Crystal structure of RNase A tandem enzymes and their interaction with the cytosolic ribonuclease inhibitor Ulrich Arnold, Franziska Leich*, Piotr Neumann, Hauke Lilie and Renate Ulbrich-Hofmann Department of Biochemistry and Biotechnology, Martin-Luther University Halle-Wittenberg, Halle, Germany Introduction About 10 million new cases of cancer are diagnosed worldwide annually, and cancer is the second most frequent cause of death after cardiovascular diseases. Treatment of unresectable cancer by traditional chemotherapy employs small molecules that interfere with DNA transcription; this type of treatment, Keywords crystal structure; proteolysis; ribonuclease inhibitor; stoichiometry; RNase A; tandem enzyme Correspondence U. Arnold, Department of Biochemistry and Biotechnology, Martin-Luther University Halle-Wittenberg, Kurt-Mothes Str. 3, 06120 Halle, Germany Fax: +49 345 5527303 Tel: +49 345 5524865 E-mail: ulrich.arnold@biochemtech. uni-halle.de Website: http://www.biochemtech. uni-halle.de/biotech Present addresses *Institute of Medical Immunology, Martin- Luther-University Halle-Wittenberg, Magde- burger Str. 2, 06097 Halle, Germany  Institute of Microbiology and Genetics, Georg-August University Go ¨ ttingen, Justus-von-Liebig-Weg 11, 37077 Go ¨ ttingen, Germany Database Structural data are available in the Protein Data Bank under the accession numbers 3MX8, 3MWR, and 3MWQ (Received 27 August 2010, revised 3 November 2010, accepted 8 November 2010) doi:10.1111/j.1742-4658.2010.07957.x Because of their ability to degrade RNA, RNases are potent cytotoxins. The cytotoxic activity of most members of the RNase A superfamily, how- ever, is abolished by the cytosolic ribonuclease inhibitor (RI). RNase A tan- dem enzymes, in which two RNase A molecules are artificially connected by a peptide linker, and thus have a pseudodimeric structure, exhibit remark- able cytotoxic activity. In vitro , however, these enzymes are still inhibited by RI. Here, we present the crystal structures of three tandem enzymes with the linker sequences GPPG, SGSGSG, and SGRSGRSG, which allowed us to analyze the mode of binding of RI to the RNase A tandem enzymes. Modeling studies with the crystal structures of the RI–RNase A complex and the SGRSGRSG-RNase A tandem enzyme as templates suggested a 1 : 1 binding stoichiometry for the RI–RNase A tandem enzyme complex, with binding of the RI molecule to the N-terminal RNase A entity. These results were experimentally verified by analytical ultracentrifugation, quanti- tative electrophoresis, and proteolysis studies with trypsin. As other dimeric RNases, which are comparably cytotoxic, either evade RI binding or poten- tially even bind two RI molecules, inactivation by RI cannot be the crucial limitation to the cytotoxicity of dimeric RNases. Abbreviations BS-RNase, bovine seminal RNase; ds-RNase A, domain-swapped RNase A; FAM-AUAA-TAMRA, 6-carboxyfluorescein-dArU(dA) 2 -6- carboxytetramethylrhodamine; RATE, RNase A tandem enzyme; RI, ribonuclease inhibitor. FEBS Journal 278 (2011) 331–340 ª 2010 The Authors Journal compilation ª 2010 FEBS 331 however, is often accompanied by severe side effects [1]. Antibody-based therapeutics, which target a variety of proteins (mostly on the cell surface), are much more selective, thereby reducing the systemic toxicity of the compounds. In the search for new cytotoxic therapeu- tics, RNases are considered to be powerful – nonmuta- genic – compounds by virtue of their RNA-digesting activity [1,2]. Whereas cell death was expected to be caused by ‘simple’ inhibition of the translation of the genetic information into proteins by unspecific RNA degradation, RNases were found to induce caspase- mediated apoptosis [3,4], probably by targeting non- coding RNAs [5]. Interestingly, members of the RNa- se A superfamily, which are basic proteins, show a specificity in their cytotoxic action for malignant cells [4]. However, it is still unclear whether the unusual intracellular trafficking of the endocytosed RNases in transformed cells [6] or the altered cell surface carbo- hydrate and lipid composition [7], which results in an increase in negative charge, and thus favors the bind- ing of the RNases [2], is responsible for the specific action. Unfortunately, the cytotoxicity of these RNases is limited by several factors at the cellular level [8], including restricted internalization into the cell, release from the endosomes, and inhibition by the cytosolic ribonuclease inhibitor (RI). RI is an abundant 50-kDa protein that binds the mammalian members of the RNase A superfamily extraordinarily tightly, with K D values in or below the picomolar range [9–12]. Conse- quently, RI evasion is considered to be crucial for cytotoxic efficacy [13]. In fact, OnconaseÔ (Tamir Bio- technology, Inc., Monmouth Junction, NJ, USA), an RNase A homolog from the Northern leopard frog, and the only naturally occurring dimeric RNase, bovine seminal RNase (BS-RNase), evade RI binding and are cytotoxic, as are genetically engineered RNase A variants with decreased affinity for RI [13–15]. Among the numerous approaches that were conceived to improve the cytotoxicity of mammalian RNases, the generation of pseudodimeric RNase A tandem enzymes (RATEs) proved to be very efficient [16]. In contrast to the noncytotoxic monomeric RNa- se A, RATEs show remarkable cytotoxicity (IC 50 val- ues ‡ 13 lm for K-562 cells [16]), as do the dimeric BS-RNase (IC 50 $ 1 lm for malignant SVT2 fibro- blasts [14]) and artificially domain-swapped RNase A (ds-RNase A) dimers (IC 50 $ 0.5 lm for HL-60 cells [17]). Like these, RATEs consist of two RNase entities. However, by means of gene duplication, RATEs consist of a single polypeptide chain [16]. In this way, dissociation of the RNase dimers is prevented. The dimeric structure was shown to be essential for a cyto- toxic effect of BS-RNase [18], and dimerization by domain swapping has been suggested to cause cytotox- icity by improved cellular uptake [19]. Tandemization considerably enhances endocytotic internalization into the cells [20], and potentially impedes binding by RI. Despite their clear cytotoxic action and their dimeric structure, however, RATEs were inhibited by RI in vitro at concentrations comparable to those that inactivate monomeric RNase A [16]. Whereas (at least C-terminally swapped) ds-RNase A dimers are sug- gested to bind two RI molecules in vitro, resulting in an inactive complex [19], BS-RNase evades RI binding [14,21]. To elucidate the mode of binding of RI and RATEs, the crystal structures of various RATEs differing in linker sequence length and amino acid composition (GPPG, SGSGSG, or SGRSGRSG) were solved. On the basis of these structures, modeling studies consider- ing the binding of one or two RI molecules to SGRSGRSG-RATE were performed, and were com- plemented by analytical ultracentrifugation, 2D elec- trophoresis, and proteolysis experiments. The studies revealed a 1 : 1 stoichiometry, with binding of the RI molecule to the N-terminal RNase A entity. Results Crystal structure of RATEs RATEs are composed of two RNase A molecules that are covalently linked by a peptide linker, resulting in a single polypeptide chain [16]. Three RATEs with linker sequences that differ in charge or flexibility (GPPG, SGSGSG, and SGRSGRSG) were selected for determination of the crystal structures. These RATEs were previously shown to be active [48–69%, relative to monomeric RNase A, with RNA as sub- strate; 1–5%, relative to monomeric RNase A, with 6-carboxyfluorescein-dArU(dA) 2 -6-carboxytetramethyl- rhodamine (FAM-AUAA-TAMRA) as substrate] and cytotoxic (IC 50 values between 12.9 lm and 40.0 lm with mammalian K-562 cells). All three variants showed thermodynamic stabilities similar to that of monomeric RNase A, and were inactivated by RI (K i £ 2.5 nm, as shown for SGRSGRSG-RATE) [16]. GPPG-RATE, SGSGSG-RATE and SGRSGRSG- RATE were successfully crystallized as described in Experimental procedures, all yielding the same crystal form. The crystals were highly isomorphic, and the structures could be solved at 2.10, 1.85 and 1.68 A ˚ resolution, respectively (Table S1). The structure of SGRSGRSG-RATE, which had been obtained at the highest resolution, was used as a model to refine the structures of the RATEs with the GPPG or SGSGSG Crystal structure of RNase A tandem enzymes U. Arnold et al. 332 FEBS Journal 278 (2011) 331–340 ª 2010 The Authors Journal compilation ª 2010 FEBS linker by the Fourier difference method. Comparison of the structures (Fig. S1) revealed no significant dif- ferences, with the exception of the linker region. As expected, the SGRSGRSG and GPPG linkers form loops between the two RNase A entities. In contrast, the SGSGSG linker could not be completely defined, indicating its particularly high flexibility. Interestingly, all RATEs showed the same alignment of the two RNase A entities. Consequently, the lengths and com- positions of the linker sequences used have no impact on their arrangement. For this reason, the following studies focused on SGRSGRSG-RATE, the structure of which was determined with the highest resolution (Fig. 1A; Table S1). Surprisingly, the orientation of the individual RNase A entities within the asymmetric unit, which were positioned almost perpendicular to each other, was found to be the same as in both the RNase A and RNase B (a glycosylated form of the enzyme) crystal structures [22,23]. The contact surface between the two RNase A enti- ties within one RATE molecule consists of a-helix III (residues 51–57), a b-strand (residues 61–63) and a loop region (residues 75–79) of the N-terminal entity, and a-helix I¢ (residues 4¢–12¢), a loop region (resi- dues 13¢–18¢) and the end of a-helix II¢ (residues 29¢–32¢) of the C-terminal entity (see Fig. 1B for num- bering of the amino acids in the RATEs). Interestingly, the crystal structure provides no clear indication of the reason for the decrease in catalytic activity as com- pared with RNase A. Neither of the two active sites is blocked by the interactions, and even though His12¢, which is an essential component of the active site [24], is part of helix I¢, its side chain points away from the interface. Nevertheless, the tandemization is undoubtedly the reason for the decreased activity, as concluded from the considerable increase in activity after liberation of the individual RNase A entities by trypsin (Table 1). Modeling of the RI–RATE complex As intensive attempts at the crystallization of the com- plex mixture of RI and RATEs have failed so far, models for RI–RATE complexes were derived from the crystal structure of SGRSGRSG-RATE (Fig. 1) and the porcine RI–RNase A complex [25] by superim- posing the RNase A structures with the program lsqman [26]. A B 124′1′1241 Linker 180° N-terminal RNase A entity C-terminal RNase A entity Fig. 1. Crystal structure and amino acid numbering of SGRSGRSG-RATE. The crystal structure of SGRSGRSG-RATE (A) was produced with PYMOL [42]. The N-terminal RNase A entity is shown in ruby, the C-terminal RNase A entity is shown in orange, and the linker is shown in blue. (B) The amino acids within the RATEs are numbered 1–124 for the N-terminal RNase A entity and 1¢–124¢ for the C-terminal RNase A entity; the amino acids of the linker, which differs in length between the various RATEs, are not numbered. Table 1. Catalytic efficiency of SGRSGRSG-RATE upon cleavage of the SGRSGRSG linker by trypsin in the absence or presence of RI. The activity assay was carried out as described in Experimental procedures. Sample k cat ⁄ K M (M )1 Æs )1 ) Before trypsin treatment After trypsin treatment RNase A (2.5 ± 0.5) · 10 7 Not determined SGRSGRSG-RATE (4.8 ± 0.8) · 10 6 (2.1 ± 0.6) · 10 7 SGRSGRSG-RATE +RI(1:1) (9.7 ± 1.5) · 10 5 (1.8 ± 0.4) · 10 7 SGRSGRSG-RATE + RI (1 : 250) (1.9 ± 0.6) · 10 5 (2.9 ± 0.5) · 10 5 U. Arnold et al. Crystal structure of RNase A tandem enzymes FEBS Journal 278 (2011) 331–340 ª 2010 The Authors Journal compilation ª 2010 FEBS 333 As shown in Fig. 2A, binding of an RI molecule to the N-terminal RNase A entity of SGRSGRSG-RATE is possible without restrictions. In contrast, superimpo- sition of the structure of the RI–RNase A complex and the C-terminal RNase A entity of SGRSGRSG- RATE indicates considerable steric hindrance (Fig. 2B). In accordance with this result, modeling of the complex between SGRSGRSG-RATE and two RI molecules (Fig. 2C) revealed severe steric clashes. The hydration shell, which was not considered in these modeling studies, may enhance the mismatch effect even further. Consequently, a 1 : 1 complex with bind- ing of the RI molecule to the N-terminal RNase A entity of the RATE seems most likely. In solution, however, the latitude of the RNase A entities may be different. As the molecules of mono- meric RNase A, which are highly ordered in the crystal [22] and are arranged like the RNase A entities in the RATEs, completely dissociate in solution, it seems likely that the close contact of the RNase A entities in the RATEs is also lost in solution. This assumption is supported by shape complementarity (sc) calculations for the two interacting molecular surfaces, performed with the program sc from the ccp4 suite [27]. The obtained sc value of 0.400 indicates weak binding of the two entities of the RATE. Therefore, the molecule might become more relaxed in solution, thereby enabling RI binding to both the N-terminal and the C-terminal RNase A entities. The distance between the C-terminus of the first and the N-terminus of the second RNase A entity was estimated to be $ 11.6 A ˚ in the crystal structures of all RATEs (Fig. S1), but it might increase up to 30 A ˚ (SGRSGRSG linker) when the peptide is fully stretched. In the closest modeled arrangement of two RI–RNase A complexes that is possible without steric clashes, the distance between the C-terminus of the RNase A in the first complex and the N-terminus of the RNase A in the second complex is about 6 A ˚ , i.e. less than the distance that would be covered by two (stretched) amino acids. As the RATEs that have been used so far [16,20] contain linker sequences of four (GPPG) to eight (SGRSGRSG) amino acids, a more relaxed conforma- tion of the RATEs in solution and, consequently, a stoichiometry for RI–RATE higher than 1 : 1, cannot be excluded by the modeling studies. However, the linker keeps the two RNase A entities close to each other not only in the crystal but also in solution, thereby supporting their interactions. As the intra- cellular concentration of macromolecules is close to the crystallization conditions [28], intensive interactions between the two RNase A entities are also likely in solution. Analytical ultracentrifugation Analytical ultracentrifugation was used to determine the stoichiometry within the RI–SGRSGRSG-RATE complex experimentally. Experiments with 1.5 lm RI and different concentrations of RATE (0–3 lm) yielded a maximum sedimentation velocity (s app )of 5.22 S at 1.5 lm SRGSGRSG-RATE (Fig. 3), i.e. at a molar ratio of 1 : 1 of RI and SGRSGRSG-RATE in the complex. Analysis of the RI-binding stoichiometry by gel electrophoresis For verification of the 1 : 1 stoichiometry in the RI– RATE complex, RNase A and SGRSGRSG-RATE were incubated in the presence of RI and separated by native PAGE (Fig. 4). Whereas a two-fold molar quantity of RI (Fig. 4, lanes 1 and 3) leaves unbound RI in both cases, no free RI was detectable at an equi- molar ratio (Fig. 4, lanes 2 and 4), confirming the 1 : 1 binding stoichiometry in the RI–RATE complex. ABC Fig. 2. Alignment of SGRSGRSG-RATE with the RI–RNase A complex. The crystal structure of the porcine RI–RNase A complex (Protein Data Bank entry: 1DFJ) was aligned with the N-terminal (A) or the C-terminal (B) RNase A entity of the tandem enzyme or with both RNa- se A entities (C). The RI molecule, which binds to the N-terminal RNase A entity, is shown in bright green, and the RI molecule, which binds to the C-terminal RNase A entity, is shown in pale green. The N-terminal and C-terminal RNase A entities are shown in ruby and orange, respectively, and the linker is shown in blue. Crystal structure of RNase A tandem enzymes U. Arnold et al. 334 FEBS Journal 278 (2011) 331–340 ª 2010 The Authors Journal compilation ª 2010 FEBS Additionally, the bands containing the RI–RNase A or RI–SGRSGRSG-RATE complexes (Fig. 4, lanes 2 and 4) were excised and subjected to SDS ⁄ PAGE, in which the noncovalent complexes dissociate (Fig. S2). The ratios of band intensities of RI–RNase A and RI–SGRSGRSG-RATE were (2.9 ± 0.5) : 1 and (1.3 ± 0.2) : 1, respectively. These values are slightly lower than calculated for a 1 : 1 stoichiometry from the molecular masses (3.6 and 1.7), because of the indi- vidual staining properties of RNase A, SGRSGRSG- RATE, and RI, but clearly indicate a 1 : 1 binding stoichiometry in the RI–RATE complex. Assessment of the functionality of the RNase A entities of SGRSGRSG-RATE by tryptic cleavage of the linker RNase A resists proteolytic attack by trypsin at 25 °C [29], whereas the flexible linker in SGRSGRSG-RATE is expected to be cleaved at the two potential cleavage sites (C-terminal to the two Arg residues). This specific cleavage should allow liberation of the RNase A enti- ties and thus allow evaluation of their activity loss by RI binding in the complex. In fact, cleavage by trypsin occurred exclusively within the linker sequence (Fig. 5), and the cleavage products were identified as RNase A-SGR and SG-RNase A by MS (13 986 and 13 829 Da in comparison with the theoretical values of 13 983 and 13 826 Da); that is, trypsin cleaves the lin- ker C-terminally to both Arg residues. Interestingly, the k cat ⁄ K M value of SGRSGRSG-RATE increased upon cleavage of the linker by trypsin to about that of RNase A (Table 1). From this activation, it can be unambiguously concluded that the RNase A entities within the RATEs are catalytically active and that the activity decrease [16] is a result of the tandemization. Accessibility of the linker of SGRSGRSG-RATE in the presence of RI The accessibility of the linker sequence in SGRSGRSG-RATE was evaluated in the presence of RI, first by SDS ⁄ PAGE (Fig. 5). As observed for the RNase A monomers, RI also resisted proteolytic attack by trypsin (not shown), whereas the linker sequence of SGRSGRSG-RATE was cleaved and Fig. 3. Analysis of the stoichiometry of the RI–SGRSGRSG-RATE complex by analytical ultracentrifugation. Formation of the complex between SGRSGRSG-RATE and RI (1.5 l M) was analyzed in the absence and in the presence of different amounts of the tandem enzyme (0–3 l M) in 0.1 M sodium phosphate buffer (pH 6.5) con- taining 2 m M dithiothreitol and 0.5 mM EDTA. The sedimentation velocity was analyzed at 40 000 r.p.m. (130 000 g) and 20 °C. RI - RATE complex RI - RNase A complex RI 12 34 Fig. 4. Analysis of the stoichiometry of the RI–SGRSGRSG-RATE complex by native PAGE. Lanes 1 and 2: complex of RNase A (100 pmol) with RI (200 and 100 pmol, respectively). Lanes 3 and 4: complex of SGRSGRSG-RATE (‘‘RATE’’, 100 pmol) with RI (200 and 100 pmol, respectively). Neither unbound RNase A nor SGRSGRSG-RATE is visible in the native PAGE gel [16]. RI RATE Fragments 1M2 Fig. 5. Analysis by SDS ⁄ PAGE of the impact of RI on the tryptic cleavage of SGRSGRSG-RATE. SGRSGRSG-RATE (‘‘RATE’’) was incubated with trypsin in the absence (lane 1) or in the presence (lane 2) of RI, as described in Experimental procedures. Lane M shows the molecular mass marker proteins lactalbumin (14.4 kDa), soybean trypsin inhibitor (21 kDa), carbonic anhydrase (30 kDa), ovalbumin (43 kDa), BSA (66 kDa), and phosphorylase b (97 kDa). The resulting cleavage products (see text) were combined and denoted as ‘fragments’. U. Arnold et al. Crystal structure of RNase A tandem enzymes FEBS Journal 278 (2011) 331–340 ª 2010 The Authors Journal compilation ª 2010 FEBS 335 yielded the same fragment pattern as in the absence of RI (Fig. 5). Next, the catalytic activity of the RI–SGRSGRSG- RATE complex was studied before and after treatment with trypsin (Table 1). Under the conditions applied, the activity of SGRSGRSG-RATE was decreased by 80% and 96% in the presence of RI at ratios of 1 : 1 and 1 : 250, respectively. After tryptic cleavage of the linker, considerable activity was regained at an equi- molar RI ⁄ SGRSGRSG-RATE ratio, indicating the release of an active RNase A entity from the RI–SGRSGRSG-RATE complex by trypsin. In con- trast, no significant increase in activity could be obtained at a 250-fold excess of RI, which proves that the released RNase A entity is, like the natural RNa- se A, sensitive to RI binding. Finally, the mode of binding of RI to SGRSGRSG- RATE was studied by cation exchange chromatogra- phy. Under the conditions applied, RNase A, SGRSGRSG-RATE and their complexes with RI elute at distinct, separated elution times (Fig. S3A). After tryptic treatment of the RI–SGRSGRSG-RATE com- plex, a new peak (peak 1 in Fig. S3A) emerged at an elution time similar to that of RNase A. Analysis by MS yielded a mass of 13 830 Da (Fig. S3B), which unambiguously identified the released RNase A entity as SG-RNase A (13 826 Da), i.e. the C-terminal RNase A entity, and thus proved binding of RI to the N-terminal RNase A entity of SGRSGRSG-RATE. Discussion The covalent linkage of two RNase A molecules, which are not cytotoxic in the monomeric form, by a peptide linker has been proven to endow cytotoxicity [16] simi- lar to that of the natural dimeric BS-RNase, in which the RNase entities are linked by two intermolecular disulfide bonds [14], or ds-RNase A dimers, in which the RNase A entities are noncovalently held together by swapping of the N-terminal or C-terminal ‘domains’ [19]. BS-RNase evades RI binding in vitro, which is regarded as a reason for its in vivo cytotoxicity. Upon reduction of the intermolecular disulfide bonds, however, the monomers slowly dissociate and become susceptible to inhibition by RI [21], losing their cyto- toxic properties [18]. In contrast to BS-RNase, ds- RNase A dimers show an affinity for RI comparable to that of RNase A, and the formed RI–ds-RNase A dimer complex apparently possesses a 2 : 1 stoichiome- try [19]. Despite slow dissociation of the ds-RNase A dimers, they proved to be cytotoxic as well [17], because of improved interaction of the dimers with the negatively charged cell membrane (dimerization increases the local concentration of RNase molecules on the cell surface), thereby favoring their endocytosis [19]. The stoichiometry of RI binding to RATEs and the role of the RI–RATE complex in the in vitro inacti- vation of RATEs have so far been obscure. The crystal structures of three different RATEs (Figs 1 and S1) revealed that the linkers do not con- strain the ability of the RNase A entities to adopt the same orientation in the crystal as monomeric RNase A. On the other hand, the decreased activity of the RATEs as compared with RNase A ([16] and Table 1) indicate a negative influence of the tandemization on the cata- lytic efficiency. The recovery of activity after proteolytic cleavage of the linker clearly proves that the decreased activity is a result of tandemization. Interestingly, the activities of both BS-RNase [30] and ds-RNase A dimers [19] are also decreased by about 60% and 70%, respectively, in comparison with RNase A. Modeling studies on the RI–RATE interactions on the basis of the crystal structure of SGRSGRSG- RATE (Fig. 2) suggest a 1 : 1 binding stoichiometry in the RI–RATE complex, with binding of the RI mole- cule to the N-terminal RNase A entity. These conclu- sions were unambiguously confirmed by experiment. The results of ultracentrifugation (Fig. 3) and electro- phoresis analyses (Fig. 4) clearly show that RATEs are able to bind one RI molecule only, corresponding to a 1 : 1 binding stoichiometry in the RI–RATE complex (at least at protein concentrations £ 6.67 lm; that is, the K D value for the C-terminal RNase A entity is ‡ 6.67 lm, whereas K D values of mammalian RI–RNase complexes are in the picomolar range or below [9,12]). By tryptic cleavage of the linker in the RI–SGRSGRSG-RATE complex and MS analysis of the released RNase A entity (Figs S3A,B), the sug- gested binding position of RI at the N-terminal RNa- se A entity was verified. The experimentally determined binding stoichiometry corroborates the ori- ginal idea in the design of RATEs [16]. As the intracel- lular concentration of RI is about 4 lm [31], sufficient activity may remain in vivo to explain the cytotoxicity of the tandem constructs, even though activity mea- surements in the presence of RI indicated a decrease in activity that was larger than expected (Table 1). More- over, tandemization has been shown to dramatically improve endocytosis efficiency [20]. In summary, the three types of dimeric RNase, which are comparably cytotoxic, differ fundamentally in their RI binding: BS-RNase binds no RI, ds- RNase A dimers supposedly bind two RI molecules, and RATEs bind one RI molecule. Therefore, RI evasion cannot be the pivotal determinant for cytotox- icity of dimeric RNase variants. Rather, improved Crystal structure of RNase A tandem enzymes U. Arnold et al. 336 FEBS Journal 278 (2011) 331–340 ª 2010 The Authors Journal compilation ª 2010 FEBS endocytosis in comparison with monomeric RNase A seems to be the decisive factor. Dimerization by a non- reducible covalent linkage, which prevents dissociation, renders RATEs superior to other types of RNase dimers as cytotoxic agents. Experimental procedures Proteins and chemicals RNase A from Sigma (Taufkirchen, Germany) was purified on a SOURCE S FPLC system (Amersham Biosciences, Uppsala, Sweden). Growth media were from Difco Labora- tories (Detroit, MI, USA). Escherichia coli strain BL21(DE3) was from Stratagene (Heidelberg, Germany). FAM-AUAA-TAMRA was from Metabion International AG (Martinsried, Germany). All other chemicals were of the purest grade commercially available. Expression, renaturation and purification of RATEs The experimental procedures for expression and renatur- ation of RATEs have been described previously [16]. The proteins were purified on a SOURCE S FPLC system (50 mm Tris ⁄ HCl, pH 7.5, with a linear gradient of 0–500 mm NaCl). Expression and purification of RI The plasmid pET–22b(+), which contains the gene for RI, was a gift from R. T. Raines (UW-Madison, WI, USA). RI was purified by procedures similar to those described previ- ously [12]. Briefly, the plasmid was transformed into E. coli BL21(DE3) cells, and a single colony was used to inoculate LB medium (25 mL) containing ampicillin (200 lgÆmL )1 ). An overnight culture, grown at 37 °C and 180 r.p.m. for 12 h, was used to inoculate cultures of TB medium (1 L) containing ampicillin (200 lgÆmL )1 ). These cultures were grown at 37 °C and 180 r.p.m. up to D 600 nm ‡ 3.0. Expres- sion of the ri gene was induced by addition of isopropyl thio-b-d-galactoside to a final concentration of 0.5 mm, and the cultures were grown at 15 °C and 120 r.p.m. for 24 h. Bacteria were harvested by centrifugation (7000 g for 15 min), and resuspended in 200 mL of 20 mm Tris ⁄ HCl buffer (pH 7.8) containing 10 mm EDTA, 10 mm dith- iothreitol, 100 mm NaCl, and 0.04 mm phenylmethanesulfo- nyl fluoride. After lysis of the bacteria by three passages through a Gaulin Lab 40 (APV, Lu ¨ beck, Germany), cell debris was removed by centrifugation (48 000 g, 20 min, 4 °C). The supernatant after the centrifugation step, which contained the soluble RI, was loaded onto an RNase A affinity column. For affinity chromatography, RNase A was attached covalently to the resin of a 5-mL HiTrap NHS-ester column (Amersham Biosciences), follow- ing the manufacturer’s protocol. RI was eluted in 100 mm sodium acetate buffer (pH 5.0) containing 3 m NaCl, 10 mm dithiothreitol, and 1 mm EDTA, after extensive washing with 50 mm KH 2 PO 4 buffer (pH 6.4) containing 1 m NaCl, 10 mm dithiotheitol, and 1 mm EDTA. The pro- tein eluted from the RNase A affinity resin was dialyzed for 16 h against 10 L of 20 mm Tris ⁄ HCl buffer (pH 7.5) containing 10 mm dithiothreitol and 1 mm EDTA, and purified further by anion exchange chromatography with a Mono Q column (Amersham Biosciences; 20 mm Tris ⁄ HCl, pH 7.5, containing 10 mm dithiothreitol and 1 mm EDTA, with a linear gradient of 0–1 m NaCl). The purity of the eluted RI was determined to be > 99% by SDS ⁄ PAGE (data not shown). Crystallization Crystals of RATEs were obtained by hanging-drop vapor diffusion over 30% (w ⁄ v) poly(ethylene glycol)-8000 con- taining 200 mm (NH 4 ) 2 SO 4 . The hanging-drop solution contained a mixture of purified RATE (2 lL; 10 mgÆmL )1 in 10 mm Tris ⁄ HCl, pH 7.0) and crystallization solution (2 lL). Diffraction-quality crystals grew within 6 days at 13 °C to a size of 0.1 · 0.1 · 0.1 mm. Structure determination A redundant dataset of a RATE crystal was collected at 100 K on a flash-frozen crystal by transferring the crystal rapidly into a cryoprotectant containing mother liquor made up to 20% (v ⁄ v) glycerol. The crystals diffracted up to 1.68 A ˚ resolution with Cu Ka radiation (k = 1.5418 A ˚ ), with a rotating-anode source (RA Micro 007; RigakuMSC, Sevenoaks, Kent, UK) and image plate detector (R-AXIS IV++; RigakuMSC). Oscillation photographs were integrated, merged and scaled with mosflm and scala, respectively (details are given in Table S1), from the ccp4 suite [32]. The crystals of RATEs crystallize in space group C2 (Table S1). The structure was determined by the molecular replacement method with data between 20 and 2.5 A ˚ , using the RNase A structure (Protein Data Bank code: 1SRN [33]) as the search model, with phaser [34]. The molecular replace- ment search solution showed two RNase molecules (corre- sponding to one tandem enzyme molecule) occupying the asymmetric unit (Matthews coefficient of 1.86 A ˚ 3 ⁄ Da, corre- sponding to 33.3% solvent content). The structure was man- ually rebuilt and verified against a simulated annealing omit map as well as SIGMA A -weighted [35] difference Fourier maps, with the o and coot programs [36,37]. The final refine- ment was performed with refmac from the ccp4 suite [32], with TLS parameterization. Both cns and refmac used the same R free set (randomly chosen 5% of the reflections). Six U. Arnold et al. Crystal structure of RNase A tandem enzymes FEBS Journal 278 (2011) 331–340 ª 2010 The Authors Journal compilation ª 2010 FEBS 337 residues displaying dual conformations were modeled. The stereochemistry of the structure was assessed with procheck [38] (Ramachandran plot statistics: $88% of all amino acids in favored regions, and $12% in allowed regions). Analytical ultracentrifugation For analysis of the stoichiometry of the RI–SGRSGRSG- RATE complex, RI (1.5 lm) and SGRSGRSG-RATE (0–3 lm) were incubated in 0.1 m sodium phosphate buffer (pH 6.5) containing 2 mm dithiothreitol and 0.5 mm EDTA. Formation of the complex between SGRSGRSG- RATE and RI was studied with a Beckman Optima XL-A analytical ultracentrifuge (Palo Alto, CA, USA), using an An-50Ti rotor. The experiment was carried out at 40 000 r.p.m. (130 000 r.p.m.) and 20 °C for calculation of the sedimentation velocity (absorption at 280 nm). Data were analyzed with the software provided by Beckman Instruments (Palo Alto, CA, USA). Analysis of the RI-binding stoichiometry by gel electrophoresis To analyze the stoichiometry of the complex between RI and RATEs, 100 or 200 pmol of RI was incubated either with 100 pmol of SGRSGRSG-RATE or with 100 pmol of RNase A at 25 °C for 15 min in 15 lL of 0.1 m sodium phosphate buffer (pH 6.5) containing 2 mm dithiothreitol and 0.5 mm EDTA. After addition of 5 lL of 125 mm Tris ⁄ HCl (pH 6.8) containing 15% (v ⁄ v) glycerol and 0.02% (w ⁄ v) bromophenol blue, electrophoretic separation was performed on 10% (w ⁄ v) polyacrylamide gels with 40 mm Tris ⁄ acetate (pH 8.0) containing 1 m m EDTA as running buffer. The gels were stained with Coomassie Bril- liant Blue R250, and the bands of the RI–RNase A or RI– SGRSGRSG-RATE complexes were excised. The gel pieces were incubated in 25 lL of 250 mm Tris ⁄ HCl (pH 8.0) con- taining 5% (w ⁄ v) SDS, 50% (v ⁄ v) glycerol, 0.02% (w ⁄ v) bromophenol blue and 5% (v ⁄ v) 2-mercaptoethanol for 15 min at 25 °C, and analyzed by SDS ⁄ PAGE. Electropho- resis was carried out on a Midget Electrophoresis Unit (Hoefer, San Francisco, CA, USA) according to Laemmli [39], using 5% and 12% (w ⁄ v) acrylamide for stacking and separating gels. The gels were stained with Coomassie Bril- liant Blue R250. After destaining, the gels were evaluated densitometrically at 560 nm with a CD 60 densitometer (Desaga, Heidelberg, Germany). Proteolysis After preincubation of SGRSGRSG-RATE (3.6 lm) in the absence or presence of RI (7.2 lm)at25°C for 15 min in 15 lLof50mm Tris ⁄ HCl buffer (pH 8.0) containing 2 mm dithiothreitol, the reaction was started by addition of 1.5 lL of trypsin (1 · 10 )2 mgÆmL )1 in 50 mm Tris ⁄ HCl buffer, pH 8.0, containing 10 mm CaCl 2 ). After 15 min (absence of RI) or 1 h (presence of RI) at 25 °C, the reaction was stopped by addition of 5 lL of phenylmethanesulfonyl fluo- ride (50 mm, dissolved in 2-propanol). The samples were dried under nitrogen and analyzed by SDS ⁄ PAGE as described above. Alternatively, SGRSGRSG-RATE was incubated with RI (1.5 lm each in 200 lL of 100 mm sodium phosphate buffer, pH 6.55, containing 2 mm dithiothreitol) at 25 °C for 15 min. Then, 2 lL of trypsin (0.1 mgÆmL )1 in 50 mm Tris ⁄ HCl buffer containing 10 mm CaCl 2 , pH 8.0) was added. After 1 h at 25 °C, the sample was subjected to cation exchange chromatography on a SOURCE S FPLC system (see above). As references, RNase A and SGRSGRSG-RATE were analyzed in the absence and presence of RI as well. Manually collected fractions were analyzed by MALDI MS (Reflex; Bruker-Franzen, Bremen, Germany) after desalting of the protein samples with ZipTip pipette tips (Millipore, Schwalbach, Germany). Activity assay For analysis of the catalytic activity of SGRSGRSG-RATE in comparison with that of RNase A, a fluorometric assay employing the low molecular mass substrate FAM-AUAA-T AMRA was used [40,41]. Values of k cat ⁄ K M were determined in 100 mm 2-(N-morpholino)ethanesulfonic acid-NaOH (pH 6.0) containing 100 mm NaCl. SGRSGRSG-RATE (100 pm, corresponding to 200 pm RNase A entities) was incubated in the absence or in the presence of RI (100 pm or 25 nm) for 5 min, and the reaction was started by addition of FAM-AUAA-TAMRA (final concentration 50 nm). After a distinct time interval, 2 lL of trypsin (0.5 mgÆmL )1 ) was added. The reaction was finalized by addition of 1 lLof RNase A (73 lm, for complete cleavage of all substrate). Flu- orescence emission at 515 nm was followed on a Fluoro- Max-2 spectrometer (Jobin Yvon) upo n excitation at 490 nm. Values of k cat ⁄ K M were determined from the equation k cat =K M ¼ DF È ðF initial À F final Þ½E É where DF is the change in the fluorescence signal of the sam- ple per second, F initial is the signal after addition of substrate, F final is the signal after cleavage of all substrate by addition of RNase A, and [E] is the concentration of enzyme. Acknowledgements The authors are grateful to R. T. Raines (University of Wisconsin-Madison, WI, USA) for providing the plas- mid for RI. The Land Sachsen-Anhalt is gratefully acknowledged for supporting this work (3537C ⁄ 0903T). A. Schierhorn, Martin-Luther University, Halle, Germany, is acknowledged for MS measurements. Crystal structure of RNase A tandem enzymes U. Arnold et al. 338 FEBS Journal 278 (2011) 331–340 ª 2010 The Authors Journal compilation ª 2010 FEBS References 1 Arnold U & Ulbrich-Hofmann R (2006) Natural and engineered ribonucleases as potential cancer therapeu- tics. Biotechnol Lett 28, 1615–1622. 2 Makarov AA & Ilinskaya ON (2003) Cytotoxic ribo- nucleases: molecular weapons and their targets. FEBS Lett 540, 15–20. 3 Iordanov MS, Wong J, Newton DL, Rybak SM, Bright RK, Flavell RA, Davis RJ & Magun BE (2000) Differ- ential requirement for the stress-activated protein kina- se ⁄ c-Jun NH(2)-terminal kinase in RNAdamage- induced apoptosis in primary and in immortalized fibro- blasts. Mol Cell Biol Res Commun 4, 122–128. 4 Spalletti-Cernia D, Sorrentino R, Di Gaetano S, Arci- ello A, Garbi C, Piccoli R, D’Alessio G, Vecchio G, Laccetti P & Santoro M (2003) Antineoplastic ribonuc- leases selectively kill thyroid carcinoma cells via cas- pase-mediated induction of apoptosis. J Clin Endocrinol Metab 88, 2900–2907. 5 Ardelt B, Ardelt W & Darzynkiewicz Z (2003) Cyto- toxic ribonucleases and RNA interference (RNAi). Cell Cycle 2, 22–24. 6 Bracale A, Spalletti-Cernia D, Mastronicola M, Cast- aldi F, Mannucci R, Nitsch L & D’Alessio G (2002) Essential stations in the intracellular pathway of cyto- toxic bovine seminal ribonuclease. Biochem J 362, 553– 560. 7 Ran S, Downes A & Thorpe PE (2002) Increased expo- sure of anionic phospholipids on the surface of tumor blood vessels. Cancer Res 62, 6132–6140. 8 Arnold U (2008) Aspects of the cytotoxic action of ribonucleases. Curr Pharm Biotechnol 9, 161–168. 9 Lee FS, Shapiro R & Vallee BL (1989) Tight-binding inhibition of angiogenin and ribonuclease A by placen- tal ribonuclease inhibitor. Biochemistry 28, 225–230. 10 Vicentini AM, Kieffer B, Matthies R, Meyhack B, Hemmings BA, Stone SR & Hofsteenge J (1990) Protein chemical and kinetic characterization of recombinant porcine ribonuclease inhibitor expressed in Saccharomy- ces cerevisiae. Biochemistry 29, 8827–8834. 11 Dickson KA, Haigis MC & Raines RT (2005) Ribonu- clease inhibitor: structure and function. Prog Nucleic Acid Res Mol Biol 80, 349–374. 12 Johnson RJ, McCoy JG, Bingman CA, Phillips GN Jr & Raines RT (2007) Inhibition of human pancreatic ribonuclease by the human ribonuclease inhibitor protein. J Mol Biol 368, 434–449. 13 Rutkoski TJ & Raines RT (2008) Evasion of ribonucle- ase inhibitor as a determinant of ribonuclease cytotoxic- ity. Curr Pharm Biotechnol 9, 185–189. 14 Antignani A, Naddeo M, Cubellis MV, Russo A & D’Alessio G (2001) Antitumor action of seminal ribonu- clease, its dimeric structure, and its resistance to the cyto- solic ribonuclease inhibitor. Biochemistry 40, 3492–3496. 15 Rutkoski TJ, Kurten EL, Mitchell JC & Raines RT (2005) Disruption of shape-complementarity markers to create cytotoxic variants of ribonuclease A. J Mol Biol 354, 41–54. 16 Leich F, Ko ¨ ditz J, Ulbrich-Hofmann R & Arnold U (2006) Tandemization endows bovine pancreatic ribonu- clease with cytotoxic activity. J Mol Biol 358, 1305– 1313. 17 Libonati M, Gotte G & Vottariello F (2008) Novel biological actions acquired by ribonuclease A through oligomerization. Curr Pharm Biotechnol 9 , 200–209. 18 Mastronicola MR, Piccoli R & D’Alessio G (1995) Key extracellular and intracellular steps in the antitumor action of seminal ribonuclease. Eur J Biochem 230, 242– 249. 19 Naddeo M, Vitagliano L, Russo A, Gotte G, D’Alessio G & Sorrentino S (2005) Interactions of the cytotoxic RNase A dimers with the cytosolic ribonuclease inhibi- tor. FEBS Lett 579, 2663–2668. 20 Leich F, Sto ¨ hr N, Rietz A, Ulbrich-Hofmann R & Arnold U (2007) Endocytotic internalization as a cru- cial factor for the cytotoxicity of ribonucleases. J Biol Chem 282, 27640–27646. 21 Murthy BS & Sirdeshmukh R (1992) Sensitivity of monomeric and dimeric forms of bovine seminal ribo- nuclease to human placental ribonuclease inhibitor. Bio- chem J 281, 343–348. 22 Leonidas DD, Shapiro R, Irons LI, Russo N & Acharya KR (1997) Crystal structures of ribonucle- ase A complexes with 5¢-diphosphoadenosine 3¢-phos- phate and 5¢-diphosphoadenosine 2¢-phosphate at 1.7 A resolution. Biochemistry 36, 5578–5588. 23 Williams RL, Greene SM & McPherson A (1987) The crystal structure of ribonuclease B at 2.5-A resolution. J Biol Chem 262, 16020–16031. 24 Raines RT (2004) Active site of ribonuclease A. In Nucleic Acids and Molecular Biology (Zenkova MA ed), pp 19–32. Springer-Verlag, Berlin. 25 Kobe B & Deisenhofer J (1996) Mechanism of ribonu- clease inhibition by ribonuclease inhibitor protein based on the crystal structure of its complex with ribonucle- ase A. J Mol Biol 264, 1028–1043. 26 Kleywegt GJ & Jones TA (1994) Halloween masks and bones. In From First Map to Final Model (Bailey S, Hubbard R & Waller D eds), pp. 59–66. SERC Dares- bury Laboratory, Warrington. 27 Lawrence MC & Colman PM (1993) Shape complemen- tarity at protein ⁄ protein interfaces. J Mol Biol 234, 946–950. 28 Zimmerman SB & Minton AP (1993) Macromolecular crowding: biochemical, biophysical, and physiological consequences. Annu Rev Biophys Biomol Struct 22, 27–65. 29 Arnold U, Ru ¨ cknagel KP, Schierhorn A & Ulbrich- Hofmann R (1996) Thermal unfolding and proteolytic U. Arnold et al. Crystal structure of RNase A tandem enzymes FEBS Journal 278 (2011) 331–340 ª 2010 The Authors Journal compilation ª 2010 FEBS 339 susceptibility of ribonuclease A. Eur J Biochem 237, 862–869. 30 Opitz JG, Ciglic MI, Haugg M, Trautwein-Fritz K, Raillard SA, Jermann TM & Benner SA (1998) Origin of the catalytic activity of bovine seminal ribonuclease against double-stranded RNA. Biochemistry 37, 4023– 4033. 31 Haigis MC, Kurten EL & Raines RT (2003) Ribonucle- ase inhibitor as an intracellular sentry. Nucleic Acids Res 31, 1024–1032. 32 Collaborative Computational Project Number4 (1994) The CCP4 suite: programs for protein crystallography. Acta Crystallogr D 50, 760–763. 33 Martin PD, Doscher MS & Edwards BF (1987) The refined crystal structure of a fully active semisynthetic ribonuclease at 1.8-A resolution. J Biol Chem 262, 15930–15938. 34 McCoy AJ, Grosse-Kunstleve RW, Storoni LC & Read RJ (2005) Likelihood-enhanced fast translation func- tions. Acta Crystallogr D 61, 458–464. 35 Read RJ (1986) Improved Fourier coefficients for maps using phases from partial structures with errors. Acta Crystallogr A 42, 140–149. 36 Jones TA, Zou JY, Cowan SW & Kjeldgaard M (1991) Improved methods for building protein models in elec- tron density maps and the location of errors in these models. Acta Crystallogr A 47, 110–119. 37 Emsley P & Cowtan K (2004) Coot: model-building tools for molecular graphics. Acta Crystallogr D 60, 2126–2132. 38 Laskowski RA, MacArthur MW, Moss DS & Thornton JM (1993) PROCHECK: a program to check the ste- reochemical quality of protein structures. J Appl Crys- tallogr 26, 283–291. 39 Laemmli UK (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227, 680–685. 40 Kelemen BR, Klink TA, Behlke MA, Eubanks SR, Leland PA & Raines RT (1999) Hypersensitive substrate for ribonucleases. Nucleic Acids Res 27, 3696– 3701. 41 Tam A, Arnold U, Soellner MB & Raines RT (2007) Protein prosthesis: 1,5-disubstituted[1,2,3]triazoles as cis-peptide bond surrogates. J Am Chem Soc 129, 12670–12671. 42 DeLano WL (2002) The PyMOL Molecular Graphics System. DeLano Scientific, Palo Alto, CA, USA. http:// www.pymol.org. Supporting information The following supplementary material is available: Fig. S1. Superimposition of the crystal structures of the RATEs with the linker sequences GPPG, SGSGSG, and SGRSGRSG. Fig. S2. Analysis of the stoichiometry of the RIÆSGRSGRSG-RATE complex by native PAGE. Fig. S3. Analysis of the tryptic cleavage in the RIÆSGRSGRSG-RATE complex by FPLC and MALDI mass spectrometry. Table S1. Crystallographic data processing and refine- ment statistics for the RATEs with the linker sequences GPPG, SGSGSG, and SGRSGRSG. This supplementary material can be found in the online version of this article. Please note: As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer-reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors. Crystal structure of RNase A tandem enzymes U. Arnold et al. 340 FEBS Journal 278 (2011) 331–340 ª 2010 The Authors Journal compilation ª 2010 FEBS . that the RNase A entities within the RATEs are catalytically active and that the activity decrease [16] is a result of the tandemization. Accessibility of. allowed us to analyze the mode of binding of RI to the RNase A tandem enzymes. Modeling studies with the crystal structures of the RI RNase A complex and