An L-RNA-based aquaretic agent that inhibits vasopressin in vivo

Werner G. Purschke, Dirk Eulberg, Klaus Buchner, Stefan Vonhoff, and Sven Klussmann*

NOXXON Pharma AG, Max-Dohrn-Strasse 8-10, 10589 Berlin, Germany

Edited by Jack Keene, Duke University, Durham, NC, and accepted by the Editorial Board February 2, 2006 (received for review November 7, 2005) A class of diuretic͞aquaretic agents based on mirror-image oligo- inadequately coupled to the decreased plasma osmolality (9), nucleotides (so-called Spiegelmers) has been identified. These but, as a consequence, patients show symptoms of chronic molecules directly bind and inhibit the neuropeptide vasopressin volume overload, , congestion, and increased sys- (AVP). AVP is the major regulatory component of body fluid temic vascular resistance (10). A central point in CHF is the homeostasis mediated through binding to the renal V2 receptor. progression of left ventricular dysfunction and chamber remod- Elevated plasma levels of AVP are implicated in several patholog- eling (11). AVP may be involved in these aspects of CHF because ical conditions, mainly cardiovascular diseases. In congestive heart AVP can induce structural changes in the myocardium through failure, AVP is part of a neuroendocrine imbalance that is respon- activation of the V1a receptor (12, 13). Elevated AVP levels are sible for progressive worsening of the disease. Employing in vitro also found in the syndrome of inappropriate antidiuretic hor- selection techniques, RNA aptamers that bind to the unnatural mone secretion (14) and late-stage liver cirrhosis (15), which D-configuration of AVP were isolated. The best aptamer displayed leads in both diseases to dilutional hyponatremia and hypoos- an affinity to D-AVP of Ϸ560 pM at 37°C. The corresponding molality. Through its mitogenic properties, AVP can also affect Spiegelmer, a 38-mer mirror-image oligonucleotide (L-RNA) termed the genesis of tumors. AVP and its receptors are expressed in all NOX-F37, inhibits vasopressin-dependent activation of V1a as well small cell lung carcinomas and breast cancer (16). In conclusion, as V2 receptors with IC50 values of 6.1 nM and 1 nM, respectively. a variety of severe human diseases exist where AVP seems to be NOX-F37 administered to healthy rats effectively neutralized AVP dysregulated and for which effective treatments are necessary. and increased dose-dependently for 24 h. The mode of Here we describe the development of a Spiegelmer that binds action was strictly aquaretic, i.e., the increase in urine volume was to AVP with high affinity and neutralizes its action. Spiegelmers not accompanied by an increase in electrolytes. These results are L-enantiomeric aptamers (17) that are identified through a clearly prove the in vivo efficacy of NOX-F37 and points out its SELEX-based (18) in vitro selection process that involves chiral potential as a drug in the treatment of diseases that are associated principles (19). First, the mirror-image configuration of AVP (all with body fluid overload. D-amino acids) is synthesized. A combinatorial nucleic acid library is screened to identify aptamers that bind to the all-D- aptamer ͉ congestive heart failure ͉ diureses ͉ water homeostasis ͉ AVP. Lastly, an individual aptamer sequence is synthesized in its Spiegelmer corresponding mirror-image configuration to give a Spiegelmer (L-oligonucleotide) that binds to the natural AVP. Spiegelmers he cyclic nonapeptide vasopressin (AVP) is a neurohormone have been identified already to bind to a variety of different Tthat is synthesized in the hypothalamus and stored in the peptide and protein targets (20–22). The mirror-image config- hypophysis from where it is released into the circulation (1). uration confers excellent biostability to Spiegelmers without the AVP is also known as antidiuretic hormone because its main need for further modifications. peripheral function is the maintenance of volume and osmolality We report the identification of a 38-mer RNA aptamer that of body fluids by regulating the free water reabsorption in the binds to AVP with an extraordinarily low dissociation constant kidneys. Under physiological concentrations, AVP acts vasocon- of 560 pM at 37°C. The corresponding Spiegelmer NOX-F37 strictive. AVP’s actions are mediated through specific G protein- inhibits AVP signaling in cell culture at the V2 and the V1a coupled receptors (2). The V receptor, mainly expressed on receptor with IC50 values of 1 and 6.1 nM, respectively. After 2 systemic administration NOX-F37 efficiently induces diuresis in cells of the basolateral membrane of the collecting duct in the a rat model, demonstrating its potential in neutralizing AVP’s kidneys, mediates the antidiuretic effect via enhanced cAMP action in vivo. levels (3). As a result, aquaporin channels are inserted into the cytoplasma membranes, and the transcription of aquaporin Results mRNA is induced. The V1a receptor, mainly expressed on D Identification of -AVP-Binding RNA Sequences. The in vitro selec- PHARMACOLOGY vascular smooth muscle cells and on cells of the myocardium, tion process to identify high-affinity RNA sequences binding to exerts vasoconstrictive effects through the second messengers D-AVP was divided into two parts: an automation-based stan- inositoltrisphosphate and diacylglycerol. Subsequently, intracel- ϩ dard selection (23) followed by a manually performed high- lular Ca2 is mobilized from the endoplasmic reticulum, and ϩ ϩ ϩ stringency protocol that included mutagenic steps within the Ca2 channels as well as Na ͞H exchangers are activated (4). amplification step. The selection was started with a library of In addition to vasoactive effects, binding of AVP to the V 1a Ϸ2.4 ϫ 1015 different RNA molecules and individual aptamer receptor also stimulates cell growth and proliferation in vascular sequences were identified after 16 selection rounds. Alignment smooth muscle cells through activation of the mitogen-activated protein kinase pathway (5). AVP release into the circulation is tightly regulated by osmo- Conflict of interest statement: All authors are employed by the biotechnology company receptors in the hypothalamus and baroreceptors in the heart NOXXON Pharma; none of the authors owns significant stock of the company. and large arteries. Whereas osmoreceptors respond to small This paper was submitted directly (Track II) to the PNAS office. J.K. is a guest editor invited changes (1%) in osmolality, baroreceptors respond only to by the Editorial Board. significant variations (10%) in blood volume and pressure (6, 7). Abbreviations: AVP, vasopressin; CHF, congestive heart failure; ITC, isothermal titration In congestive heart failure (CHF) patients, AVP plasma levels calorimetry; PEG, polyethylene glycol. are approximately two to three times higher than in healthy *To whom correspondence should be addressed. E-mail: [email protected]. subjects (8). It is not well understood why the AVP release is © 2006 by The National Academy of Sciences of the USA

www.pnas.org͞cgi͞doi͞10.1073͞pnas.0509663103 PNAS ͉ March 28, 2006 ͉ vol. 103 ͉ no. 13 ͉ 5173–5178 Downloaded by guest on September 28, 2021 Fig. 1. AVP binding sequences. (A) 134-A9 is the best sequence that resulted from automated selection whereas 157-B4 was obtained as the best sequence after applying a mutagenic high-stringency (mhs) protocol; other variants of the mhs protocol are listed in Supporting Appendix, which is published as supporting information on the PNAS web site. (B) Consensus motif of AVP-binding sequences composed of terminal helical regions, two G rich boxes and an AU-rich (6–9 nt) region. The four G stretches in box2 are shown in bold.

of the sequences revealed one family of relatives that differ by Although the sequences that resulted from the high stringency point mutations (data not shown). selection were not much different from the originally identified Truncation of the primer binding sites resulted in aptamers molecules, comparative ranking showed a significant improve- consisting of 47–49 nucleotides. The best binding sequence ment in binding affinity. In a competitive binding assay with an 134-A9 (Fig. 1A) was synthesized in its L-configuration. It excess of nonlabeled RNA, the best aptamer from the high inhibits L-AVP-induced cAMP formation in cell culture with an stringency selection, 157-B4, was determined to bind Ϸ12-fold IC50 of Ϸ50 nM (data not shown). To identify sequences with better to biotinylated D-AVP than 134-A9 (data not shown). even higher binding affinity, the enriched pool of round 16 was subjected to additional 13 selection rounds with increased Optimization of Aptamer 157-B4 by Site-Directed Modification with stringency, and a mutagenic amplification protocol was applied Hexaethyleneglycol. Because shorter oligonucleotides are easier in six selection rounds. Sequencing revealed further variants of and more cost effectively produced, we truncated the best the originally identified family. An alignment unveiled a 46- to binding aptamer 157-B4 (47-mer) further by replacing nucleo- 49-nt consensus motif (Fig. 1B; CLUSTAL X; ref. 24) that contains tides with a hexaethyleneglycol linker (26). The sequence vari- distinct secondary structure elements as predicted by a software ability of the AU-rich region was the most promising target to algorithm (ref. 25; Fig. 2). The 5Ј and 3Ј end of the sequences are introduce the linker (Fig. 2). Indeed, all different variants in self-complementary (Helix 1 and Helix 2) and seem to form a which up to seven nucleotides were replaced are still active (e.g., helix. Box 1, directly attached to the helix, is joined to Box 2 by F-11, a 43-mer). However, replacement of more than three a 6- to 9-nt-long region mainly composed of adenosine and nucleotides showed a slight decrease in affinity toward AVP. uridine residues. This AU-rich region shows the highest degree Therefore, the structure was ‘‘opened’’ at this site. Seven nucle- of diversity. Box 2, defined by four conserved stretches of two otides were deleted and new 5Ј and 3Ј ends were introduced, (one time), three (two times), or four (one time) guanosine whereas the structure was ‘‘closed’’ at the former helical 5Ј and nucleotides, is joined to the helix region at the 3Ј end. The 3Ј ends with the hexaethylene glycol linker. This operation guanosine stretches of Box 2 can potentially form a G quartet. preserved the high affinity to AVP. Moreover, the former 7-bp helix could be truncated additionally to a 6-bp helix so that the resulting RNA oligonucleotide consists of 38 nucleotides plus the linker. The final sequence was synthesized in its mirror-image configuration as Spiegelmer NOX-F37 (Fig. 2) and was further characterized in vitro and in vivo.

Affinity of NOX-F37 Toward AVP. The affinity of aptamer NOX-F37 was measured in a competitive pulldown assay that allowed the determination of affinity constants in solution (27). The con- centrations of radioactively labeled aptamer and biotinylated D-AVP were kept constant, whereas the concentration of the nonbiotinylated D-AVP competitor was varied. Assuming a 1:1 stoichiometry, the dissociation constant (Kd) of aptamer NOX- F37 to D-AVP at 37°C was determined to be 560 Ϯ 52 pM (Fig. 3A). Reducing AVP’s disulfide bridge (100 mM DTT) and, thus, destroying its 3D structure resulted in a total loss of aptamer binding, which underlines the oligonucleotides’ ability to specif- ically recognize a structural element rather than a short se- quence stretch of amino acids (data not shown). The affinity of the Spiegelmer NOX-F37 was determined by isothermal titration calorimetry (ITC), and the apparent Kd was calculated to be 1.7 Ϯ 0.8 nM at 37°C (Fig. 3B). The 3Ј PEGylated [40 kDa polyethylene glycol (PEG)] variant of NOX-F37, which was used for in vivo experiments (22, 28), displayed a Kd of 1.3 Ϯ 0.5 nM, which is in good agreement with Fig. 2. Secondary structure predictions of AVP-binding sequences. Site- the value determined for the unmodified molecule. directed modification of 157-B4 with a hexaethylene glycol linker (PEG18) led to a series of intermediate sequences with a PEG18 linker located in the AU-rich region (e.g., F-11). In the final sequence NOX-F37, the AU-rich region was Inhibition of L-AVP by NOX-F37 in Cell Culture. The potential of completely removed, the secondary structure was opened at this site and NOX-F37 to block receptor-mediated cell signaling of L-AVP closed with the PEG18 linker at the former 5Ј and 3Ј end. was tested with two different cell lines. Whereas LLC-PK1 cells

5174 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0509663103 Purschke et al. Downloaded by guest on September 28, 2021 Fig. 4. Inhibition of L-AVP mediated receptor activation by NOX-F37 (Œ) and its 40-kDa PEGylated variant (■). (A)V2 receptor was stimulated with 1 nM L-AVP in the presence of the indicated concentrations of Spiegelmer, the resulting cAMP formation was measured, and the IC50 (1 nM) was determined at the Spiegelmer concentration of half-maximal cAMP formation. (B)V1a receptor was stimulated with 5 nM L-AVP in the presence of the indicated concentrations of Spiegelmer, the resulting Ca2ϩ release was measured, and the IC50 (6.1 nM) was determined at the Spiegelmer concentration of half- maximal Ca2ϩ release. Fig. 3. Affinity of the aptamer and Spiegelmer NOX-F37 sequences to AVP. (A) Competitive pulldown assay of aptamer NOX-F37. The binding of radio- actively labeled aptamer to biotinylated D-AVP(C-Lys) was measured in the presence of the indicated concentrations of nonbiotinylated peptide [D- rats i.v. At different time points, urine volume, osmolality, AVP(n.b.)]. The dissociation constant (560 Ϯ 52 pM) was calculated (GRAFIT)by sodium concentration, and water intake was measured. using a 1:1 stoichiometry. (B) Calorimetric profile of Spiegelmer NOX-F37 Two hours after NOX-F37 administration, a dose-dependent binding to L-AVP. Results of a typical ITC measurement are shown. The increase in urine volume with a peak of 6.5 ml for the low dose dissociation constant was determined to be 1.7 Ϯ 0.8 nM assuming a 1:1 and peaks of 10.5 ml for the middle and high dose was observed stoichiometry. (Fig. 5A). These effects are specific to NOX-F37, because rats treated with vehicle or with control Spiegelmer showed a urine volume of only Ϸ1 ml in this time interval. For the low and (expressing V receptors) respond with an increase of intracel- 2 middle dose, the effect of NOX-F37 decreased in the following lular cAMP release, A7r5 smooth muscle cells (expressing V1a 2ϩ time interval (2–4 h), and the urine volume reached the level of receptors) mobilize intracellular Ca after L-AVP stimulation. the control animals in the third interval (4–6 h). However, the The inhibitory potency of NOX-F37 was determined by using highest dose of NOX-F37 induced a very strong response not a constant L-AVP stimulus in the EC50 range and increasing only in the first (0–2 h) and second (2–4 h), but also in the third concentrations of NOX-F37. The NOX-F37 concentration at time interval (4–6 h). In the final time interval (6–24 h), no PHARMACOLOGY 50% inhibition (IC50) was deduced from the respective dose– major differences between the groups were detectable. response curve and calculated to be 1 nM for the V2 receptor and Concomitant to the increase in urine volume, the osmolality 6.1 nM for the V1a receptor, respectively (Fig. 4). The inhibition and the sodium concentration of the urine decreased dose was specific because a nonfunctional Spiegelmer that does not dependently after NOX-F37 administration (Fig. 5 B and C). bind to AVP did not show any inhibitory effect in the test systems These effects seem to persist longer than the increase in urine (data not shown). The 3Ј PEGylated NOX-F37 showed essen- volume. After 6 h, all three doses showed a clear trend to a lower tially the same inhibition constant (Fig. 4), confirming the results urine osmolality and sodium concentration, and this effect was of the ITC measurements. The different IC50 values of NOX-F37 still significant in the final time interval 6–24 h for the highest in the two cell culture assays may be due to the divergent EC50 dose. The decrease of the electrolyte concentration is clear proof values. of the Spiegelmers’ ability to interfere with the animal’s AVP system. The urine volume was increased and diluted compared Effect of NOX-F37 on Diuresis. In vivo activity of the L-AVP binding to the control groups, indicating that the inhibition of AVP and inhibiting Spiegelmer NOX-F37 was tested in a rat diuresis results in a reduction of reabsorption of water in the collecting model. Three doses of the 3Ј PEGylated NOX-F37 (80, 400, and duct of the kidneys. 2,000 nmol͞kg), a nonfunctional control Spiegelmer (3Ј- During the study, the animals had free access to water. In PEGylated), and vehicle were administered to conscious, healthy parallel to the drastic increase in urine volume, the water intake

Purschke et al. PNAS ͉ March 28, 2006 ͉ vol. 103 ͉ no. 13 ͉ 5175 Downloaded by guest on September 28, 2021 dissociation constants at 37°C. Truncation and site-directed modifications, including the introduction of an internal (18 atoms) PEG linker spanning the former 5Ј and 3Ј end yielded the final candidate sequence Spiegelmer NOX-F37. NOX-F37 and its terminally PEGylated derivative (40 kDa) display dissociation constants of 1.7 nM and 1.3 nM (measured at 37°C), respectively, which is three orders of magnitude better than the dissociation constant of a previously published 55-mer mirror-image DNA molecule that binds to AVP with a Kd of 1.2 ␮M (measured at room temperature) (29). In both cases, the binding to nonmodi- fied AVP was determined in solution but with different assay formats (ITC vs. equilibrium dialysis). Elevated AVP levels have been correlated with severe dis- eases. The most important and widespread is CHF, a major cardiac disorder that is increasing in prevalence (30–32). The standard therapy comprises the medication with angiotensin- converting enzyme inhibitors, ␤-adrenoreceptor antagonists, and (33). Despite the effectiveness of standard medi- cations, the morbidity and mortality of patients suffering from CHF remains dissatisfactorily high (34). Evidence that points to a significant role of AVP in the progressive worsening of CHF has stimulated the development of AVP receptor antagonists. Various nonpeptidic AVP recep- tor antagonists have been discovered (35). Most advanced are , described as a specific V2 receptor antagonist, and , a dual V1a and V2 receptor antagonist. Both antag- onists are benzazepine derivatives that show inhibition constants to the human V2 and V1a receptors of 0.4 and 12.3 nM for tolvaptan and 1.1 and 6.3 nM for conivaptan, respectively (36, 37). In vivo efficacy was demonstrated for both antagonists in preclinical and clinical studies (7, 38–40). We have demonstrated in vivo efficacy of NOX-F37, a com- pound that directly binds and inhibits AVP. A single i.v. admin- istration of NOX-F37 increased the urine volume of normally hydrated conscious rats in a 2-h interval Ͼ10-fold and decreased urine osmolality and sodium concentration Ͼ5-fold at the same time. These aquaretic effects are comparable with reports on the activity of tolvaptan (36) and conivaptan (41) in rats. The advantage of aquaretic drugs in the treatment of CHF is the Fig. 5. Time course and dose–response of NOX-F37 on diuresis and water removal of excessive water from the body without causing serum intake in conscious rats. The different bars represent the following groups: electrolyte imbalances. A loss of electrolytes, which usually is open bars, vehicle (PBS); vertical hatched bars, 2,000 nmol͞kg control excited by conventional diuretics (31, 42) can induce or even ͞ Spiegelmer; left hatched bars, 80 nmol kg NOX-F37; right hatched bars, 400 worsen hyponatremia, one of the most important indicators of nmol͞kg NOX-F37; horizontal hatched bars, 2,000 nmol͞kg NOX-F37. Deter- mined parameters are as follows: urine volume (A), urine osmolality (B), urine cardiovascular mortality (43). Quite the contrary, the normal- sodium concentration (C), and water intake (D). Data are the mean Ϯ SEM (n ϭ ization of plasma sodium concentrations will result with 5) with the following exceptions: For the urine sodium concentration, only aquaretically acting compounds. Therefore, NOX-F37 may have one measured value could be generated in the vehicle group in the interval the potential for a more optimal therapy in diseases with 2–4 h, and in the control Spiegelmer group and the NOX-F37 group (80 increased water load. nmol͞kg) for the interval 4–6 h. *, P Ͻ 0.05; #, P Ͻ 0.01 indicate statistical In addition to its function in water homeostasis, a solid body significance of effects observed in the treated groups compared with the of evidence exists that the AVP system(s) may be involved in vehicle group. cardiac muscle hypertrophy and remodeling of the left ventricle, a key aspect in the progression of CHF (10). In this context, the importance of AVP signaling at the V receptor still remains to increased dose dependently in the first measured time interval 1a be elucidated. With respect to the low dissociation constant, (0–6 h) (Fig. 5D). In the time interval 6–24 h, the differences NOX-F37 might be even more effective in inhibiting AVP between the groups in water intake leveled off. signaling via the V1a receptor than the receptor antagonist Discussion conivaptan. For conivaptan, a Ki of 6.3 nM has been reported (44). Moreover, the IC50 value in a cell culture assay is higher for Spiegelmers and aptamers are 3D nucleic acid structures that are conivaptan (44) than for NOX-F37 (14.3 nM vs. 6.1 nM), which able to bind to target molecules conceptually similar to antibod- is further strengthened by the fact that the IC50 of NOX-F37 was ies. Unlike aptamers and other oligonucleotide-based structures, determined in an Ϸ4-fold less-sensitive cell assay (EC50 Ϸ 5 nM) Spiegelmers are nuclease-resistant because of their unnatural compared with conivaptan (EC50 Ϸ 1.3 nM; ref. 37). It can be L-enantiomeric configuration and, therefore, are biostable with- concluded that NOX-F37 might block AVP-induced V1a recep- out any further modification. This property makes them very well tor activation very efficiently and could reduce AVP-mediated suited for in vivo applications and, thus, for therapeutic use. remodeling in the heart of CHF patients accordingly. After a combination of high-stringency selection protocols NOX-F37 is a nucleic acid-based compound with potentially and mutagenization steps, we have identified very affine vaso- high therapeutic value for the treatment of severe AVP- pressin-specific RNA oligonucleotides that display picomolar conveyed diseases of which CHF remains the most important. To

5176 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0509663103 Purschke et al. Downloaded by guest on September 28, 2021 examine whether NOX-F37 can, in addition to its strong reaction without target peptide were conducted. Only the eluate aquaretic effects, reduce cardiac hypertrophy, further studies in from the most stringent binding reaction that still showed a appropriate animal models will be necessary. higher signal than the control reaction without peptide was processed further. In parallel, the RNA concentration was Materials and Methods decreased from 10 nM (round 17) to 330 pM (round 29). To Peptides and Nucleic Acids. All-L-AVP (CYFQNCPRG-NH2) and increase the complexity of the binding pool, the DNA templates all-D-AVP were obtained from Bachem (Bubendorf, Switzerland); for rounds 17, 18, 20, 22, 24, and 27 were amplified by mutagenic the latter was custom synthesized nonbiotinylated as well as PCR in the presence of MnCl2 and an unbalanced mix of dNTPs C-terminally biotinylated. The biotin group was linked to the according to Fromant et al. (45). The mutation rate of an D-peptide with a PEG linker D-AVP(C-Lys) (linker: amino- amplified reference aptamer was determined to be 6%. DNA ethyloxy-ethyloxy-acetylamino-ethyloxy-ethyloxy-acetyl) and a obtained from round 29 was cloned and sequenced. double-long chain linker D-AVP(LC-LC) (LC-LC: ␧-aminohexyl- ␧-aminohexyl), respectively. Oligonucleotides were synthesized at Site-Directed Modification and Truncation. The most affine oligo- NOXXON by using standard phosphoramidite chemistry. L- nucleotide was modified to identify shorter variants. Potential Phosporamidites and hexaethylene glycol phosporamidites (PEG- dispensable positions were bridged with a hexaethylene glycol linker) were from ChemGenes Corp. (Wilmington, MA). The linker. The resultant variants were analyzed with a ranking assay DNA library with 34 internal random positions had the sequence using competitive binding of a radioactively labeled reference 5Ј-GTGGAACCGACTCACCTGAGCGTGC-N34-GCACGCT- aptamer and an excess of nonlabeled variant aptamer to D-AVP. GCTGTTGTCTAAGCTCC-3Ј and was amplified with the forward primer 5Ј-TCTAATACGACTCACTATAGGAGCT- Determination of Binding Affinities. The affinity of aptamers to TAGACAACAGCAG-3Ј and the reverse primer 5Ј-GTG- D-AVP was measured by competition binding to biotinylated GAACCGACTCACCTGAG-3Ј. For in vivo application, the D-AVP vs. nonbiotinylated D-AVP in a pulldown assay at 37°C. AVP-binding Spiegelmer NOX-F37 (5Ј-GGGGUAGGG- After de- and renaturing, samples of 0.2 nM radioactively labeled CUUGGAUGGGUAGUACAC[PEG]18GUGUGCGUGGU-3Ј) D-RNA were incubated in selection buffer at 37°C with a and a nonfunctional control Spiegelmer (5Ј-UAAGGAAAC- constant amount of 0.8 nM biotinylated D-AVP and varying UCGGUCUGAUGCGGUAGCGCUGUGCAGAGCU-3Ј) amounts of nonbiotinylated D-AVP for 2–3 h to reach equilib- were modified with 40-kDa PEG at the 3Ј and 5Ј terminus, rium at low concentrations. Selection buffer was supplemented respectively (22). with 1 ␮g͞ml human serum albumin (Sigma-Aldrich) and 1 ␮g͞ml yeast RNA (Roche, Mannheim, Germany). After immo- In Vitro Selection. In vitro selection to identify aptamers to D-AVP bilization and washing without any competition, Ϸ10–15% was started with an automation-based standard protocol (23) binding to the biotinylated peptide was observed. To immobilize followed by 13 rounds of a mutagenic selection protocol. After peptide and peptide–aptamer complexes, a constant amount of annealing the forward primer, the ssDNA library (Ϸ2.4 ϫ 1015 25% SAulϩ matrix, preequilibrated in selection buffer (corre- molecules) was made double-stranded by using TaqDNA poly- sponds to 1 ␮l of 100% SAulϩ), was added and kept suspended merase (Invitrogen). The RNA library was transcribed and by mixing for 30 min. After detaching the supernatant and labeled with T7 RNA polymerase (Stratagene) from the dsDNA washing, the matrix-associated radioactivity was determined in a by using [␣-32P]-ATP and -GTP (Hartmann Analytic, Braun- scintillation counter (LS6500; Beckman Coulter). The percent- schweig, Germany). RNA with higher specific radioactivity was age of binding was plotted against the concentration of nonbio- obtained by transcription in the presence of 2.5 mM guanosine tinylated D-AVP and dissociation constants were calculated and 1 mM NTPs and subsequent labeling with [␥-32P]-ATP assuming a 1:1 stoichiometry (GRAFIT, Erithacus Software, (Hartmann Analytic) and T4 polynucleotide kinase (Invitrogen). Surrey, U.K.). Transcribed and labeled RNA was purified on denaturing (8 M The affinity of NOX-F37 to L-AVP was measured in solution urea) polyacrylamide gels. Before each selection step, RNA was by isothermal titration calorimetry on a VP-ITC instrument denatured in 20 mM Tris, pH 7.4͞150 mM NaCl͞5 mM KCl at (MicroCal, Northampton, MA) at 37°C (23). After de- and 94°C for 2 min, and was immediately placed on ice before it was renaturing of NOX-F37, Spiegelmer as well as L-AVP solutions supplemented with 1 mM MgCl2͞1 mM CaCl2͞0.1% Tween 20 were degassed and temperature adjusted in selection buffer for renaturing at 37°C. (without detergent). 3–5 ␮M NOX-F37 (1.4 ml) solution was Selection rounds 1–3 were carried out manually with two loaded to the instrument’s cell, and 0.25 ml of 25 ␮M L-AVP was complexities of the RNA library at 25 ␮M RNA͞25 ␮M peptide injected in 7.5-␮l portions at 6-sec duration for each injection (rounds 1–2) and 10 ␮M RNA͞10 ␮M peptide concentration and 5-min intervals between the injections for equilibration. (round 3). The library was incubated with biotinylated D-AVP at After each injection, the generated heat of the binding reaction 37°C, and complexes were immobilized on NeutrAvidin agarose was recorded. ITC binding curves were analyzed by using the PHARMACOLOGY (NAag) or Streptavidin ultralink plus (SAulϩ) (Pierce). Non- single-site binding equation in the MicroCal ORIGIN software bound RNA was separated by washing the matrix with selection package. buffer. Bound RNA was eluted, reverse transcribed (SuperScript II, Invitrogen), amplified by TaqDNA polymerase, and in vitro Inhibition of AVP Signaling in Cell Culture. The inhibitory activity of transcribed for the next selection round. A preincubation step L-AVP binding NOX-F37 was determined by using cell lines with pure matrix was performed starting with round 3. In every expressing the renal V2 receptor (LLC-PK1 from porcine kidney round, a different combination of matrix (NAag͞SAulϩ) and epithelium; ATCC-CL101) and the vascular V1a receptor (A7r5 peptide [D-AVP(C-Lys)͞D-AVP(LC-LC)] was used. The strin- from rat aortic smooth muscle; ATCC-CRL-144), respectively. gency was increased by reducing the peptide concentration to 20 LLC-PK1 cells were stimulated with 1 nM L-AVP, and cAMP nM. DNA from round 16 was cloned and sequenced (Agowa, formation was determined in the presence of different NOX-F37 Berlin). The same DNA was used as the starting library for the concentrations. Cells were grown overnight at a density of 6 ϫ mutagenic protocol. 104 per well in a 96-well plate in Medium 199 (Invitrogen) In the mutagenic protocol, the stringency was increased supplemented with 10% heat inactivated FCS͞4mML-alanyl- further by decreasing the peptide concentration from 5 nM L-glutamine͞50 units/ml penicillin͞50 ␮g/ml streptomycin at (round 17) to 0.4 pM (round 29). In each round, binding 37°C and 5% CO2 atmosphere. Twenty minutes before stimu- reactions with different peptide concentrations and a control lation, the cells were pretreated with 1 mM isobutyl-1-

Purschke et al. PNAS ͉ March 28, 2006 ͉ vol. 103 ͉ no. 13 ͉ 5177 Downloaded by guest on September 28, 2021 methylxanthin (IBMX). The IBMX-containing medium was ground fluorescence, and plotted against the concentration of replaced by the stimulation mix (L-AVP with varying concen- NOX-F37. IC50 values were drawn from the graphs. The EC50 trations of Spiegelmer in Hank’s balanced salt solution (HBSS)͞ value (5 nM) had been determined previously with varying 0.1% BSA͞1 mM IBMX), which had been preincubated at 37°C peptide concentrations as described (data not shown). for 30–60 min. After 30 min, the stimulation solution was replaced by lysis buffer from the cAMP-Screen System (Applied Diuresis Study in a Rat Model. The effect of NOX-F37 on diuresis Biosystems). The cAMP content of the cells was measured by was tested in a rat model (Aurigon Life Science, Tutzing, luminescence detection in a POLARstar Galaxy multidetection Germany). Male Sprague–Dawley rats (239–290 g; Elevage microplate reader (BMG Labtech, Offenburg, Germany). IC50 Janvier, Le Genest St. Isle, France) were acclimatized to the lab values were determined graphically by plotting the cAMP conditions for 9 days. Two days before the treatment, they were amount against the concentration of NOX-F37. The EC50 value placed in individual metabolic cages. Throughout the whole (1 nM) was determined previously by stimulation with varying experiment, animals had free access to water and food. The day concentrations of L-AVP without NOX-F37 (data not shown). before treatment, water intake and urine volume was measured. A7r5 cells were stimulated with 5 nM L-AVP, and the mobi- Groups of five animals were treated with NOX-F37 (80, 400, or ͞ lization of calcium was determined in the presence of different 2,000 nmol kg), a nonfunctional control Spiegelmer (2,000 ͞ ͞ NOX-F37 concentrations. Approximately 4 ϫ 104 cells per well nmol kg), or vehicle (PBS) by i.v. bolus injection (2 ml kg) into were grown overnight in a dark 96-well plate with clear bottom the tail vein. After administration, the urine was collected for (Greiner, Frickenhausen, Germany) in DMEM supplemented time intervals of 0–2 h, 2–4 h, 4–6 h, and 6–24 h. Water intake was determined for time intervals 0–6 h and 6–24 h. Urine with 10% heat-inactivated FCS͞4mML-alanyl-L-glutamine͞50 osmolality and sodium concentration was measured at Synlab.vet units/ml penicillin͞50 ␮g/ml streptomycin at 37°C and 5% CO2 atmosphere. Cells were washed once with 200 ␮l HBSSϩ (HBSS (Augsburg, Germany) by using a freezing point depression supplemented with 0.1% BSA͞5 mM probenecid͞20 mM osmometer and a flame photometer, respectively. The experi- mental design was approved by the local government. Hepes) and were preloaded for 60 min at 37°C with 50 ␮lof10 ␮M fluo-4 indicator dye solution͞0.08% pluronic 127 (both from Statistical Analysis. All data are expressed as mean Ϯ SEM. The Molecular Probes) in HBSSϩ. Cells were then washed three values of the diuresis study were analyzed by one-way ANOVA times with 180 ␮l HBSSϩ,90␮l HBSSϩ was added, and the between treated groups and the vehicle group. P Ͻ 0.05 was background fluorescence was measured at 485 nm excitation and considered as statistically significant. 520 nm emission in a POLARstar Galaxy multidetection micro- plate reader. Ten microliters of stimulation mix (L-AVP with ϩ We thank the NOXXON chemistry group for providing the oligonucle- varying concentrations of NOX-F37 in HBSS ), which had been otides, the biology group for carrying out the cell culture assays, preincubated at 37°C for 15–30 min, was added and the maxi- Christine Kaduk and Britta Wlotzka for helpful discussions, and mum fluorescence signal was measured, corrected for back- Jenny Fischer for critical reading of the manuscript.

1. Zimmerman, E. A. (1981) Adv. Biochem. Psychopharmacol. 28, 63–75. 24. Thompson, J. D., Gibson, T. J., Plewniak, F., Jeanmougin, F. & Higgins, D. G. 2. Thibonnier, M., Berti-Mattera, L. N., Dulin, N., Conarty, D. M. & Mattera, R. (1997) Nucleic Acids Res. 25, 4876–4882. (1998) Prog. Brain. Res. 119, 147–161. 25. Zuker, M. (2003) Nucleic Acids Res. 31, 3406–3415. 3. Lolait, S. J., O’Carroll, A. M., McBride, O. W., Konig, M., Morel, A. & 26. Hicke, B. J., Marion, C., Chang, Y. F., Gould, T., Lynott, C. K., Parma, D., Brownstein, M. J. (1992) Nature 357, 336–339. Schmidt, P. G. & Warren, S. (2001) J. Biol. Chem. 276, 48644–48654. 4. Thibonnier, M., Auzan, C., Madhun, Z., Wilkins, P., Berti-Mattera, L. & 27. Friguet, B., Chaffotte, A. F., Djavadi-Ohaniance, L. & Goldberg, M. E. (1985) Clauser, E. (1994) J. Biol. Chem. 269, 3304–3310. J. Immunol. Methods 77, 305–319. 5. Thibonnier, M., Conarty, D. M. & Plesnicher, C. L. (2000) Am. J. Physiol. 279, 28. Ostendorf, T., Kunter, U., Eitner, F., Loos, A., Regele, H., Kerjaschki, D., H2529–H2539. Henninger, D. D., Janjic, N. & Floege, J. (1999) J. Clin. Invest. 104, 913–923. 6. Morris, M. & Alexander, N. (1989) Hypertension 13, 110–114. 29. Williams, K. P., Liu, X.-H., Schumacher, T. N. M., Lin, H. Y., Ausiello, D. A., 7. Paranjape, S. B. & Thibonnier, M. (2001) Expert Opin. Investig. Drugs 10, Kim, P. S. & Bartel, D. P. (1997) Proc. Natl. Acad. Sci. USA 94, 11285–11290. 825–834. 30. Yamani, M. & Massie, B. M. (1993) Mayo Clin. Proc. 68, 1214–1218. 8. Goldsmith, S. R., Francis, G. S., Cowley, A. W., Jr., Levine, T. B. & Cohn, J. N. 31. Schrier, R. W. & Abraham, W. T. (1999) N. Engl. J. Med. 341, 577–585. (1983) J. Am. Coll. Cardiol. 1, 1385–1390. 32. Doggrell, S. A. & Brown, L. (2002) Expert Opin. Pharmacother. 3, 915–930. 9. Goldsmith, S. R., Francis, G. S. & Cowley, A. W., Jr. (1986) Am. J. Cardiol. 58, 33. Heart Failure Society of America (2000) Congest. Heart Fail. 6, 11–39. 295–299. 34. Russell, S. D. & DeWald, T. (2003) Am. J. Cardiovasc. Drugs 3, 13–20. 10. Lee, C. R., Watkins, M. L., Patterson, J. H., Gattis, W., O’Connor C. M., 35. Thibonnier, M., Coles, P., Thibonnier, A. & Shoham, M. (2001) Annu. Rev. Gheorghiade, M. & Adams, K. F., Jr. (2003) Am. Heart J. 146, 9–18. Pharmacol. Toxicol. 41, 175–202. 11. Sabbah, H. N. (2004) Heart Fail Rev. 9, 91–97. 36. Yamamura, Y., Nakamura, S., Itoh, S., Hirano, T., Onogawa, T., Yamashita, 12. Nakamura, Y., Haneda, T., Osaki, J., Miyata, S. & Kikuchi, K. (2000) Eur. T., Yamada, Y., Tsujimae, K., Aoyama, M., Kotosai, K., et al. (1998) J. Phar- J. Pharmacol. 391, 39–48. macol. Exp. Ther. 287, 860–867. 13. Fukuzawa, J., Haneda, T. & Kikuchi, K. (1999) Mol. Cell. Biochem. 195, 93–98. 37. Tahara, A., Saito, M., Sugimoto, T., Tomura, Y., Wada, K., Kusayama, T., 14. Baylis, P. H. (2003) Int. J. Biochem. Cell Biol. 35, 1495–1499. Tsukada, J., Ishii, N., Yatsu, T., Uchida, W. & Tanaka, A. (1998) Br. J. 15. Martin, P. Y. & Schrier, R. W. (1997) Kidney Int. Suppl. 59, S43–S49. Pharmacol. 125, 1463–1470. 16. North, W. G. (2000) Exp. Physiol. 85, 27S–40S. 38. Adis Data Information BV (2004) Drugs R D 5, 94–97. 17. Ellington, A. D. & Szostak, J. W. (1990) Nature 346, 818–822. 39. Doggrell, S. A. (2004) Curr. Opin. Investig. Drugs 5, 977–983. 18. Tuerk, C. & Gold, L. (1990) Science 249, 505–510. 40. Miyazaki, T., Yamamura, Y., Onogawa, T., Nakamura, S., Kinoshita, S., 19. Klussmann, S., Nolte, A., Bald, R., Erdmann, V. A. & Furste, J. P. (1996) Nat. Nakayama, S., Fujiki, H. & Mori, T. (2005) Endocrinology 146, 3037–3043. Biotechnol. 14, 1112–1115. 41. Wada, K., Tahara, A., Arai, Y., Aoki, M., Tomura, Y., Tsukada, J. & Yatsu, 20. Purschke, W. G., Radtke, F., Kleinjung, F. & Klussmann, S. (2003) Nucleic T. (2002) Eur. J. Pharmacol. 450, 169–177. Acids Res. 31, 3027–3032. 42. Golestaneh, L., Talreja, A. & Le Jemtel, T. H. (2004) Curr. Heart Fail. Rep. 1, 21. Vater, A., Jarosch, F., Buchner, K. & Klussmann, S. (2003) Nucleic Acids Res. 190–196. 31, e130. 43. Anderson, R. J., Chung, H. M., Kluge, R. & Schrier, R. W. (1985) Ann. Intern. 22. Wlotzka, B., Leva, S., Eschgfaller, B., Burmeister, J., Kleinjung, F., Kaduk, C., Med. 102, 164–168. Muhn, P., Hess-Stumpp, H. & Klussmann, S. (2002) Proc. Natl. Acad. Sci. USA 44. Tahara, A., Saito, M., Sugimoto, T., Tomura, Y., Wada, K., Kusayama, T., 99, 8898–8902. Tsukada, J., Ishii, N., Yatsu, T., Uchida, W. & Tanaka, A. (1998) Naunyn 23. Eulberg, D., Buchner, K., Maasch, C. & Klussmann, S. (2005) Nucleic Acids Schmiedebergs Arch. Pharmacol. 357, 63–69. Res. 33, e45. 45. Fromant, M., Blanquet, S. & Plateau, P. (1995) Anal. Biochem. 224, 347–353.

5178 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0509663103 Purschke et al. Downloaded by guest on September 28, 2021