used substrate for glandular kallikrein) and shown), consistent with the observation that kininase is also present in 5-2302 (substrate for plasma kallikrein). C} Comparison of kinin generation human eccrine sweat (see B) . from bovine LMW kininogen and hydrolysis ofS-2266 and S-2302. Sweat samples from each subject were pooled and concentrated. The correlation between the hydrolysis ofS-2266 and that ofS-2302 is R = 0.90, P < 0.001, Y = 0.95X -0.12, not shown. MATERIALS AND METHODS Materials Trizma base, phenylmethyl sulfonyl fluoride (PMSF), diiso­ propylfluorophosphatc (DFP), l,lO-phenanthroline, trypsin (type III), pa­ ular weight (LMW) kininogen was purchased from Seikagaku Kogyou, pain (type III), benzoyl-arginine-p-nitroanilide (BANA), aprotinin, soybean Tokyo, Japan. Purified bovine pancreatic kallikrein was a gift of Showa trypsin inhibitor (SBTI), bradykinin, Iys-bradykinin (kallidin), angiotensin Denko Co. Ltd., Tokyo, Japan. StrAvigen was purchased from Biogenex converting enzyme (ACE) inhibitor (Gln-Trp-Pro-Arg-Pro-Gln-I1e-Pro­ Laboratories, San Ramon, CA. Rabbit antiserum to human urinary ka.Ilikrein Pro), and angiotensin I were obtained from Sigma Chemicals, St. Louis, was obtained from Japan Chemical Research, Kobe,Japan. Norit A was the MO. Ethylenediamine tetraacetic acid (EDTA) was purchased from MCB product of Fisher Scientific, Fair Lawn, NJ. USP Vaseline was used as white Manufacturing Chemicals, Cincinnati, OH. Sephacryl S-200, diethylamin­ petrolatum. All other reagents were purchased from Sigma. oethyl (DEAE) Sepharose, cyanogenbrornide (CNBr)-activated Sepharose 4B, Dextran T 70, and prepackaged Mono Q column (HR 5/5) were pur­ Sweat Collection To obtain thermally induced sweat samples with a chased from Pharmacia (Uppsala, Sweden). VyDAC reverse-phase col­ minimum of epidermal contamination and evaporative water loss, sweat was umn C\8 was obtained from Bio-Rad, Hercules, CA. Val-Leu-Arg-pNA collected inside a plastic film or bag over a white petrolatum barrier placed (S-2266), and Pro-Phe-Arg-pNA (S-2302) were purchased from Kabi, on the skin [20,21]. For the time-course study (Fig 1), sweat was collected Stockholm, Sweden. The enzyme immunoassay kit for kinins was obtained from a rectangular area, 30 X 35 cm2 , on the upper back. The room temper­ from Dainippon Pharmaceuticals (Suita,Japan). Purified bovine low molec- ature was increased stepwise from 45 to 52.5 °C (see Fig 1, top) over the 216 HIBINO ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

C- cording to the method of Ueno et al [23], using an assay kit (Markit A o Bradykinin). 20 ;: u E 1.6 A co Synthetic Substrate Assay Kallikrein-like activity was also measured c:_ ~ 15 using the synthetic peptide substrates S-2266 and S-2302. Fifty- to one hundred-microliter samples (sweat or eluates from columns) were incubated ~ \ 1.2 A LMW K-gen fnc: in test tubes containing 400 III of 50 mM Tris-HCl (pH 8.5) and 50,u1 of 4 mM S-2266 or S-2302 at 37"C for 10-30 min. The reaction was termi­ co- • H.P. 10 c: G) - nated by adding 50 III acetic acid. The absorbance at 405 nm was deter­ 0.8 0 u ;: c: co mined. Enzyme activity was expressed as nmol/min/ml of p-nitroaniline co ~ liberated, assuming E 0S = 10,500 [24]. .0 5 G) 4 ~ 0.4 o :e III Kininase Assays Kininase activity was expressed as the rate of degradation .0 .5 of exogenous bradykinin by samples in the absence of EDTA or 1,10-phen_ < o Ir---ra~~,--,--~~--~ 0 c: anthroline. The reaction mixture consisted of 20 III of sample (typically 0.2 ~ containing around 0.1 ng of bradykinin), 40 III of anti pain mixture (20 B mg/ml of antipain in 0.4 M hydroxyethylpiperazine ethanesulfonic acid 05-2266 [HEPES], pH 8.0) + 0.4 M NaCI), and 20 III of bradykinin (400 ng/mI). After incubation at 37 · C for 30 min, the reaction was stopped by the addi_ • 5-2302 tion of 20 III TCA. The residual kinin was measured by the enzyme-linked 0.1 immunoassay. Angiotensin I Degradation The fifty-microliter sweat sample and 25-pl buffer (0.4 M HEPES + 0.4 M NaCl, pH 8.0) were incubated with 25,u1 of angiotensin 1(20 mg/ml) for 3 hat 37"C. After boiling the reaction mixture o for 5 min, anti-angiotensin I serum (200 Ill) and !2SI-angiotensin I (50 pi, 5000 cpm) were added and incubated for 24 h at 4· C. The antibody-bound 20 C angiotensin I was removed by adding 200 III of dextran-coated charcoal Bradykinin (1.25% Norit A and 0.25% Dextran T 70) and the radioactivity of free angiotensin I in the supernatant determined using a Beckman gamma Angiotensin 1 counter. 10 Purification of Sweat Kallikrein Concentrated pooled sweat samples (5-7.5 ml of 100X) obtained from 15 subjects were applied to a Sephacryl S-200 gel column equilibrated with 20 mM phosphate buffer (pH 7.2) con­ taining 0.14 M NaCI. The kallikrein-rich fractions were pooled (bar in Fig o~~~~~~~~--~ 3) and applied to a DEAE-Sepharose column equilibrated with 20 ruM 40 50 60 70 80 90 phosphate buffer (pH 7.0) . Elution was performed by a linear gradient of NaCI (see Fig 3). The pooled kallikrein-rich fractions (bar in Fig 4) were Fraction number further purified by fast flow chromatography (FPLC) using an anion-ex_ change (Mono Q) column equilibrated with 20 mM Tris-HCl (pH 8.0). Figure 3. Sephacryl S-200 gel chromatography of concentrated clean sweat. The column, 3.2 X 80 em, was eluted with phosphate-buffered sa­ Purification of Salivary Kallikrein Salivary kallikrein was also puri­ line and 5.5-m1 fractions collected. A) Kinin generation from bovine LMW fied from human mixed saliva by acetone precipitation followed by sequen_ kininogen (opell triangles) and that from heated plasma (solid triallgles) . M,. for tial column chromatography as described above. The final kallikrein frac- the peak is around 40,000. Kallikrein-rich fractions indicated with a bar were pooled and further purified. B) Hydrolysis of S-2266 (open circles) and S-2302 (solid circles) by the same fractions as in (A). Note that the peaks of hydrolytic activiry are around M, 50,000. C) Bradykinin-degrading activity (solid squares) and angiotensin I-degrading activity (open squares) in the same .6 A 16~ fractions as in (A) . T - j :s E c 12~ c .4 .6 , 00 oS -, 50-min period while the relative humidity was kept at 70%. Sweat was N , collected continuously by suction from the reservoir of the bag, pooled every 0; 8~ ...... 5 min, and placed on ice. Collected sweat was filtered through a Millipore l!! ~ flc HA membrane (0 .22 mm) and concentrated about 10 times using an Amicon .. .2 11 .2 .E t::' YM 10 membrane (the exact concentration factor was calculated for each -eo 4~ U sweat sample) at 4·C and stored at -70·C until use . Pooled sweat samples f/) 'c .. .c ~ used for the chromatographic purification of kallikrein and kininase were S2 I collected in the same fashion but from the entire upper trunk and with the '" ,. 0 0 trunk and arms (painted with Vaseline) tightly occluded with a large plastic B bag [21]. Pooled sweat samples were concentrated about 100 times by filtra­ tion with an Amicon YM 10. The absence of epidermal contamination was • S-2266 confirmed by the absence of aminopeptidase activity (which is derived al­ o S-2302 most exclusively from the epidermis) in sweat samples [21].

Enzyme-Linked Immunoassay for Kinins Human citrated plasma that contains both LMW and high molecular weight (HMW) kininogens was heated at60· Cfor30 minin the presenceof4 mM 1,10-phenanthroline o II .,I;); r r=I/--',..,..-,-,..-';P¥~~ to inactivate plasma kininases. It was used as the substrate for determination 10 20 30 70 80 90 100 of the total (i .e., plasma and glandular) kallikrein activity. Purified bovine Fraction number LMW kininogen was used to determine glandular kallikrein activity. Twenty microliters of concentrated sweat and 20 III of buffer (100 mM Figure 4. DEAE-Sepharose chromatography of the kallikrein-rich fraction Tris-HCI, pH 8.5, + 5 mM EDTA + 5 mM 1,10-phenanthroline) were from Fig 3. A) Kinin liberation from LMW kininogen. B) Synthetic sub­ incubated with 20 III of heated plasma (HP) or LMW bovine kininogen (0.5 strate hydrolysis. Note that the majority of S-2266 and S-2302 hydrolytic mg/ml) at 37· C for 60 min. The reaction was stopped by the addition of activities were separated from the peak kallikrein activity. Fraction volume 20 III of trichloroacetic acid (TCA). Generated kinins were determined ac- was 2.8 mi. VOL. 102, NO.2 FEBRUARY 1994 KALLIKREIN IN SWEAT 217

RESULTS A 20 Because the kallikrein (or lcininogenase) levels in some unconcen­ .. trated sweat samples were near or below the detection limit of the T .2 .4 E ~.c assay system [17,19]' we chose to concentrate all of our sweat sam­ c <:> ples by ultrafiltration. The loss of enzyme activity during ultrafil­ = tration may be minimal, if any, and to similar degrees in different N t <;j .s :;: samples. In fact, kallikrein was not recovered from the freshly used Q) 10 -c c g .1 ~ .2 ff YM 10 membranes by exhaustive elusion (unpublished) and the III ~ III .t:I Q) mean kallikrein activity (Fig IB) was rather comparable to those 15 g ~ reported by Mayfield et af [19], who used unconcentrated sweat c «.t:I'" 'c samples. Figure 1B (solid symbols) shows that the mean kallikrein S2 activity remained constant with time of sweating, suggesting that 0 0 kallikre.in was not an epidermal contaminant (because epidermal contammants tend to show higher concentrations in the first few sweat samples than in later samples) [20] but was secreted by the sweat gl~nd on a continuous basis. Also shown in Fig 1B (open :ymbols) IS ~hat s:veat also contains kininase (kinin-degrading) activ­ Ity. As depicted m Fig 1 C, when kallikrein activities were compared with total sweat volumes (collected from the upper trunk over a 50-min period as an indirect measure of one's maximal sweat rate), the correlation was absent, indicating that the kallikrein level in Figure 5. FPLC Mono Q chromatography of the peak fraction from Fig 4. sweat is not determined by the factors that influence the sweating Note that the S-2266 hydrolytic activity (B) now eluted with the peak rate, e.g., degree of fitness or acclimatization. kallikrein activity (A). Fraction volume was 1 ml. S-2302 hydrolase activity We then addressed whether or not the synthetic substrates was undetectable (not shown). The specific activity of kallikrein increased 300 times, with a yield of 9%. S-2266 and S-2302, widely used for the determination of glandular and plasma kallikrein activities [26-28], respectively, are feasible for the determination of kallikrein activity in sweat. If hydrolysis of these synthetic substrates by sweat enzymes reflects kallikrein activ­ tions showed two protein bands at M, 38,000 and 41,000 on the sodium ities in sweat, then the superior sensitivity and simplicity of these dodecyl sul fate (SOS) polyacrylamide gel. Both bands reacted with anti-ur­ substrates would make them extremely useful for further physio­ inary kallikrein IgG on immunoblotting. When enzyme-linked immunoas­ logic studies of sweat kallikrein. Thus we determined kinin genera­ say for kinins was performed, 1 ng of purified salivary kallikrein liberated tion from both HP (which contains both HMW and LMW kinino­ 36 ng of kinins/h. This preparation was also used as a positive control and a gens; glandular kallikrein hydrolyzes both kininogens but plasma standard in the sweat kallikrein assays. kallikrein works only on HMW kininogen) and bovine LMW kin-

Purification of Human HMW Kininogen Human HMW kininogen was partially purified as previously described [25] . The amount of kininogen was measured by papain inhibition using benzoyl-arginine-naphthylamide (BANA) as a substrate [25]. The final preparation liberated 0.5 mg of kinin/ 100 mg protein/h when incubated with an excess amount of human urinary ..... : Aprotinin kallikrein. c: ~DO: SBTI 0 Determination of Kinin Species Fifty microliters of the sweat kal­ ;; likrein fraction from the Mono Q chromatography was mixed with 25 III of ...CU human HMW kininogen and incubated for 60 min at 37 °C. After the CD addition of 25 ml TCA, the supernatant was lyophilized and redissolved in :e 20 ml of 10% acetonitrile containing 0.1% trilluoroacetic acid. Samples were isolated by reverse-phase high-performance liquid chromatography c: (HPLC) using a VyOAC CIS column. The sequence of the isolated peptide c: was determined using an Applied Biosystem 470 A Protein Sequencer. :i ... ~ : Sweat kallikrein -0 50 • 0 : Bovine Pancr. Kal. Effect of Aprotinin, Soybean Trypsin Inhibitor, and Anti-Human c: • 0 : Salivary Kal, Urinary Kallikrein IgG on Sweat Kallikrein Twenty microliters of 0 the sweat kallikrein fraction (from the Mono Q chromatography) was mixed :e with various concentrations of inhibitors (20 Ill), or phosphate-buffered .c saline (PBS) mixed with 20 III of assay buffer (as control) and incubated for .c 30 min at 25 ° C. After the addition of 20 III of heated plasma, the reaction c: mixture was incubated for 60 min at 37" C and the generated kinins deter­ -;!!.. mined by enzyme immunoassay. Salivary and pancreatic kallikrein prepara­ 0 tions were diluted to yield a kininogenase activity comparable to that of the sweat Mono Q fraction in Fig 5. 0 Purification of Sweat Kininase After Sephacryl S-200 gel chromato­ graphy of the concentrated sweat, the kini11ase-rich fractions were further .1 .33 1.0 3.3 10 33 )lglml purified by Mono Q and Superose 12 chromatography. Inhibitor concentration

In1munohistochemical Localization of Sweat Kallikrein Anti-hu­ Figure 6. Effect of aprotinin and SBTI on sweat kallikrein and on other man urinary kall ikrein IgG was purified with a protein A-Sepharose column glandular kallikreins (i.e., bovine pancreatic kallikrein and human salivary followed by an aprotinin-Sepharose column to remove contaminating plas­ kallikrein). Kallikreins were incubated with various concentrations of apro­ min. Five-rnicrometer-thick cryostat sections of human skin were stained tinin (solid symbols) or SBT! (ope" symbols). The residual kallikrein activity with anti-urinary kallikrein IgG diluted variously from 1: 100 to 1: 500, was measured by enzyme immunoassay for kinins and the result expressed then with streptavidin. Appropriate controls were used. the percent inhibition. 218 HIBINO ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Figure 7. Immwlohistochem_ A B ical localization of sweat kal­ likrein. Skin sections were treated with anti-urinary (iden­ tical with glandular) kal­ likrein IgG. The luminal ductal cells, especially near the cuticular border, are pos­ itive for kallikrein (arrow in A). Secretory coils are shown in (B), where the peripheral cells, presumably myoepith­ elial cells, are most densely stained (arrow in B). Sections treated with the antibody pre incubated with excess sal­ ivary kallikrein showed no staining (not shown). Scale bar, 100 11m.

inogen (substrate for glandular kallikrein) to examine whether inhibitors, sweat kallikrein is most likely of the glandular type. In there was a quantitative correlation between kinin generation and fact, the antibody inhibited salivary kallikrein activity by 97%, kal­ the hydrolysis of the synthetic substrates. As expected, the correla­ likrein activity of the concentrated sweat by 96%, and Mono Q_ tion between kinin production from bovine LMW kininogen and purified sweat kallikrein by 99% (not illustrated), further suppon­ from human HP was significant (Fig 2A), although HP produced ing the idea that sweat kallikrein is of the glandular type (because more kinins than did LMW kininogen. This suggests that either urinary kallikrein is identical with glandular kallikrein [32]). One bovine LMW kininogen is a poor substrate for human glandular final bit of evidence to establish the identity of sweat kallikrein as kallikrein [29] or that sweat contains both types of kallikrein and the glandular type would be to confirm that the kinin produced by thus produced more kinins from the HP (which will also be ad­ sweat kallikrein is Iys-bradykinin (or kallidin) rather than brady_ dressed later). As shown in Figs 2B,C, the correlation between hy­ kinin. In fact, when the sweat kallikrein fraction was incubated with drolysis of the synthetic substrates and kinin liberation from both human HMW kininogen and the supernatant run on a HPLC natural substrates was absent, suggesting that sweat enzymes in­ VyDAC C 1S column (B), a sharp peak corresponding to Iys-brady_ volved in hydrolysis of these synthetic substrates may be different kinin was observed (data not illustrated). Furthermore, sequence from kallikrein. Such a problem is readily resolved because the analysis of the kinin liberated by sweat kallikrein was that of elution profile from the Sephacryl S-200 gel chromatography of Iys-bradykinin, i.e., Lys-Arg-Pro-Pro-Gly-Phe-Ser-Pro-Phe-Arg. concentrated sweat showed only one peak of kallikrein activity at Having identified sweat kallikrein as the glandular type, we then around Mr 40,000 regardless of whether HP or LMW kininogen localized kallikrein in the sweat gland using immunohistochemis­ was used as the substrate (Fig 3A). This indicates that a single en­ try. The kallikrein activity was localized in the luminal membrane zyme, most likely glandular kallikrein, is involved in the generation of the duct cells (arrow in Fig 7 A) and the peripheral rim of the of kinin from both LMW and HMW kininogens (i.e., human secretory coil segment (arrow in Fig 7B), presumably along the plasma kallikrein has a Mr of 90,000 [24] and the chromatography peripheral membranes of the myoepithelial cells. should have shown two peaks at MrS 40,000 and 90,000 when HP Finally, a limited attempt was made to characterize the sweat was used). The hydrolytic activities of both S-2266 and S-2302 kininase already shown in Figs 1 and 3. The kininase-rich fraction were eluted slightly to the left of the kallikrein peak, with the peak obtained from the Sephacryl S-200 gel chromatography (see Fig activity at around Mr 50,000 (Fig 3B), but because of the wide 3C) was applied to a Mono Q column. As shown in Fig 8, kininase overlap with kallikrein activity, no definitive conclusion can be and angiotensin I-degrading activity were separated into two peaks, drawn from these data alone. Also shown in Fig 3C is the kininase F-I and F-II. Both F-I and F-II were further purified by Superose 12 (bradykinin-degrading) activity and angiotensin I - degrading ac­ gel chromatography. Both F-I and F-U eluted at Mr 160,000 (not tivity, both of which eluted in the same fractions at approximately shown). In fact, as listed in Table I, F-I and F-II showed comparable Mr 160,000 (Fig 3C). This is of interest because ACE (which hy­ responses to various proteinase inhibitors. EDTA and 1, lO-phen­ drolyzes angiotensin I to generate angiotensin II) is identical with anthroline completely abolished the kinin-degrading activities of kininase II, a dipeptidyl carboxypeptidase [30,31]. When the kal­ both fractions, indicating that sweat kininase is a metalloproteinase. likrein-rich fraction in Fig 3 was further purified by DEAE-Sephar­ Inhibition of sweat kininase by the inhibitors of ACE, captopril (SQ ose chromatography, the kallikrein activity was eluted at about 14,225), and Gln-Trp-Pro-Arg-Pro-Gln-Ile-Pro-Pro, further sup­ 0.35 M NaCI concentration (Fig 4A), whereas major S-2266 and ported the thesis that the enzyme is kininase II (which is identical S-2302 hydrolytic activities were eluted at 0.1-0.2 M NaCl, indi­ with ACE) [30,31]. cating that sweat contains abundant non-kallikrein proteases that DISCUSSION hydrolyze these synthetic substrates. Further purification of the kallikrein-rich fraction from DEAE-Sepharose chromatography Kallikrein-like enzymes have been reported to occur in human ec­ (bar in Fig 4A) with a Mono Q column removed most of the non­ crine sweat [16-19,33]; however, no ultimate proof has been kallikrein S-substrate hydrolases so that the small peak of S-2266 presented to date. Thus the initial goal of the present study was to hydrolase now co-eluted with the kallikrein activity (Fig 5). The confirm the presence of kallikrein in sweat. In the course of the sensitivity of the Mono Q-purified kallikrein fraction to aprotinin study, we have also observed that the sweat gland secretes kal­ and soybean trypsin inhibitor (SBTI) is shown in Fig 6. As shown, likrein-like enzymes on a continuous basis, that sweat kallikrein is sweat kallikrein, as well as bovine pancreatic and human salivary of the glandular type, and that immunoreactive kallikrein can be kallikrein, was inhibited by aprotinin, but not by SBTI, in a dose­ localized in luminal ductal cells and in the peripheral rim of the dependent fashion. Because tissue kallikrein is sensitive to apro­ eccrine secretory coil, presumably in the outer membrane and/or tinin, but not to SBTI, and plasma kallikrein is sensitive to both cytoplasm of the myoepithelial cells. Furthermore, we have also VOL. 102, NO.2 FEBRUARY 1994 KALLIKREIN IN SWEAT 219

Tissue (or glandular) kallikrein is actually a generic term for a F 1 F 2 family of serine proteases of closely related structure that are found in many mammalian tissues [34,35]. In the rat and mouse, at least 20 five major gene products are intensely studied, i.e., the y subunit of .8 7S nerve growth factor (N GF),,ON GF, which forms a complex with ~ and yN~F,,ONGF-endopeptidase, epidermal growth factor-bind­ 16 ~ 16 • o ~ng protem, and glandular kallikrein [10,34]. It remains to be stud­ Ied whether human sweat and salivary kallikrein belong to a similar c c o subgroup of a gene family. .3 12 ~ 12 ,g T "C co "C . ~ysyl-bradykinin (kallidin, an octapeptide cleaved from LMW E '" e c g> kmmogen by g l a~dular kallikrein) ~nd bradykinin (nonapeptide) f "tI :5 .2 8 c 8 are strong vasodtlators. These kinms also stimulate membrane N ,... c fti ·2 c2 ·iii tr.a nsport [~1- ~3]. In the present immunohistochemical study (see ., c <.J Fig 7), kallI~em was localized in the luminal ductal cell especially lii .1 ...... •./ 4 ~ ne~r the cut1c~lar border, which is a ring of dense tonofilaments D- .6> : o c g c( ~nslde the lummal membrane, suggesting that the enzyme is located D- mtracellularly near the ductal lumen. This is analogous to the kid­ ei: 0 o 20 30 40 50 60 ml ney collectmg tubule and the salivary duct [3,36], where kallikrein Elution volume is produced in the cell interior and secreted into urine or saliva respe.ctive!y. It ren:ains unknown where kallikrein plays a majo; Figure 8. FPLC anion-exchange chromatography of sweat kininase. The physlOlog~c role, either intracellularly or in the ductal lumen (in kininase-rich fractions obtained from Sephacryl S-200 gel chromatography sweat). If l~tracellular kallikrein is the enzyme that is involved in (Fig 3) were applied to a Mono Q column. Kinin-degrading activity was the regulatIOn of cellular function, then its substrate LMW kinino­ separated into two fractions (F-I and F-I1, bars) . Both kinin and angiotensin gen.or i~s equivalent, m~st be present in the sweat giand cell. If not, degradation are expressed as ng degradation/him!. kallIkrem mu~t be play~ng a role different from kinin generation, e.g., proteolyt~c p:o~essmg of other peptide hormones and growth factors. If k~l~lkrem m sweat plays a major functional role, then its observed that sweat contains abundant S-substrate hydrolases of substrate (kmmogen) must be in the luminal membrane of the duct non-kallikrein origin and kininase II, which is the same as ACE. We and be readily accessible to sweat kallikrein because we failed to confirmed that sweat does not contain plasma kallikrein, a HMW, detect kininogen-like activity in human sweat collected from the complex protease involved in intrinsic blood coagulation [8]. Al­ skin surface (unpublished; Poblete et ai, however, were able to local­ though the present study does not address the possible function of ize kininogen in the dark cells [37]). In cultured sweat ductal cells the kallikrein/kinin system, the occurrence of both kinin-generat­ bradykinin !s ~own to stimulate sodium transport [38]. The baso~ ing (kallikrein) and kinin-eliminating (kininase II) enzymes sug­ !aterallocallzatlon of the kallikrein-like immunoreactivity is also of gests its functional roles in the regulation of glandular function. For ~nterest be~ause the e~yme. may be accessible to LMW kininogen example, the presence of kallikrein in the luminal ductal cell of the l~ t.he pe~lglandular. mterstItlum for the generation of lys-brady­ sweat duct suggests its role in ductal function. Likewise, the pres­ kuun, w~lch may ~e mstrumental in enhancing peri glandular blood ence of kallikrein in the basolateral rim of the secretory coil suggests flow dUrIng sweatmg. The presence of kininase II (ACE), an en­ that kallikrein can act on plasma LMW kininogen to produced zyme that degrades bradykinin and Iys-bradykinin (or the same lys-bradykinin, which could be instrumental in increasing periglan­ enzyme that con."~r~s angiotensin I to angiotensin II), indirectly dular blood flow. If this thesis is proved to be the case, it can readily supvorts t.he POSSibilIty that the kallikrein/kinin system (or renin/ explain why patients with anhidrotic ectodermal dysplasia lack anglOtensm s'ystem) may be functionally involved and that kinin ARV [14] during heat exposure. The consequence of this lack may be fun~t~onal only locally near the site of production. Locally would make these patients extremely susceptible to hyperthermia produced bnms would be immediately inactivated by the abundant because heat transfer from the body core to the skin surface is also kininase II in sweat. compromised due to the lack of ARV, minimizing the effectiveness . In. summary, we have. confirmed the presence of glandular kal­ of evaporative cooling when they wet the skin with water. !tkrem and klIlmase II 111 human sauna sweat and observed that immunoreactive kallikrein is localized in the luminal cell and in the basol~teral site. of ~he ~ecretory coil, raising the possibility that a functIonal kallIkrem/km1l1 system exists in the sweat gland. Thus Table I. Effect of Proteinase Inhibitors on Kininase Activity in the 'present study pr?vid.es the basis for future studies on the physio­ Fractions F-I and F-II" logiC role of the kallikre1l1/kinin system in the sweat gland in heal th and disease. F-I F-I1 Inhibition (%) REFERENCES None o o 1. Sato K, Ohtsuyama M, Sato F: Whole cell Kand Cl currents in dissociated eccrine secretory coil cells during stimulation.] Me",b Bioi 134:93-106, 1993 DFpb (1 mM) o o o 2. Petersen OH: Stimulus-secretion coupling: cytoplasmic calcium signal and the PMSF' (1 mM) o control of Ion channels in exocrine acinar cells. ] PI,ysio/448:1-51, 1992 Iodoacetamide (1 mM) o o 3. Schachter M, Peret MW, Moriwaki C, Rodrigues JAA: Localization of kallikrein EDTA (1 mM) 100 100 in submandibular gland of cat, guinea pig, do g, and man by the immunoperoxi­ 1, 10 phenanthroline (1 mM) 92 98 dase method.] Histoc!wn Cytoehetn 28:1295-1300, 1980 Pepstatin (10 mg/ml) 10 5 4. Carretero OA, Scicli AG: The renal kallikrein-kinin system. Arn ] Physiol ACE Inhibitord (10 mg/ml) 96 90 238:F247 - F255, 1980 Captopril (1 mg/ml) 98 100 5. Fiedler FE, Fink E. Tschesche H, Fritz H: Porcine glandular kallikreins. In: Lorand L (ed.). Methods ill Enzymology, Vol. 45. Academic Press, New York/ • The inhibitors were added to the reaction mixture (see Materials and Methods) and London, 1976, pp 493 - 532 incubated for 30 min at 37°C. The reaction was stopped by the addition of 20 pI 6. Fujimoto Y, Moriya H. Moriwaki C: Studies on human sali vary kallikrein. trichloroacetic acid and the residual kinin determined by enzyme-linked immunoassay. ] Bioehern 74:239-246,1973 The result is the mean of a triplicate assay. 7. Geiger R, Fritz H. Human urinary kallikrein. In: Lorand L (cd.). Methods i" ; Diisopropyl f1uorophosphate. E"zymology, Vol. 45. Academic Press, New York/ London, 1976, PP 466-492 , Phenylmethylsulfonyl fluoride . 8. Sampaio C, Wong SC, Shaw E: Human plasma kallikrein. Arch Bioehern Biophys J Gln-Trp-Pro-Arg-Pro.Gln-ile-Pro.Pro. 165:133-139, 1974 220 HIBINO ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

9. Orstavik TB, Carretero OA, Scicli AG: Kallikrein-kinin system in regulation of 24. Hibino T , Izaki S, Kimura H, Izaki M: Partial purification of plasma and tiss ue submandibular gland blood flow. A m] Pllysio/242:HI010-HI014, 1982 kallikreins in psoriatic epidermis. ] I"vest Derlllato/90:505-510, 1988 10. Bothwell MA, Wilson WH, Shooter EM: The relationship between glandular 25. Hibino T, lza ki S, lzaki M: Detection of se rine proteinase inhibitors in human kallikrein and growth factor-processing proteases of mouse submaxillary gland. cornified cells. Biocllelll Biopllys R es Co",,,,,,,, 101:948-955, 1981 ] Bioi Chem 254:7287 - 7294, 1979 26. Powers CA, Nasjletti A: Activation of a latent kinin-generating proteinase in the 11 . Baird AW, Margolius HS: Bradykinin stimulates electrogenic bicarbonate secre­ porcine anterior pituitary. Bioe/rem Biopllys Acta 790:182 -187, 1984 tion by the guinea pig gall blad der.] pharm Exp Ther 248:268-272, 1989 27. Yamada K, Erdos EG: Kallikrein and prekallikrein of the isolated basolateral 12. Smith JJ, McCann JD, Welsh MJ: Bradykinin stimulates airway epithelial Cl­ membrane of rat kidney. Kid"ey I"te", 22:331-337, 1982 secretion via two second messenger pathways. Am] Pllysio/258:L369 - L377, 28. Lija H: A kallikrein-like serine protease in prostatic fluid cleaves the predominant 1990 seminal vesicle protein.] C/ill lowest 76:1899 - 1903, 1985 13. Cuthbert AW , Margolius H: Kinins stimulate net chloride secretion by the rat 29. Webster ME: In: Erdos EG (cd.). Hawlbook oj Experilll ellta l Pharmacology. Vol. colon. B r] Pllarmaco/75:587 -598, 1982 XVII. Springer-Verlag, Heidelberg, 1970, pp 131-155 14. Lave AHG , Shanks RG: The relationship between the onset of sweating and 30. Yang HY, Erdos EG, Levin Y: A dipeptidyl carboxypeptidase that converts an­ vasodilation in the forearm during body heating. ] Physiol (Land) 162:121- giotensin I and inactivates bradykinin. Bio, helll Biophys Acta 214:374-376, 128,1962 1970 15. Brengelmann GL, Freund PR, Rosell LB, Olerud JE, Kraning KK: Abse nce of 31. KosrisJB: Angiotensin converting enzyme inhibitors. 1: Pharmacology. A m H eart active cutaneous vasod ilation associated with congenital absence of sweat ] 118:1580-1590, 1988 glands in human. Am ] PhysioI240:HS71- H577, 1981 32. Jovov B, Will NK, Donaldson PJ, Lewis SA: Vectorial secretion of a kallikrein_ 16. Fox RH, Hilton SM : Bradykinin fonnation in human skin as a factor in heat like enzyme by cultured renal cel ls. J: General properties. Alii] Pllysiol vasodilation.] Physiol (Land) 142:219-232, 1958 259:C869-C882,1990 17. Frewin DB, McConnel DJ, Downey JA: Is a kininogenase necessary for human 33. Horie N, Yokozeki H, Sato K: Proteolytic enzymes in human eccrine sweat: a sweating? La" cet n :744, 1973 screening study. Alii] P/rysio/250:R691 - 1\698, 1986 18. Fraki JE, Jansen CT, Hopsu-Havu VK: Human sweat kallikrein. Acta Dermatol 34. Ashley PL, MacDonald RJ: Kallikrein-related mRNAs of the rat submaxillary Ve" ereoISO:321 -326, 1970 gland: nucleotide sequences of four distinct types including tonin. Biochemistry 19. Mayfield RK, Sens DA, Jaffa AA, Margolium HS: Studies of sweat kallikrein in 24:4512- 4520, 1985 normal human subjects. Adv Exp Med Bioi 247B:649 - 655, 1990 35. van Leeuwen BH, Evans BA, Tregear GW, Richards RI: Mouse glandular kal­ 20. Boysen TC, Yanagawa S, Sato F, Sato K: A modified anaerobic method of sweat likrein genes. Identification, structure, and expression of the renal kallikrein collection.] Appl Physiol 56: 1302 - 1307, 1984 gene.] Bioi CI,elll 261:5529 -5535, 1986 21. Yokozeki H, Sato K: Thiol-dependent aminopeptidase (350,000 mol wt) as a 36. Beasley 0, Oza NB, Levinsky NG: Micropuncture localization of kallikrein marker of epidermal contamination in sweat. ] Appl Physiol 63: 1040-1048 , secretion in the rat nephron. Kidlley IlIIem 32:26 - 30, 1987 1987 37. Poblete M, Reynolds NJ, Figueroa CD, BurtonJL, Muller-Esterl W , Bhoola ICD: 22. Takemura T, Wertz PW, Sato K: Free fatty acids and sterols in human eccrine Tissue ka llikrein and kininogen in human sweat glands and psoriatic skin. BrJ sweat. Br] DermatoI 120:43-47, 1989 DermatoI124:236-24I,1991 23. Ueno A, O hishi S, Kitagawa T , Katori M: Enzyme immunoassay of bradykinin 38. Brayden OJ, Cuthbert AW: A novel method for cultur;ng sweat gland epithelia: using p-D-galactosidase as labeling enzyme. Biochelll Pharmacol 30:1659 - comparison of normal and cystic fibrosis tissues. Br] Pharlllacol 29:235 - 238, 1664, 1981 1990