The isolation and characterization of a novel corticostatin/defensin-like peptide from the kidney.

We report the isolation and characterization of RK-1, a novel peptide found in the kidney. RK-1 is related to the corticostatin/defensins and has the sequence MPC-SCKKYCDPWEVIDGSCGLFNSKYCCREK but differs from the very cationic corticostatins/defensins in having only one arginine and a calculated charge at pH 7 of +1. Like some myeloid corticostatin/defensins RK-1 inhibits the growth of Escherichia coli. Since corticostatin/defensins effect ion flux in responsive epithelia we used volume changes in villus enterocytes as a model system to study the effects of RK-1 on ion channels in epithelial cells. At concentrations > or = 10(-9) M RK-1 decreased enterocyte volume in a dose-dependent manner through a pathway that requires extracellular calcium and is inhibited by niguldipine, a dihydropyridine-sensitive "L"-type Ca(2+)-channel blocker. In other assay systems for corticostatin-defensins, such as the inhibition of adrenocorticotropin-stimulated steroidogenesis, or cell lysis, RK-1 was inactive or only weakly active. These results demonstrate the existence of a novel system of biologically active peptides in the kidney represented by RK-1 which is antimicrobial and can activate epithelial ion channels in vitro.

Many CS/defs stimulate a calcium-dependent volume reduction in villus enterocytes at concentrations of 10 Ϫ10 to 10 Ϫ8 M (19) in response to changes in the ion flux across the cell membrane. The maintenance of cell volume in villus epithelial cells in isotonic medium reflects a balance between the rates of salt influx and efflux (20). The addition of calcium ionophore to these cells under isotonic conditions activates charybdotoxinsensitive K ϩ channels (21) and calcium-activated Cl Ϫ channels (21). The consequent loss of these osmolytes together with water results in a reduction in cellular volume. The volume response to CS/defs requires extracellular calcium and is dihydropyridine-sensitive implying a pivotal role for L-type calcium channels in this process.
The expression of genes for CS/def-like polypeptides is not confined to myeloid cells. The intestine contains antimicrobial CS/def-like peptides called cryptdins found in Paneth's cells of the crypt but not in cells of the inflammatory system (22)(23)(24)(25). The ability of some CS/defs to elicit water movement in enterocytes (19) may contribute to the onset of diarrhea, as has been suggested for guanylin, another peptide found in Paneth's cells (26). Interestingly a further system of cationic cystine-rich peptides, the ␤-defensins, that are structurally distinct from CS/defs occurs in bovine tracheal epithelia (27) and in bovine neutrophils (28), suggesting that specialized peptidergic defense systems may occur in several tissues.
The existence of two specialized families of CS/def-like peptides, the defensins and the cryptdins, raises the possibility of other tissue-specific CS/def-related systems. Urinary tract infections may lead to kidney failure and are a significant source of Gram-negative septicemia (29). Neither myeloid nor enteric CS/defs occur in the kidney, but other kidney-specific CS/deflike peptides might contribute to the defense mechanisms of the kidney.
We used a homology screening system based on the unusual physicochemical properties of the CS/defs to detect renal-specific peptides. Chemical assays have been successful in the isolation of biologically active gastroendocrine peptides (30) but have rarely been attempted in other fields. Here we used three selection criteria: the elution position on reverse phase HPLC, an apparent size of approximately 30 residues assessed by size-exclusion HPLC and amino acid analysis, and a cystine content of about 20%. Using these criteria we isolated a peptide, RK-1, that is abundant in the kidney and is a distant relative of the myeloid/enteric CS/defs. Having proven the existence of a renal CS/def-like peptide, RK-1 was then subjected to a series of assays characteristic of the myeloid CS/defs to compare its activities to those of the classic CS/def system. These experiments show that RK-1 possesses the Ca 2ϩ channel activating effects associated with the myeloid CS/defs.

MATERIALS AND METHODS
Tissue Extracts-Freshly dissected whole rabbit kidneys, or frozen kidneys or kidney cortex (Pel-Freez, Rogers, AK), were extracted by homogenization between 1 and 5 min (depending on amount of tissue being homogenized) using a Polytron homogenizer in a high salt low pH buffer (1 M HCl, 5% formic acid, 1% NaCl, and 1% trifluoroacetic acid, 4°C). A volume to weight ratio of approximately 10:1 extraction medium:tissue was used. Fresh tissue was diced before homogenization, and frozen tissue was homogenized whole while still frozen. The high acidity and high ionic strength of this buffer solubilizes peptides but precipitates most higher molecular weight proteins and inhibits most proteolytic enzymes. In previous studies we showed that the structure of human myeloid defensins extracted from lung tissues was identical to that obtained from peripheral neutrophils when using this extraction protocol (17). After centrifugation at 2000 ϫ g for 20 min at 4°C, the pellet was re-extracted as above, and the pooled supernatants filtered over glass wool in a cold room to remove low density solids. The filtered extract was passed through ODS cartridges (Sep-Pak C 18 Bondapak, Waters, Milford, MA) connected in series and sequentially eluted first using 15% acetonitrile in 0.1% trifluoroacetic acid, then 40% acetonitrile in 0.1% trifluoroacetic acid, and finally 80% acetonitrile in 0.1% trifluoroacetic acid. The sequential elution allows us to take advantage of the differential retention of some peptides between C 18 Bondapak cartridges and C 18 Bondapak columns. This greatly simplifies the purification. The eluates were lyophilized and stored at Ϫ40°C until use.
HPLC Purification Procedures-The Sep-Pak eluates were resuspended in 0.1% trifluoroacetic acid and fractionated by reversed-phase HPLC using a Waters C 18 Bondapak column (7.8 mm ϫ 30 cm) eluted over a 3-h period using a gradient of 0 -80% acetonitrile in 0.1% trifluoroacetic acid and an elution rate of 1.5 ml min Ϫ1 with a 10-min equilibration period in 0.1% trifluoroacetic acid before the gradient. Fractions eluting close to the elution positions of rabbit, rat, and human CS/defs were subjected to amino acid analysis to determine the cystine content. Fractions with a high cystine content were subsequently purified by size-exclusion HPLC using two Waters I-125 Protein-Pak columns connected in series and eluted in 40% acetonitrile with 0.1% trifluoroacetic acid at a flow rate of 1 ml min Ϫ1 . UV absorbing fractions were again subjected to amino acid analysis. Fractions with a high cystine content, and which eluted on size exclusion chromatography in the position of unreduced CS/defs, were further purified using a second reversed-phase gradient of 12-24% acetonitrile in 0.1% trifluoroacetic acid over 60 min at 1.5 ml min Ϫ1 using a Waters C 18 Bondapak column (3.9 mm ϫ 30 cm). At this stage the peptide was pure as judged by amino acid analysis, electrospray mass spectrometry, and gas-phase microsequencing.
Amino Acid Analysis and Microsequencing-Amino acid analysis, and S-pyridylethylation were performed as described previously (6). Purified S-pyridylethylated peptides were then subjected to gas-phase Edman microsequencing using an Applied Biosystems (Foster City, CA) Model 477A Gas-Liquid protein sequencer with a microscale reaction cartridge and Applied Biosystem's MICFST program cycles (31).
Villus Cell Assay-Villus cells were isolated as described elsewhere (20). Briefly, segments of adult guinea pig jejunum were placed over metal spiral rods and vibrated in Ca 2ϩ -free phosphate-buffered physiologically balanced salt solution. Isolated cells were collected by centrifugation at 50 ϫ g for 5 min and resuspended at 0.8 -1.5 ϫ 10 6 cells ml Ϫ1 in RPMI 1640 medium without HCO 3 , containing bovine serum albumin at 1 mg ml Ϫ1 and 20 mM NaHepes, pH 7.3, at 37°C. Viability was assessed by Trypan Blue exclusion. Cell volume was measured using a Coulter counter (model ZM) with a Coulter Channelyzer (C-256) as described (20). The electronically determined cell volume correlated positively (r ϭ 0.967) over a range of tonicities with direct measurements of cell water (20). Relative cell volume was determined as the ratio of cell volume under the study conditions to the volume under basal conditions in an isotonic medium immediately before challenge. Initial rates of volume reduction were calculated by comparing relative cell volumes at 0.5, 2, and 3 min and expressing the result as percent volume change per min.
Rat Adrenal Cell Bioassay-The effect of the purified rabbit kidney peptide on ACTH-stimulated adrenal corticosterone synthesis was assayed as described previously (3).
Antimicrobial Assay-The effect of RK-1 and the rabbit neutrophil defensin CS-4/NP-2 on the survival of Escherichia coli (strain Y1090, a smooth phenotype, Promega, Madison, WI) was assessed using a modification of the methods described in Ref. 32. Cells were grown overnight in LB broth, an aliquot of 0.5 ml was then taken and grown for 2-4 h in 25-30 ml of fresh LB broth, washed in azide free Hematall Isotonic diluent (0.15 M NaCl, 3.0 mM KCl, 15 mM phosphate buffer, pH 7.5, Fisher, Montreal), and then resuspended in the same buffer containing 5% LB to a titer of approximately 10 6 colony-forming units ml Ϫ1 (the required dilution being determined from the absorbance at 600 nm compared with previously determined values). Peptides at the appro-priate concentration (Fig. 3) were added to 1.5-ml Eppendorf tubes in 10 l of 0.01% acetic acid (the zero control was 10 l of acetic acid alone). To each tube 80 l of Hematall Isotonic diluent was added followed by a 10-l aliquot of the cell stock. The final volume was therefore 100 l of Hematall diluent containing 0.5% LB, 10 5 colony-forming units ml Ϫ1 , pH approximately 6.0. After 12 h aliquots of the cell suspension were diluted in a 10 ϫ serial fashion, streaked onto agar plates using a flamed loop, and then incubated overnight at 37°C.
Cell Growth Assays-All cell lines were obtained from the American Type Culture Collection (ATCC, Rockville, MD). [ 3 H]Thymidine incorporation was performed as follows (33). 0.8 ϫ 10 5 cells/well of a 24multiwell plate were incubated for 24 h in RPMI 1640 medium (Flow Laboratories, McLean, VA) supplemented with 10% fetal bovine serum. The medium was removed and washed twice with serum-free medium, and the cells incubated in serum-free medium with various concentrations of the human defensin HP-1 or the purified rabbit kidney peptide for 24 h. [ 3 H]Thymidine incorporation was then determined as described (see Ref. 33). Cell counts were not performed because we have previously shown that the decrease in thymidine incorporation in response to HP-1 is a function of cell death (33). Cells used were A549, SK-MES-1, and CHO-K1, which are derived from a lung carcinoma, a pleural effusion from a patient with a squamous carinoma of the lung, and Chinese hamster ovaries, respectively. These cells were chosen because we had previously shown they are among the most responsive cell lines to the cytotoxic effects of myeloid defensins (5,17).

RESULTS
Purification of RK-1-Rabbit kidney extracts were fractionated by RP-HPLC and analyzed for cystine content across the region of the chromatogram corresponding to the elution positions of the myeloid CS/defs (Fig. 1A). A cystine-rich compound was found eluting at fractions 70 -73 (about 24 -26% acetonitrile) in the 15-40% acetonitrile eluate of the C 18 Sep-Pak (Fig.  1A). No cystinyl compounds were found in the 0 -15% acetoni- trile eluate of the C 18 Sep-Pak although when chromatographed this material overlaps the cystine-rich region in Fig.  1A. Clearly therefore the sequential elution of the Sep-Paks greatly simplifies the purification. The major cystine-rich component from Fig. 1A (fraction 72-75, indicated by an arrow) was fractionated by size-exclusion HPLC (not shown), and eluted in the position of the known CS/defs. The less abundant cystine containing component in Fig. 1A (fraction 67, 68, and 69) eluted by size-exclusion chromatography in the approximate position of insulin (6 kDa, not shown). The CS/def-sized material, which we called RK-1, was then purified to homogeneity by a second step of RP-HPLC fractionation (Fig. 1B) and its purity assessed by amino acid analysis. The final yield of RK-1 is 6.04 nmol/kidney or 0.81 nmol/g wet weight (n ϭ 4, 40 kidneys per determination). We were unable to detect RK-1 in other tissues (bone marrow, lung, and liver) using the chromatographic and analytical techniques used to identify RK-1 in the kidney.
Sequence Analysis of RK-1-Approximately 80 pmol of the intact pyridylethylated derivative of RK-1 was subjected to gas-phase microsequence analysis which identified all residues except residue 29. The average repetitive yield was 89%. To identify residue 29 a total of 200 pmol of the pyridylethylated peptide was digested with Asp-N-protease, which digests on the N-terminal side of aspartic acid, and the digest was then fractionated by RP-HPLC. The C-terminal digestion fragment was identified by amino acid analysis. Approximately 50 pmol were sequenced with an average repetitive yield of 86.3% giving a sequence corresponding to residues 16 -32 in the sequence given below. The deduced amino acid composition of the intact peptide corresponds closely with the composition determined by amino acid analysis (Table I). RK-1 and pyridylethylated RK-1 were also analyzed by electrospray mass spectrometry which gave masses of 3701 and 4338, respectively, agreeing with the calculated masses of 3702.1 and 4338.1. This confirms the sequence assignment and shows that all six cysteines are oxidized. The sequence of RK-1 is: Met-Pro-Cys * -Ser-Cys * -Lys-Lys-Tyr-Cys * -Asp-Pro-Trp-Glu * -Val-Ile-Asp-Gly * -Ser-Cys * -Gly-Leu-Phe-Asn-Ser-Lys-Tyr-Ile-Cys * -Cys * -Arg-Glu-Lys (where * indicates a residue conserved with the myeloid/enteric CS/defs).
This sequence is compared with representative members of the myeloid and enteric CS/defs in Table II and as a dendrogram in Fig. 2. The distance between peptides along the branches represents their relatedness expressed in arbitary units. For example, HP-1 and HP-3 which differ by one amino acid are close, whereas HP-1 and HP-4 which have only 11 amino acids in common are placed far apart on the dendrogram. In this analysis there is no more difference between defensins and cryptdins than between defensins of different species. In contrast RK-1 maps on a separate limb of the dendrogram even though it possesses the cystine motif of the CS/defs.
Effect of RK-1 on the Growth of E. coli in Culture-As shown in Fig. 3 RK-1 is antimicrobial at concentrations between 15 and 150 g/ml. This is compared with the bone marrow corticostatin/defensin NP-2 in Fig. 3 which is approximately 10-fold more active in this assay.
Effect of RK-1 on Villus Cell Volume in the Presence of Na ϩ or in a Na ϩ -free Medium-The addition of RK-1 (10 Ϫ8 M) to guinea pig villus enterocytes suspended in isotonic Na ϩ containing medium caused a volume reduction within 2 min, and by 5 min the enterocytes shrank to a relative volume (0.94 Ϯ 0.01) that did not change for the rest of the experiment (Fig. 4A). The effect of RK-1 on enterocyte volume was dose responsive: 10 Ϫ11 M and 10 Ϫ10 M had no effect on cell volume, while 10 Ϫ9 M stimulated volume reduction which was slower (1.1 Ϯ 0.2% cell Ϫ1 min Ϫ1 versus 2.0 Ϯ 0.4% cell Ϫ1 min Ϫ1 , p Ͻ 0.05) and reduced (final relative volume, 0.96 Ϯ .01 versus 0.93 Ϯ .01, p Ͻ .05) in comparison with 10 Ϫ8 M RK-1. Volume reduction was also observed at 10 Ϫ6 M which was the highest concentration tested. In Ca 2ϩ -free medium containing 100 M EGTA the effect of 10 Ϫ8 M RK-1 on enterocyte volume was prevented (final relative volume, 0.98 Ϯ .01 versus 0.93 Ϯ 0.1, p Ͻ .001). In Ca 2ϩ (1 mM) containing medium, in the presence of the dihydropyridine Ca 2ϩ channel blocker niguldipine (0.2 M), 10 Ϫ8 M RK-1 did not cause a volume reduction (final relative volume 1.00 Ϯ 0.01, p Ͻ .001, Fig. 4A). Previously we have shown that suspending the villus enterocytes in isotonic Na ϩ -free medium, which hyperpolarizes the membrane, provides an increased driving force for conductive Cl Ϫ efflux, and results in larger volume changes (20,21). We therefore examined the effect of 10 Ϫ8 M RK-1 on isotonic enterocyte volume in medium where Na ϩ was replaced isotonically with NMDG ϩ (Fig. 4B). When suspended in NMDG ϩ medium, without agonist, the villus cells loose Ϸ8% of their volume, which is consistent with Na ϩ influx having been prevented while Na ϩ efflux together with anion efflux continues through a leak pathway. The addition of RK-1 caused an increase in the initial rate of cell shrinkage compared with NMDG ϩ medium alone (4.1 Ϯ 0.4% cell Ϫ1 min Ϫ1 versus 1.9 Ϯ 0.4% cell Ϫ1 min Ϫ1 , p Ͻ 0.001) as well as a greater volume reduction (final relative volume 0.84 Ϯ 0.01 versus 0.92 Ϯ .01. p Ͻ 0.001, Fig. 4B). The volume reduction caused by RK-1 in NMDG ϩ was prevented by Ca 2ϩ -free (EGTA, 100 M) NMDG ϩ medium (final relative volume 0.93 Ϯ 0.02 versus 0.84  (3), and the enteric CS/def murine cryptdin 1 (22) The sequences are aligned at the cysteines with dashes introduced to maximize the alignment. Fig. 4B).
Effect of RK-1 on ACTH-stimulated Adrenal Steroidogenesis-A total of 12 M RK-1 reduced the corticosterone output of rat adrenal cell suspensions in response to 150 pg of ACTH by 40% (not shown). Because of the high levels of peptide required we were unable to reach the I.D. 50. In comparison CS-1 has an I.D. 50 of 25 nM (3). RK-1 is therefore not corticostatic.
Effect of RK-1 on Mammalian Cell Survival-RK-1 had no effect on the growth of the mammalian epithelial cell lines SK-MES-1, CHO-K1, and A549, although the related peptide HP-1 was cytotoxic at the concentrations assayed (Fig. 5) as previously reported (33). DISCUSSION Here we demonstrate the existence of a novel peptide found in the kidney which we call RK-1 and which has antimicrobial activity against E. coli, and activates ion channel activity. RK-1 possesses the cystine motif of the CS/defs and has potent biological activities. It is, however, structurally distinct from both the myeloid and enteric peptides indicating the evolution of at least three families of CS/def-like peptides from a common ancestor with each family restricted in its distribution to either neutrophils and alveolar macrophages, the intestinal crypts, or the kidney. RK-1 is the least cationic of the CS/def-like peptides and unlike the myeloid and enteric CS/defs the positive charge is carried mainly on lysine residues. To distinguish RK-1 and related peptides, peptides from the myeloid or enteric CS/defs, and to avoid the potential problems associated with naming a peptide from a tissue source or a bioassay, we propose the generic term lysyl-CS/def with the individual peptides given initials according to their species.
Having proven the existence of a CS/def-related peptide in the kidney we performed several assays to compare its biological characteristics with those of myeloid and enteric CS/defs. Among these we examined the antimicrobial activity of RK-1 against E. coli, and its ability to kill mammalian epithelial cells. E. coli was selected because it is a common bacterial pathogen in urinary tract infections (29) and is a standard test organism for CS/defs and cryptdins. RK-1 is antimicrobial in this assay, although less so than the bone marrow-derived peptide CS-4/NP-2. Although the role of CS/defs in host defense is well established (1) some CS/defs are only weakly antimicrobial (32). The conditions used here were based on those described by Lehrer et al. (32) and may not reflect the conditions in the kidney. Further experiments using other conditions and organisms will be required to show the optimal conditions for RK-1 antimicrobial activity.
The villus enterocyte volume assay is a useful model to study the effects of biological mediators on ion channels in epithelial cells (19 -21). RK-1, like the corticostatins (19), stimulates the shrinkage of villus cells in isotonic media both in the presence and absence of sodium. The decrease in volume requires extracellular calcium and is inhibited by niguldipine, an L-type calcium channel blocker (34). Thus the effects of RK-1 on calcium dependent volume changes in villus enterocytes and, by implication on dihydropyridine-sensitive L-type calcium channels in these cells, are similar to those previously reported for the CS/defs (19).
There is considerable evidence that dihydropyridine-sensitive voltage-dependent Ca 2ϩ channels are important in the kidney. The dihydropyridine-sensitive voltage-dependent Ca 2ϩ channel is composed of five subunits of which the ␣ 1 -subunit forms the ion pore. At least five forms of the ␣ 1 -subunit exist with various electrophysiological properties and sensitivities to pharmacological probes and each encoded by separate genes (35). RNA transcripts of four of the ␣ 1 -subunit genes have been demonstrated in the kidney and are differentially distributed between the cortex and the medulla (36). Possible roles for these channels include Ca 2ϩ -reabsorption (37), tubular epithelial cell volume regulation (38 -41), in mesangial cell function (42), and in glomerular hemodynamics (43). This does not prove that the properties of RK-1 as a functional agonist of Ca 2ϩ channels in epithelial cells are physiologically relevant but indicates that the molecular and cellular machinery exists for RK-1-like peptides to act through Ca 2ϩ channels in the kidney.
In summary, the major finding of this paper is the existence of a novel and potently biologically active peptide in the kidney. The similarity of this molecule with the CS/defs reveals a network of CS/def-like peptides with each branch of the family restricted to a very narrow range of tissues. The evidence presented suggests two hypotheses for the function of RK-1. First, based on its homology with the defensins and cryptdins and its antimicrobial activity, RK-1 may be involved in disease resistance in the kidney. Second, the biological effects of RK-1 on calcium dependent volume regulation may indicate a role for RK-1 as a local regulator of calcium channels. Our demonstration of a novel peptide in the kidney exemplified by RK-1 should open important new avenues of investigation into renal physiology and pathology.