DDBJ/EMBL/GenBank accession no. U88827
ABSTRACT
The human eosinophil-derived neurotoxin (hEDN) is a secretory effector protein from eosinophilic leukocytes that is a member of the ribonuclease A (RNase A) family of ribonucleases. EDN is a rapidly evolving protein, accumulating non-silent mutations at a rate exceeding those of most other functional coding sequences studied in primates. Although all primate EDNs retain the structural and functional residues known to be prerequisites for ribonuclease activity, we have shown previously that recombinant EDN derived from a New World monkey sequence (Saguinus oedipus) had significantly less catalytic activity than the human (hEDN) ortholog. In this work, we have prepared recombinant proteins from EDN from sequences derived from orangutan (Pongo pygmaeus, oEDN) and Old World monkey (Macaca fascicularis, mcEDN) genomic DNAs, and from a second New World monkey sequence (Aotus trivirgatus, omEDN) as well. The catalytic efficiencies [kcat/Km (M-1 s-1)] determined for both oEDN and mcEDN were similar to that determined previously for hEDN, while omEDN displayed ~100-fold less catalytic activity. The relative ribonuclease activities of hEDN/omEDN chimeras pointed to a C-terminal segment as crucial to the enhanced catalytic activity hEDN, and substitution of Arg 132-Ile 133 of hEDN with the Thr-Thr pair at the analogous position in omEDN resulted in an ~10-fold reduction in hEDN's catalytic efficiency. However, the reverse substitution, Arg-Ile for Thr-Thr in omEDN, did not enhance the catalytic efficiency of this relatively inactive protein. These results indicate that the Arg and/or Ile residues adjacent to the C-terminus are necessary (but not sufficient) for enhanced ribonuclease activity among the primate EDNs, and will permit prediction of the relative ribonuclease activities based on differences in primary structure.
The eosinophil-derived neurotoxin (EDN) is a small, glycosylated protein found in the large specific granules of eosinophilic leukocytes. Durack and colleagues (1 ) were first to report the isolation of EDN, and to determine that eosinophil-related neurotoxicity-a syndrome of ataxia and paralysis associated with Purkinje cell degeneration (the Gordon phenomenon, 2 )-wasmediated in part by the activity of this secretory protein (1 ,3 ). Gleich and colleagues (4 ) reported the N-terminal sequence of purified EDN, and noted the similarity between this peptide and the N-terminal sequence of bovine ribonuclease A (RNase A). EDN's membership in the RNase A family of ribonuclease genes was later confirmed by molecular cloning (5 ,6 ). In terms of enzymatic activity, EDN is a catalytically efficient ribonuclease (7 -9 ) and exhibits some degree of preference among experimental substrates (10 ). Both Sorrentino and colleagues (10 ) and Newton and colleagues (11 ) have shown that EDN's neurotoxic effects are directly dependent on ribonuclease activity.
In an earlier study, we traced the evolutionary history of EDN and the closely related ribonuclease/toxin, eosinophil cationic protein (ECP), and found that both genes accumulated non-silent mutations at rates exceeding those of all other functional coding sequences studied in primates while retaining all the structural and catalytic components known to be prerequisites for ribonuclease activity (12 ). With this in mind, we were surprised to find that one of the novel isolates, EDN encoded by the New World monkey Saguinus oedipus, was 100-fold less catalytically active than its human ortholog (13 ). In the work presented here, we have examined the ribonuclease activity of recombinant proteins derived from three additional non-human primate species and compared them to the activities of recombinant human EDN and ECP. Information derived from these experiments has permitted us to create interspecies chimeras, and to identify amino acid sequence elements that support the enhanced level of ribonuclease activity characteristic of human EDN.
The intronless coding sequence of owl monkey EDN (omEDN) was isolated by polymerase chain reaction (PCR) as described (11 ); in this case, a 3' primer encoding a segment of the 3' untranslated region of human EDN was used in order to identify precise sequence at the 3' end of the coding sequence. The source of PCR template was genomic DNA isolated from the owl monkey kidney cell line OMK (637-69) from the American Type Culture Collection (cat. no. CRL-1556). All sequence analysis and comparisons were performed with the assistance of the Wisconsin Genetics Computer Group programs available on-line at the National Institutes of Health.
Chimeras were created by overlapping PCR mutagenesis as described (13 ,14 ), with amplification primers designed to facilitate direct cloning into the pFCTS bacterial expression vector (International Biotechnologies, Inc., New Haven, CT) as described below. All chimeras were confirmed by dideoxy-sequencing.
All EDN and ECP coding sequences were PCR-amplified with primers containing restriction sites facilitating direct cloning into the pFCTS bacterial expression vector (IBI); all constructs were confirmed by dideoxy-sequencing. The pFCTS vector adds the octapeptide DYKDDDK (`FLAG') to the recombinant protein which permits its isolation and detection using the M2 anti-FLAG monoclonal antibody (mAb). We have shown previously that the FLAG octapeptide does not interfere with the folding or the catalytic activity of recombinant ribonucleases (13 ,15 ). Recombinant proteins were isolated from 2-4 l of bacterial cultures after a 1 h induction with isopropyl-1-thio-[beta]-galactoside (IPTG; 1 mM for EDNs, and 1 [mu]M for hECP). After harvest and sucrose lysis (EDNs) or harvest and cell lysis by freeze-thaw and sonication (hECP), recombinant proteins were concentrated and isolated by M2 mAb-agarose affinity chromatography (IBI) as described in detail in reference 13 . The concentration of recombinant protein was determined by comparison to serial dilutions of a known concentration of FLAG-conjugated standard as described (13 ).
Reactions were carried out with varying concentrations of yeast tRNA (Sigma Chemical Co., St Louis, MO, cat. no. R-9001) added in separate reactions to 0.8 ml of 40 mM sodium phosphate, pH 7.0, containing recombinant EDN at concentrations indicated. The assay, solutions, conditions and t = 0 controls were as described in reference 13 . The ribonuclease activity from sham isolates (M2-resin equilibration and glycine elution of periplasmic proteins isolated from equivalent volumes of pFCTS-vector alone bacterial transfectants) was determined; the sham isolates had levels of ribonuclease activity that were insignificant when compared to hEDN, oEDN and mcEDN, and represented no more than 20% of the experimentally determined initial rates for mEDN, omEDN and the lower-activity chimeras. All double- reciprocal plots were constructed from appropriately corrected initial rates. All time points represent the average of triplicate samples. Calculations included the following approximations: the average molecular weight (Mr) of tRNA as Mr = 28 100 (75-90 ribonucleotides/tRNA molecule * Mr = 341/ribonucleotide), with A260 of 1.0 corresponding to 40 [mu]g of RNA (16 ). Bestfits and correlation coefficients (r2) were determined with the assistance of Cricket Graph software on-line at the National Institutes of Health.
The estimated evolutionary distances between the human and non-human primate species discussed in this work are shown inFigure 1 A (13 ,17 ). The double reciprocal plots shown inFigure 1 B demonstrate the relationship between substrate concentration and ribonuclease activity for recombinant hECP and EDNs prepared from several non-human primate sequences: orangutan (Pongo pygmaeus, oEDN), the Old World monkey, macaque (Macaca fascicularis, mcEDN), and the New World monkey, owl monkey (Aotus trivirgatus, omEDN). The values for Km ([mu]M), as determined from the x-intercepts, and kcat (s-1), as determined from the y-intercepts of these plots, are tabulated inFigure 1 C and compared to those obtained previously for both human (Homo sapiens, hEDN)- and New World monkey (S.oedipus, mEDN)- derived recombinant proteins (13 ). Although the amino acid sequences of both oEDN and mcEDN differ significantly from that of hEDN (6.8 and 16.5% divergence, respectively), the catalytic constants and the calculated catalytic efficiency (kcat/Km) remain unchanged; the catalytic efficiences of recombinant hEDN, oEDN and mcEDN are all on the order of 106 M-1 s-1. In contrast, the catalytic constants determined for the two EDNs isolated from New World monkeys (mEDN and omEDN) show significant reductions in kcat, and thus reductions in overall catalytic efficiencies (~100-fold), with kcat/Km calculated for each at 0.96 * 104 M-1 s-1 and 1.2 * 104 M-1 s-1 , respectively. The catalytic efficiency of recombinant hECP was even lower, with kcat/Km determined at 0.59 * 103 M-1 s-1 , ~2000-fold less activity than that observed for recombinant hEDN and ~20-fold less than for omEDN.
An alignment of the predicted amino acid sequences of hEDN and omEDN is shown inFigure 2 A. The omEDN sequence retains the eight cysteines as well as the catalytic histidines and lysine that are characteristic of the RNase A gene family (18 ,19 ). Similarly, the omEDN sequence contains the `CKXXNTF' motif (amino acids 37-44) also found to be invariant among these proteins (18 -20 ). The calculated amino acid sequence divergence between hEDN and omEDN is 29.2% (Fig. 1 C). The two sequences were found to be identical within the two bracketed areas shown, permitting construction of chimeras A and B (Fig. 2 B). The ribonucleolytic activities of both chimera A (first part hEDN followed by second and third parts omEDN) and chimera B (first and second parts hEDN followed by third part omEDN) were determined as `+', reflecting their similarity to the lower level of activity observed for omEDN (and mEDN). These results suggested that a sequence element (or elements) present in the C-terminus of hEDN was necessary for full catalytic activity. There are only two regions of significant divergence within this final segment of EDN sequence: the gap in omEDN in place of Arg 117 of hEDN, and the penultimate Thr-Thr in omEDN in place of Arg 132-Ile 133 of hEDN (Fig. 1 A). Interestingly, both oEDN and mcEDN are more closely related to hEDN at these sites; oEDN is identical to hEDN, and in mcEDN, a Val replaces Arg 117.
As shown in Figure 3 A, chimera C was created by exchanging the C-terminal Arg-Ile-Ile (132-134) of hEDN with Thr-Thr-Ile from omEDN, and chimera D, by exchanging the Thr-Thr-Ile of omEDN for Arg-Ile-Ile. Double reciprocal plots of substrate concentration versus initial rates yielded the catalytic constants listed in Figure 3 B. Comparison of the values calculated for kcat/Km for hEDN (1.3 * 106 M-1 s-1) and chimera C (1.6 * 105 M-1 s-1) demonstrates that the C-terminal sequence exchange resulted in an ~8-fold reduction in catalytic activity. In contrast, kcat/Km determined for chimera D (1.3 * 10 4 M-1 s-1) does not differ significantly from that determined for wild type omEDN. Taken together, these results indicate that the C-terminal sequence Arg 132-Ile 133 is necessary to sustain the full catalytic activity of hEDN, but at the same time it has no effect on the relatively inactive omEDN.
We would like to thank Dr Jaap Beintema for his helpful comments, and Dr John I. Gallin for his continuing suppport of our work.
*To whom correpondence should be addressed at: LHD/NIAID/NIH, Building 10, room 11N104, 9000 Rockville Pike, Bethesda, MD 20892, USA. Tel: +1 301 402 9131; Fax: +1 301 402 4369; Email: hr2k@nih.gov
REFERENCES