A novel form of neurotensin post-translationally modified by arginylation.

A novel bioactive form of neurotensin post-translationally modified at a Glu residue was isolated from porcine intestine. Purification of the peptide was guided by detection of intracellular Ca2+ release in SK-N-SH neuroblastoma cells. Using high resolution accurate mass analysis on an ion trap Fourier transform mass spectrometer, the post-translational modification was identified as arginine linked to the gamma-carboxyl of Glu via an isopeptide bond, and we named the newly identified peptide "arginylated neurotensin" (R-NT, N-(neurotensin-C5-4-yl)arginine). Although arginylation is a known modification of N-terminal amino groups in proteins, its presence at a Glu side chain is unique. The finding places neurotensin among the few physiologically active peptides that occur both in post-translationally modified and unmodified forms. Pharmacologically, we characterized R-NT for its ligand activity on three known neurotensin receptors, NTR1, -2, and -3, and found that R-NT has similar pharmacological properties to those of neurotensin, however, with a slightly higher affinity to all three receptors. We expressed the intracellular receptor NTR3 as a soluble protein secreted into the cell culture medium, which allowed characterization of its R-NT and neurotensin binding properties. The creation of soluble NTR3 also provides a potential tool for neutralizing neurotensin action in vivo and in vitro. We have shown that SK-N-SH neuroblastoma cells express NTR1 and NTR3 but not NTR2, suggesting that the Ca2+ mobilization elicited by R-NT is via NTR1.

Orphan G-protein-coupled receptors are present in established cell lines and may serve as screening systems to find their endogenous ligands. In the present study, SK-N-SH neuroblastoma cells were utilized to screen for novel peptides in porcine duodenal extracts using intracellular Ca 2ϩ release as a reporter system. However, SK-N-SH cells also express other G-protein-coupled receptors and may respond to a number of neuropeptides. The bioactivities were followed by peptide purification and subsequent structural characterization, yielding peptides of potential physiological significance, among them a unique form of neurotensin.
Neurotensin is a 13-residue peptide that was first isolated from bovine hypothalamus during purification of substance P and was named from its neuronal localization and a vasodilatation after local administration (1). Neurotensin is localized to both the central nervous system and peripheral tissues mainly in the gastrointestinal tract (2). In the central nervous system, neurotensin plays the role of a neurotransmitter or neuromodulator of dopamine transmission and of anterior pituitary hormone secretion (3). It also shows potent hypothermic and analgesic effects in the brain (4). Neurotensin has been implicated in the pathophysiology of several brain diseases, such as schizophrenia and Huntington and Parkinson diseases (5). In the periphery, neurotensin acts as a local hormone exerting a paracrine and endocrine modulation of the digestive tract (3). It also stimulates the growth of various gastrointestinal tissues, including the pancreas and the gastric antrum (6).
Three subtypes of neurotensin receptors (NTRs) 2 have been reported. Neurotensin receptor 1 (NTR1) (7) and receptor 2 (NTR2) (8 -10) are two related G-protein-coupled receptors sharing 43% amino acid sequence identity and 64% structural identity. Neurotensin receptor 3 (NTR3) (9) is a single transmembrane receptor previously characterized as sortilin (11). NTR1 has been reported to regulate the turning behavior in mice (12). For signal transduction, NTR1 is coupled with Ca 2ϩ signaling. Activation of NTR1 by neurotensin leads to intracellular Ca 2ϩ mobilization. NTR2 was discovered based on its homology to NTR1. Recent studies show that NTR2 is involved in neurotensin-induced analgesia (12). NTR2 has been shown to bind neurotensin and can be activated by neurotensin when NTR2 is expressed in Xenopus oocyte. However, unlike NTR1, which stimulates Ca 2ϩ in mammalian cells, recombinant NTR2, when expressed in HEK293 cells, does not activate Ca 2ϩ signaling in response to neurotensin stimulation. NTR3 normally exists in intracellular compartments and not on the cell surface. NTR3 expression has been reported in rat adipocytes in vesicles containing the glucose transporter GLUT4 and mouse neurons (13). Insulin or neurotensin stimulation of rat adipocytes or mouse neurons, respectively, has been reported to trigger the translocation of NTR3 from the intracellular compartment to the cell surface. So far, very little is known about the physiological function and the signal transduction pathway of NTR3.
Earlier structural analyses of hormonal peptides have shown that several peptides are post-translationally modified. The most frequently occurring post-translational modification is C-terminal amidation found in peptide hormones such as secretin (14,15), cholecystokinin (16) and vasoactive intestinal peptide (17,18). This feature has been utilized for the discovery of hormonal peptides such as peptide histidine isoleucine amide (19,20), peptide YY (19,21), neuropeptide Y (22,23), and galanin (24). However, other types of post-translational modifications also exist in peptide hormones, including tyrosine sulfation of * This work was supported by the Swedish Institute, the Swedish Society for Medical Research, the Swedish Science Council, and Karolinska Institutet. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1  cholecystokinin (16), serine phosphorylation of peptide YY (25), phosphorylation of adrenocorticotrophic hormone (26), and pancreastatin (27). We now describe, in isolation from porcine intestine, the structural and pharmacological characterizations of neurotensin post-translationally decorated with arginine (R-NT; N-(neurotensin-C5-4-yl)arginine)). We have shown that it has a potent biological activity. During this purification, two N-terminally truncated neurotensin forms, neurotensin-(2-13) and neurotensin- (3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13), both biologically active, were also isolated and characterized.
Ca 2ϩ Assay-SK-N-SH cells were plated onto 96-well black viewplates (Packard, CT) and incubated at 37°C in 5% CO 2 overnight. The culture medium was removed, and the cells were treated for 1 h at room temperature with 4 M Fluo-3 dye (Teflabs, Austin, TX), 0.04% pluronic F-127, 2.5 mM probenecid in Dulbecco's modified Eagle's medium/F-12 without phenol red. Chromatographic fractions were lyophilized and dissolved in Dulbecco's modified Eagle's medium/F-12 and plated onto 96-well V-bottom plates (Greiner, Longwood, FL). Several receptor ligands were used as positive and negative controls. From these plates, the solutions were removed by an automatic 96-tip pipette and added to the 96-well plates with Fluo-3-loaded cells. Fluorescence intensities were recorded during 2 min and plotted as a dependence of fluorescence intensity (%) versus time (2 min full scale). The ligand additions and intracellular Ca 2ϩ measurements were carried out on a FLIPR (fluorometric imaging plate reader) instrument (Molecular Devices, Sunnyvale, CA).
Purification of R-NT-The general procedure is shown by the flowchart in Fig. 1. The starting material was obtained as a side fraction from the purification of secretin and its precursor forms (28) after chromatography on CM-cellulose (Whatman) with 22.5 mM sodium-phosphate (pH 6.4) and a linear gradient of 0 -0.3 M NaCl. Fractions were desalted on a Sephadex G-25 Coarse column (Amersham Biosciences) using 0.2 M acetic acid and purified further onÄKTA Explorer and Purifier HPLC systems (Amersham Biosciences). The lyophilized material from fractions 1 and 2 (in total 118 mg) was dissolved in 40 ml of sodium-acetate, pH 5.2, after which 26 ml was chromatographed on a Resource S 6-ml column (Amersham Biosciences) using a solvent system consisting of A (20 mM sodium-acetate, pH 5.2, containing 20% acetonitrile) and B (A ϩ 2 M NaCl) with a gradient of 1-6% B in 35 column volumes (CV). Fractions of 4.5 ml were collected. Fractions 1-21 were pooled, diluted with water to 336 ml, adjusted to pH 2.5, and further chromatographed on the Resource column with a solvent system consisting of A at pH 2.5 and B with a gradient 4 -23% B over 30 CV, with collection of 4.5-ml fractions.
The material was purified further by three steps of reverse phase HPLC (Fig. 2). Fractions 12-15 were pooled, diluted with 2 volumes of water, and chromatographed on ODS-AP (10 ϫ 100 mm, 5 m) (YMC, Schermbeck, Germany) using a solvent system consisting of A, 0.1% TFA/water, and B, 0.1% TFA/water/95% acetonitrile, and a gradient of 14 -42% B over 30 CV, with collection of 3.5-ml fractions. This chro-matography was performed three times, and two active regions (fractions 19 -20 plus fractions 22-23) were observed and further purified separately. Use of fractions 19 and 20 is given in the next section, whereas fractions 22 and 23 were pooled, diluted with water, and chromatographed on the Source 5RPC (ST 4.6/150, Amersham Biosciences) column with the solvent system as in the previous step. The gradient was 19 -31% B in 15 CV, and 500-l fractions were collected. Fraction 36 was diluted with water and chromatographed on RPC C2/C18 (ST 4.6/100, Amersham Biosciences) using the same solvent system as in the previous step and a gradient of 20 -30% B over 15 CV. Fraction 28 was subjected to structural analysis.
Purification of Neurotensin-(2-13)-Fractions 19 and 20 from the YMC ODS-AP chromatography ( Figs. 1 and 2), with the activity as described above, were pooled and diluted with water to 31 ml. The resulting solution was chromatographed on the Source 5RPC column using the same solvent system as above. A gradient of 17-32% B over 15 CV was employed, and 0.5-ml fractions were collected. Fraction 35 was diluted with 400 ml of water and chromatographed on Vydac C18 column (4.6 ϫ 50 mm, 3 m, 300 Å) using a solvent system consisting of A, 0.1% TFA/water, and B, 0.1% TFA/water/95% methanol. A gradient of 28 -44% B over 30 CV was used, and 300-l fractions were collected. Fraction 11 was subjected to structural analysis.
Purification of Neurotensin-(3-13)-The starting material for this part was prepared as described previously (29) (Fig. 1C). The material (1.5 g) from the methanol extraction and precipitation at neutral pH (acidic methanol-soluble neutral insoluble fraction) (29) was dissolved in 500 ml of degassed water; the pH was adjusted to 5.2 with 2 M NaOH and passed through a 0.45-m Millipore filter. The liquid was divided into 10 aliquots and applied onto the Resource S column. After equilibration with 6 CV of 20 mM sodium-acetate, pH 5.2, peptides were eluted with a gradient of 0 -15% B (2 M NaCl in the same buffer) in 20 CV, and in total, 10 runs were performed. The activity was detected in fraction 12 and collected for the next step.
The active fractions were diluted 10-fold with degassed water, and the pH was adjusted to 2.5. The sample was applied onto the same column and run with a solvent system of 20 mM sodium-phosphate, pH 2.5, 20% acetonitrile (eluent A). The peptides were eluted with a gradient of 9 -23.5% B (2 M NaCl in the same buffer) over 30 CV. The active fractions 9 -11 were pooled. These combined fractions were applied onto the YMC ODS-AP column. The solvents used were A, 0.1% TFA/water, and B, 0.1%TFA/95% acetonitrile/water. The peptides were eluted with a gradient of 20 -40% B over 30 CV. The active fractions from 2 runs were pooled for the next step. The material was then diluted 2-fold with 0.1% TFA/water (eluent A) and applied onto the RPC C2/C18 column. The solvent system was the same as in the previous step. Peptides were eluted with a gradient of 20 -30% B over 15 CV. The active fractions 17-20 were analyzed by matrix-assisted laser desorption ionization (MALDI) mass spectrometry (MS), and fraction 17 was analyzed by Edman degradation for 20 cycles. The remaining material from fractions 17-20 after MALDI MS was pooled and purified further.
The peptides were applied onto cation exchange Mini S PE 4.6/50 (4.6 ϫ 50 mm, 3 m, Amersham Biosciences) in 20 mM acetate, pH 4.1, 20% acetonitrile (eluent A). Peptides were eluted with a gradient of 0 -15% eluent B (2 M NaCl in the same buffer) over 25 CV. Activity was detectable in fractions 31-34. The most active fractions, 32 and 33, were pooled for the next purification step on a RPC C2/C18, as described above, with the same solvent system. The material was eluted with a gradient 20 -30% B in 15 CV. Fraction 13 was active.
Mass Spectrometry and Sequence Analysis-MALDI-TOF mass spectra were recorded on a Voyager DE Pro instrument (Applied Bio-systems, Foster City, CA). Samples from the HPLC separations (0.7-1 l in 0.1% TFA/water/acetonitrile) were mixed on a stainless steel MALDI plate with 0.7 l of standard solution and 0.7 l of matrix solution and allowed to dry at room temperature. A standard solution was prepared in 0.1% TFA/water yielding Arg 8 -vasopressin at 0.28 pmol/l and human insulin at 18 pmol/l. The spectra were calibrated using the values for monoisotopic singly charged ions of Arg 8 -vasopressin (m/z 1084.4457) and insulin (m/z 5804.6455). Accuracy of mass measurement was 0.05 Da. Electrospray ionization time-of-flight MS spectra were recorded on an Ettan electrospray ionization time-of-flight mass spectrometer (Amersham Biosciences). Collision-induced dissociation spectra were recorded on a quadrupole-TOF tandem mass spectrometer equipped with a Z spray nano-electrospray interface (Waters/ Micromass, Manchester, UK) as described previously (30).
For C-terminal ladder sequence analysis (31), samples from the HPLC separation (0.7-1 l) were mixed on a MALDI plate with 0.7 l of CPY solution and allowed to dry at room temperature. 0.7 l of standard solution and 0.7 l of matrix solution was added to the same spot and allowed to dry again at room temperature. N-terminal sequence analysis was carried out on an Applied Biosystems Procise HT sequencer.
Accurate measurements of mass were carried out on a Finnigan LTQ-FT (Thermo Electron GmbH, Bremen, Germany) instrument. The LTQ-FT is a hybrid instrument and consists of two mass analyzers, a linear ion trap, and a Fourier transform ion cyclotron resonance cell. The linear ion trap is characterized by high scan rates also in the MS/MS mode of operation. The Fourier transform ion cyclotron resonance analyzer routinely achieves a mass resolution of r ϭ 100.000 at m/z 400 with a scan rate of 1 scan/s, while providing mass measurement accuracies Ͻ2 parts/million utilizing external calibration.
Detection of mRNA Expression of Neurotensin Receptors in the Human Cell Line SK-N-SH-Total RNA from SK-N-SH cells was treated with DNase I (Epicenter Technologies) to remove genomic DNA. cDNA was synthesized from 2 g of total RNA using SuperScript III reverse transcriptase (Invitrogen) with 100 ng of oligo(dT) 18 -21 (Amersham Biosciences) according to the manufacturer's protocol. The reaction was incubated at 50°C for 30 min and then heat-inactivated at 80°C for 3 min and chilled on ice. The cDNA was diluted 20-fold, and 2 l of sample was analyzed by quantitative PCR using the SmartCycler (Cepheid) in duplicates. The PCR mix consisted of 0.2ϫ Sybr Green I (Invitrogen), 10 mM Tris-HCl, pH 8.8, 50 mM KCl, 1 unit of TaqStart antibody (BD Biosciences), 3 units of AmpliTaq DNA polymerase (Veridex), 200 mM trehalose (Sigma), and 200 M dNTPs (Amersham Biosciences). PCR primers were from GenBase (San Diego, CA). PCR cycle parameters were: initial hold at 95°C for 90 s, 40 cycles of 95°C for 5 s, 62°C for 7 s, and 72°C for 12 s. At the end, a melt curve analysis was done. PCR products were also analyzed on a 2% agarose gel. The primer sequences were: NTR1, 5Ј-CTG GAT GAG ACT GTC CTG GA-3Ј and 5Ј-TCA ACG GGA AAG CCG ACT TC-3Ј; NTR2, 5Ј-AGC CAG GGC GAC TTC TAT CA-3Ј and 5Ј-CGC ACC AGG TTC TGT GCT AA-3Ј; and sortilin 1, 5Ј-AGG AAG GGA TTC GGC TTT CA-3Ј and 5Ј-CCC ATG GCA GAC AGA ACA GA-3Ј.
Molecular Cloning of Human NTR1, NTR2, and NTR3-The human NTR1 complete coding region was amplified by PCR using forward primer (5Ј-TCT CGC TAG CGC CAC CAT GCG CCT CAA CAG CTC CGC GCC GGG AAC CCC GGG CAC GC-3Ј) and reverse primer (5Ј-ACT AGA GCG GCC GCC TAG TAC AGC GTC TCG CGG GTG GCA TTG-3Ј) and human brain cDNA pool (BD Biosciences) as the template. Human NTR2 cDNA was PCR-amplified from the human brain cDNA pool using two primers (forward primer 5Ј-ACG GAA TTC GCC ACC ATG GAA ACC AGC AGC CCG CG-3Ј and reverse primer 5Ј-ACG AGT GCG GCC GCT CAG GTC CGG GTT TCT GGG GGA TCC-3Ј). Similarly, the human NTR3 extracellular domain (amino acids 1-756) coding region was amplified using forward primer 5Ј-ATA GAA CTC GAG GCC ACC ATG GAG CGG  CCC TGG GGA GCT GCG GA-3Ј and reverse primer 5Ј-TCT CTA  TCT AGA TTA CTT GTC ATC GTC GTC CTT GTA GTC AGA  ATT TGA CTT GGA ATT CTG TTT TTC CGG A-3Ј. NTR1, -2, and -3 extracellular domain coding regions were then cloned into the mammalian expression vector pCIneo (Promega, Madison, WI), respectively. All DNA constructs were sequenced to confirm the sequence identities.
Characterization of Neurotensin Receptors Using Radioligand Binding Assays-For NTR1 and NTR2, the binding assays were performed as described previously (32) with the membranes, radioligand, and unlabeled ligands as described below. NTR1-and NTR2-expressing DNA constructs were transiently expressed in COS-7 cells using Lipofectamine as the transfection reagent. Two days after transfection, the cell membranes were prepared and used for 125 IY 0 -neurotensin (PerkinElmer Life Sciences, Boston, MA) binding assays in the presence of different concentrations of neurotensin or R-NT as the competitors. The final concentration used for NTR1 and NTR2 binding assays is 30 pM. For NTR3, which was expressed as a soluble form with a FLAG tag at the C terminus, was transiently expressed in COS-7 cells. Three days after transfection, the conditioned medium was collected, and the soluble NTR3/ 125 IY 0 -neurotensin binding was performed similarly as described previously (49). Briefly, the conditioned media, either from mock or NTR3 cDNA-transfected cells, were centrifuged (10,000 ϫ g for 10 min at 4°C) to remove the cell debris. The clear supernatants were aliquoted in 1.5-ml tubes at 1 ml/tube; 125 IY 0 -neurotensin was added to each tube at a final concentration of 30 pM. Different concentrations of neurotensin or R-NT were added to the tubes, respectively, as the competitors. 20 l of anti-FLAG-agarose beads (Sigma) was added to each tube to capture the NTR3-FLAG-soluble receptor. The tubes were incubated on a rocking platform at 4°C overnight. The NTR3-FLAG-soluble receptor with the bound 125 IY 0 -neurotensin complex, which was captured by anti-FLAG beads, was precipitated by centrifugation. The supernatants, which contain the unbound ligand, were aspirated, and the pellets, containing the bound radioligand, were counted in a ␥ counter (Quantum Cobra, Packard Instrument).
Functional Characterization of NT1 and NT2 Receptors-Recombinant NTR1-and NTR2-expressing DNA constructs were transiently expressed in HEK293 cells either alone or in co-expression with G␣ 16 (33) using Lipofectamine as the transfection reagent. Two days after transfection, cells were detached from the cell culture plates using phosphate-buffered saline plus 10 mM of EDTA and then seeded in 96-well black well poly-D-lysine-coated tissue culture plates as described above. After Ca 2ϩ dye Fluo-3 was loaded to the cells, different concentrations of neurotensin or R-NT were used to stimulate the intracellular Ca 2ϩ mobilization, which was monitored using a FLIPR instrument. Figs. 1 and 2. The peptide from fraction 28 of the RPC step, which eluted at 19 ml, was active in the Ca 2ϩ release assay. Initially, the peptide was analyzed by MALDI-TOF and electrospray ionization time-of-flight MS, which gave an m/z of 1829.04, indicating a 156.1 Da higher mass than that of intact neurotensin. N-terminal sequence analysis was attempted several times without result, indicating that it was N-terminally blocked. The peptide was then subjected to C-terminal ladder sequence analysis with CPY-producing fragments at m/z 1602.86, 1715.96, and 1829.01, which gave mass values 113.10 and 113.05 Da for the two C-terminal residues, consistent with Leu/Ile for the last two residues. Further digestion with higher concentrations of CPY revealed an additional amino acid with 163.05 Da, consistent with Tyr (163.06 Da). Thus, the three-residue C-terminal sequence is Y(I/ L)(I/L), and is in accordance with the neurotensin structure.

Characterization of Post-translationally Modified Neurotensin-Purification was carried out as described under "Experimental Procedures" and shown in
The peptide was digested on the MALDI target with trypsin, after which analysis by MALDI-TOF MS showed two values of m/z 1186.65 and 661.43. The latter value is in good agreement with the theoretical mass value of the C-terminal neurotensin fragment RPYIL (calculated value, 661.40), but the former m/z value (1186.65) is 156.1 Da higher than that for the full-length peptide, indicating that a modification is located within the N-terminal tryptic fragment and that the peptide may contain one extra arginine residue.
To further characterize this structure and to obtain an accurate mass of the peptide, experiments were carried out on an LTQ-FT, a linear ion trap Fourier transform MS instrument. First the molecular mass of the whole peptide was measured (Fig. 3). Ion species with doubly, triply, and quadruply charged ions were detected in the mass spectrum. From these results, it became clear that the modified peptide is capable of accommodating four charges, one more than for unmodified neurotensin, suggesting that the modification contains a group capable of proton attachment, further supporting that it may be arginine. Molecular mass of the peptide was calculated from four ion species of a single scan, and it gave 1828.01172 Da (calculated value 1828.01077). Further confirmation was obtained from the comparison of the isotopic patterns of the measured peptide and a simulated isotopic pattern (Fig. 4). The patterns were found to be very similar.
To exactly locate the modification, both the intact peptide and its N-terminal tryptic fragment were subjected to MS/MS analysis on a LTQ-FT instrument and on a quadrupole-TOF MS instrument. Fragmentation of the [M ϩ 4H] 4ϩ ion of the intact peptide in the LTQ-FT showed identifiable fragments b 4 ϩ , b 5 ϩ , y 10 3ϩ , and y 10 2ϩ , suggesting that the modification is located at Glu-4 (Fig. 5A). Collision-induced dissociation analysis of the N-terminal tryptic fragment with the quadrupole-TOF instrument detected a, b, c, and y ions spanning the whole sequence. The b 4 , b 5 , b 6 , and b 7 ions were found to be shifted by the mass value of the modification and clearly indicated that the modification was     (2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13). Column used was a Source RPC 4.6/150. Solvent system was the same as in A. Gradient was 17-32% in 15 CV. Active fraction 35 is indicated. C, final purification of neurotensin- (2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13). Column used was a Vydac C18 (4.6 ϫ 50 mm, 3 m). Eluents used were A (0.1% TFA/water) and B (0.1% TFA/water/95% methanol). Gradient was 28 -44% B over 30 CV. Active fraction 11 is indicated. D, purification of R-NT. Column used was a Source RPC 4.6/150. Solvent system was the same as in A. Gradient was 19 -31% B in 15 CV. Active fraction 36 is indicated. E, final purification of R-NT. Column was a RPC C2/C18 (4.6 ϫ 100). Eluent system used was as in A. Gradient was 20 -30% B in 15 CV. Active fraction 28 is indicated. AU, absorbance units. ϩ , doubly charged y 10 2ϩ , and triply charged y 10 3ϩ are observed corresponding to the fragments indicated. B, quadrupole-TOF tandem mass spectrum of the N-terminal tryptic peptide of the post-translationally modified neurotensin. Peak #1 is an arginine-related ion (m/z 112.11). The spectrum was recorded using a quadrupole-TOF instrument (Micromass). C, LTQ-FT MS/MS data of the tryptic fragment at m/z 593.8. The produced and identified singly charged fragments are indicated. As for the spectrum shown in A, the data were obtained with a mass resolution of r ϭ 100.000 at m/z 400 (1 scan/s). Peaks indicated with an asterisk in B and C correspond to the loss of Arg from the tryptic fragment of R-NT. located at the Glu residue (Fig. 5B). At the same time, the result confirmed the whole sequence of the tryptic fragment. A complementary experiment with the tryptic fragment was carried out on the LTQ-FT MS instrument at a considerably higher mass accuracy. A similar fragmentation pattern for b ions was observed (Fig. 5C), but most of the y ions were absent from this spectrum. From these fragmentation data, we conclude that the post-translational modification is located at Glu-4.
For elemental composition analysis, data from accurate mass measurements of M ϩ , M 2ϩ , and M 3ϩ ions were combined with the theoretical values for unmodified neurotensin. These calculations gave 156.0990, 156.10118, and 156.10115 Da yielding an average of 156.1004 Da for the modification. Elemental composition calculations at a mass measurement accuracy of 0.001 Da retrieved two alternatives, C 4 H 10 N 7 and C 6 H 12 N 4 O, with almost equal mass deviations at 0.0006 and 0.0007 Da, respectively. The combination of this result with the nitrogen rule leaves the latter as the only alternative, which corresponds to the elemental composition of an arginine residue. Together all these facts show that neurotensin is post-translationally modified by arginine coupled to the ␥-carboxyl group of Glu-4 by an isopeptide bond (Fig. 6).
Characterization of Neurotensin-(3-13)-Bioactive fractions from the final purification step (Fig. 1C) were subjected to N-terminal sequence analysis, and a sequence for 6 cycles was determined (YEN-KPR). This sequence corresponds to that of neurotensin starting at position 3. The fraction was also subjected to MALDI-TOF MS analysis, which gave m/z 1448.82 for the singly charged ion. This m/z value is in accordance with the theoretical value m/z 1448.80 of neurotensin- (3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13) with the sequence YENKPRRPYIL. The peptide was subjected to C-terminal ladder sequence analysis by digestion with CPY and subsequent MALDI-TOF MS, which gave 113.11 and 113.01 Da for the two C-terminal residues confirming the C-terminal end Ile-Leu (theoretical value for Ile/Leu 113.08 Da). The structures of the peptides isolated in the present study are given in TABLE ONE.
Neurotensin-As mentioned above, we were searching for peptides that cause intracellular Ca 2ϩ release in SK-N-SH cells. A number of active fractions were detected early in the purification, and the activities were followed through several steps until pure peptides were obtained and identified. In the course of this work, ordinary neurotensin was also purified, and it showed strong Ca 2ϩ -releasing activity in SK-N-SH cells (not shown).

Synthetic R-NT Stimulates Ca 2ϩ Response in SK-N-SH Cells-To
confirm our finding that R-NT is capable of stimulating Ca 2ϩ response in SK-N-SH cells, R-NT was synthesized and tested in Ca 2ϩ mobilization assay in comparison with neurotensin using untransfected SK-N-SH and HEK293 cells. The results show that R-NT, similar to neurotensin, potently stimulates Ca 2ϩ mobilization in SK-N-SH cells, with an EC 50 value of 0.75 nM but not in HEK293 cells (Fig. 7A). In the same experiment, neurotensin stimulated Ca 2ϩ response in SK-N-SH cells with an EC 50 value of 2.7 nM.
SK-N-SH Cells Express NTR1 and NTR3 but Not NTR2 mRNA-To investigate which neurotensin receptor SK-N-SH cells express, cDNA was made from SK-N-SH cells and PCR was employed to detect the neurotensin receptor mRNA expression. The results show that SK-N-SH cells express NTR1 and NTR3 mRNAs (Fig. 7B). In the same experiment, no detectable NTR2 mRNA was found in SK-N-SH cells, suggesting that the Ca 2ϩ mobilization stimulated by R-NT in SK-N-SH cells is via NTR1, because NTR3 is a single transmembrane receptor and not known to couple with Ca 2ϩ signaling.
R-NT Binds NTR1, NTR2, and NTR3 at High Affinity-To investigate whether R-NT binds NTR1, NTR2, or NTR3, unlabeled R-NT, in comparison with neurotensin, was used as the competitor in binding assays using 125 IY 0 -neurotensin as the tracer and different recombinant neurotensin receptors. The results show that R-NT and neurotensin bind NTR1 at IC 50 values of 1 and 2 nM, respectively (Fig. 8A). Similarly, R-NT and neurotensin bind NTR2 with IC 50 values of 1.8 and 0.49 nM, respectively. Compared with NTR1 and NTR2, R-NT and neurotensin bind NTR3 at much lower affinity with IC 50 values of 7.7 and 21 nM, respectively (Fig. 8B).
R-NT and Neurotensin Stimulate Ca 2ϩ Response in NTR1-but Not NTR2-expressing HEK293 Cells-To further characterize which neurotensin receptor R-NT activates, R-NT in parallel with neurotensin was tested on HEK293 cells expressing recombinant NTR1 or NTR2. The results indicated that both R-NT and neurotensin potently stimulate Ca 2ϩ mobilization (Fig. 8C) in cells expressing NTR1 with EC 50 values of 65 pM and 0.2 nM, respectively. However, neither neurotensin nor R-NT evoked significant Ca 2ϩ response in HEK293 cells expressing NTR2 (Fig. 8C).

DISCUSSION
In this study, we have isolated and characterized a novel post-translationally modified neurotensin and shown that it is biologically active. The structural analysis indicates that the peptide is modified by arginylation of the side chain of Glu-4 (Fig. 6). To our knowledge, such posttranslational modification of peptides or proteins has not been previ-  ously described. However, arginylation of N-terminal residues of proteins is a known phenomenon, occurring in all eukaryotic cells and constitutes a vital function in cell physiology (34 -36). In the presently known N-terminal arginylation, the modification occurs at the N-terminal ␣-amino group, not at the side chain. tRNA-Arg is used as the Arg source, and the reaction is carried out by arginyl-tRNA-protein transferase. The most common amino acids that undergo N-terminal arginylation are Asp and Glu and rarely also Cys (37,38). The biochemical mechanisms and physiological significance of Glu arginylation of neurotensin remain to be investigated. Nevertheless, our data indicate that the modified neurotensin peptide is active and elicits Ca 2ϩ release in SK-N-SH cells.
Overall, R-NT has very similar pharmacology on all three neurotensin receptors compared with neurotensin but with higher potency. In the Ca 2ϩ mobilization assay, we observed higher potency for R-NT (EC 50 ϭ 65 pM) and neurotensin (EC 50 ϭ 0.2 nM) in HEK293 cells recombinantly expressing NTR1 than in SK-N-SH cells naturally expressing NTR1 (EC 50 ϭ 0.75 nM for R-NT and EC 50 ϭ 2.7 nM for neurotensin). The EC 50 values for R-NT and neurotensin derived from the Ca 2ϩ assay in HEK293 cells recombinantly expressing NTR1 are also much lower than the IC 50 values (1 and 2 nM, respectively for R-NT and neurotensin) derived from the radioligand binding assays, although in the radioligand binding assay, the IC 50 value or K i values do not depend on the receptor expression level. The EC 50 values for the Ca 2ϩ assay could change depending on the receptor expression levels, because Ca 2ϩ assay is not an assay in equilibrium. For NTR2, we were able to detect receptor-specific binding; however, we were not able to show significant ligand-stimulated Ca 2ϩ mobilization in NTR2-expressing cells. One possible reason for the lack of Ca 2ϩ response in NTR2-expressing cells may be because of the poor coupling of NTR2 with G-proteins. We co-expressed G␣ 16 (a promiscuous G-protein that can couple with many different G-protein-coupled receptors and shift the signal transduction into Ca 2ϩ response) with NTR2 and hoped to see better Ca 2ϩ response in those cells. However, our results showed no improvement for NTR2-related signaling in our recombinant system. Although the reason for the lack of signaling in the recombinant NTR2expressing cell is not clear, our results are consistent with a previous report (48).
To characterize NTR3, we expressed NTR3 as a soluble receptor for radioligand binding assays. NTR3 has been reported to be expressed in the intracellular compartments and not on the cell surface, which makes the radioligand binding assay difficult. Previously, NTR3 binding assays have been reported using NTR3 solubilized from membrane preparations using detergent. In this report, by expression of the extracellular domain without the transmembrane and the intracellular domains and containing an endoplasmic reticulum retention signal sequence, we were able to express the recombinant NTR3 secreted into the medium. The C-terminal FLAG tag of the soluble NTR3 allows efficient capture of the soluble receptor/ligand complex simplifying the binding assays.