The wide binding properties of a wheat nonspecific lipid transfer protein. Solution structure of a complex with prostaglandin B2.

The 3D solution structure of wheat nonspecific lipid transfer protein (ns-LTP) complexed with prostaglandin B2, a lipid with both vinyl and hydroxylated groups, has been determined by 1H 2D NMR. The global fold of the protein is close to the previously published structures of wheat, maize, barley and rice ns-LTPs. The ligand is almost completely embedded in the hydrophobic core of the protein. Structure comparisons of free and bound wheat ns-LTP reveal that the binding of prostaglandin B2 hardly affects the global fold of the protein. The structural data on this unusual complex are discussed and compared with other known ns-LTP lipid-complexes.     

Solution structure of the functional domain of Paracoccus denitrificans cytochrome c552 in the reduced state.

In order to determine the solution structure of Paracoccus denitrificans cytochrome c552 by NMR, we cloned and isotopically labeled a 10.5-kDa soluble fragment (100 residues) containing the functional domain of the 18.2-kDa membrane-bound protein. Using uniformly 15N-enriched samples of cytochrome c552 in the reduced state, a variety of two-dimensional and three-dimensional heteronuclear double-resonance NMR experiments was employed to achieve complete 1H and 15N assignments. A total of 1893 distance restraints was derived from homonuclear 2D-NOESY and heteronuclear 3D-NOESY spectra; 1486 meaningful restraints were used in the structure calculations. After restrained energy minimization a family of 20 structures was obtained with rmsd values of 0.56 +/- 0. 10 A and 1.09 +/- 0.09 A for the backbone and heavy atoms, respectively. The overall topology is similar to that seen in previously reported models of this class of proteins. The global fold consists of two long helices at the N-terminus and C-terminus and three shorter helices surrounding the heme moiety; the helices are connected by well-defined loops. Comparison with the X-ray structure shows some minor differences in the positions of the Trp57 and Phe65 side-chain rings as well as the heme propionate groups.     

NMR structure of cysteinyl-phosphorylated enzyme IIB of the N,N'-diacetylchitobiose-specific phosphoenolpyruvate-dependent phosphotransferase system of Escherichia coli

The determination by NMR of the solution structure of the phosphorylated enzyme IIB (P-IIB(Chb)) of the N,N'-diacetylchitobiose-specific phosphoenolpyruvate-dependent phosphotransferase system of Escherichia coli is presented. Most of the backbone and side-chain resonances were assigned using a variety of mostly heteronuclear NMR experiments. The remaining resonances were assigned with the help of the structure calculations.NOE-derived distance restraints were used in distance geometry calculations followed by molecular dynamics and simulated annealing protocols. In addition, combinations of ambiguous restraints were used to resolve ambiguities in the NOE assignments. By combining sets of ambiguous and unambiguous restraints into new ambiguous restraints, an error function was constructed that was less sensitive to information loss caused by assignment uncertainties. The final set of structures had a pairwise rmsd of 0.59 A and 1.16 A for the heavy atoms of the backbone and side-chains, respectively.Comparing the P-IIB(Chb) solution structure with the previously determined NMR and X-ray structures of the wild-type and the Cys10Ser mutant shows that significant differences between the structures are limited to the active-site region. The phosphoryl group at the active-site cysteine residue is surrounded by a loop formed by residues 10 through 16. NOE and chemical shift data suggest that the phosphoryl group makes hydrogen bonds with the backbone amide protons of residues 12 and 15. The binding mode of the phosphoryl group is very similar to that of the protein tyrosine phosphatases. The differences observed are in accordance with the presumption that IIB(Chb) has to be more resistant to hydrolysis than the protein tyrosine phosphatases. We propose a proton relay network by which a transfer occurs between the cysteine SH proton and the solvent via the hydroxyl group of Thr16.     

The Three-Dimensional NMR Solution Structure of α-Cobratoxin at pH 7.5 and Conformational Differences With the NMR Solution Structure at pH 3.2

Abstract The 3D solution structure of α-cobratoxin, a neurotoxin purified from the Naja naja siamensis snake venom, has been determined by Nuclear Magnetic Resonance spectroscopy, in conjunction with distance geometry and restrained molecular dynamics, at pH 7.5. A total of 490 distance restraints were obtained from NOE intensities and 25 φ dihedral angle restraints deduced from J- coupling data. The generated structures are well defined with root mean square deviations from a geometrical mean structure of 0.107 ± 0.036 nm for the backbone atoms and 0.128 ±0.073 nm for the side-chain atoms (considering residues 1 to 66 minus 26 to 35). A comparison between the generated structures at pH 7.5 and the mean NMR solution structure at pH 3.2 revealed that the 3D structure of α-cobratoxin is more compact at neutral pH. This major difference is mainly due to the pH-dependant conformational variations of three residues His(18), Thr(44) and Thr(59).     

The three-dimensional structure of guanine-specific ribonuclease F1 in solution determined by NMR spectroscopy and distance geometry.

Two-dimensional 1H-NMR studies have been performed on ribonuclease F1 (RNase F1), which contains 106 amino acid residues. Sequence-specific resonance assignments were accomplished for the backbone protons of 99 amino acid residues and for most of their side-chain protons. The three-dimensional structures were constructed on the basis of 820 interproton-distance restraints derived from NOE, 64 distance restraints for 32 hydrogen bonds and 33 phi torsion-angle restraints. A total of 40 structures were obtained by distance geometry and simulated-annealing calculations. The average root-mean-square deviation (residues 1-106) between the 40 converged structures and the mean structure obtained by averaging their coordinates was 0.116 +/- 0.018 nm for the backbone atoms and 0.182 +/- 0.015 nm for all atoms including the hydrogen atoms. RNase F1 was determined to be an alpha/beta-type protein. A well-defined structure constitutes the core region, which consists of a small N-terminal beta-sheet (beta 1, beta 2) and a central five-stranded beta-sheet (beta 3-beta 7) packed on a long helix. The structure of RNase F1 has been compared with that of RNase T1, which was determined by X-ray crystallography. Both belong to the same family of microbial ribonucleases. The polypeptide backbone fold of RNase F1 is basically identical to that of RNase T1. The conformation-dependent chemical shifts of the C alpha protons are well conserved between RNase F1 and RNase T1. The residues implicated in catalysis are all located on the central beta-sheet in a geometry similar to that of RNase T1.     

Determination of the three-dimensional solution structure of the histidine-containing phosphocarrier protein HPr from Escherichia coli using multidimensional NMR spectroscopy.

We recorded several types of heteronuclear three-dimensional (3D) NMR spectra on 15N-enriched and 13C/15N-enriched histidine-containing phosphocarrier protein, HPr, to extend the backbone assignments [van Nuland, N. A. J., van Dijk, A. A., Dijkstra, K., van Hoesel, F. H. J., Scheek, R. M. & Robillard, G. T. (1992) Eur. J. Biochem, 203, 483-491] to the side-chain 1H,15N and 13C resonances. From both 3D heteronuclear 1H-NOE 1H-13C and 1H-NOE 1H-15N multiple-quantum coherence (3D-NOESY-HMQC) and two-dimensional (2D) homonuclear NOE spectra, more than 1200 NOE were identified and used in a step-wise structure refinement process using distance geometry and restrained molecular dynamics involving a number of new features. A cluster of nine structures, each satisfying the set of NOE restraints, resulted from this procedure. The average root-mean-square positional difference for the C alpha atoms is less than 0.12 nm. The secondary structure topology of the molecule is that of an open-face beta sandwich formed by four antiparallel beta strands packed against three alpha helices, resembling the recently published structure of Bacillus subtilis HPr, determined by X-ray crystallography [Herzberg, O., Reddy, P., Sutrina, S., Saier, M. H., Reizer, J. & Kapafia, G. (1992) Proc. Natl, Acad. Sci. USA 89, 2499-2503).     

A new strategy for structure determination of large proteins in solution without deuteration

So far high-resolution structure determination by nuclear magnetic resonance (NMR) spectroscopy has been limited to proteins <30 kDa, although global fold determination is possible for substantially larger proteins. Here we present a strategy for assigning backbone and side-chain resonances of large proteins without deuteration, with which one can obtain high-resolution structures from (1)H-(1)H distance restraints. The strategy uses information from through-bond correlation experiments to filter intraresidue and sequential correlations from through-space correlation experiments, and then matches the filtered correlations to obtain sequential assignment. We demonstrate this strategy on three proteins ranging from 24 to 65 kDa for resonance assignment and on maltose binding protein (42 kDa) and hemoglobin (65 kDa) for high-resolution structure determination. The strategy extends the size limit for structure determination by NMR spectroscopy to 42 kDa for monomeric proteins and to 65 kDa for differentially labeled multimeric proteins without the need for deuteration or selective labeling.     

Conotoxin TVIIA, a novel peptide from the venom of Conus tulipa 2. Three-dimensional solution structure.

The three-dimensional solution structure of conotoxin TVIIA, a 30-residue polypeptide from the venom of the piscivorous cone snail Conus tulipa, has been determined using 2D 1H NMR spectroscopy. TVIIA contains six cysteine residues which form a 'four-loop' structural framework common to many peptides from Conus venoms including the omega-, delta-, kappa-, and muO-conotoxins. However, TVIIA does not belong to these well-characterized pharmacological classes of conotoxins, but displays high sequence identity with conotoxin GS, a muscle sodium channel blocker from Conus geographus. Structure calculations were based on 562 interproton distance restraints inferred from NOE data, together with 18 backbone and nine side-chain torsion angle restraints derived from spin-spin coupling constants. The final family of 20 structures had mean pairwise rms differences over residues 2-27 of 0.18+/-0.05 A for the backbone atoms and 1.39+/-0.33 A for all heavy atoms. The structure consists of a triple-stranded, antiparallel beta sheet with +2x, -1 topology (residues 7-9, 16-20 and 23-27) and several beta turns. The core of the molecule is formed by three disulfide bonds which form a cystine knot motif common to many toxic and inhibitory polypeptides. The global fold, molecular shape and distribution of amino-acid sidechains in TVIIA is similar to that previously reported for conotoxin GS, and comparison with other four-loop conotoxin structures provides further indication that TVIIA and GS represent a new and distinct subgroup of this structural family. The structure of TVIIA determined in this study provides the basis for determining a structure-activity relationship for these molecules and their interaction with target receptors.     

Determination of the three-dimensional solution structure of the antihypertensive and antiviral protein BDS-I from the sea anemone Anemonia sulcata: a study using nuclear magnetic resonance and hybrid distance geometry-dynamical simulated annealing

The three-dimensional solution structure of the antihypertensive and antiviral protein BDS-I from the sea anemone Anemonia sulcata has been determined on the basis of 489 interproton and 24 hydrogen-bonding distance restraints supplemented by 23 phi backbone and 21 chi 1 side-chain torsion angle restraints derived from nuclear magnetic resonance (NMR) measurements. A total of 42 structures is calculated by a hybrid metric matrix distance geometry-dynamical simulated annealing approach. Both the backbone and side-chain atom positions are well defined. The average atomic rms difference between the 42 individual SA structures and the mean structure obtained by averaging their coordinates is 0.67 +/- 0.12 A for the backbone atoms and 0.90 +/- 0.17 A for all atoms. The core of the protein is formed by a triple-stranded antiparallel beta-sheet composed of residues 14-16 (strand 1), 30-34 (strand 2), and 37-41 (strand 3) with an additional mini-antiparallel beta-sheet at the N-terminus (residues 6-9). The first and second strands of the triple-stranded antiparallel beta-sheet are connected by a long exposed loop (residues 17-30). A number of side-chain interactions are discussed in light of the structure.     

Protein stability and plasticity of the hydrophobic cavity in wheat ns-LTP

Plant ns-LTPs display an original structure with four helices and a flexible C-terminus, maintained together by four disulphide bridges and delineating an elongated central hydrophobic cavity. In order to relate these structural features to the protein stability and plasticity, combined molecular mechanics and simulated annealing calculations were undertaken on a wheat ns-LTP "mutant" with Cys-Ala replacement and with the application of core inter-residue restraints up to 2 A, reducing the cross-section size of the hydrophobic cavity. Analysis of the energy-minimized structures shows that removal of the disulphide bridges results in structures with a lower total energy and a smaller cavity volume. A 1-ns MD simulation at 300K in water, underlines that, despite the absence of a well-packed hydrophobic core, the native structure is extremely stable at room temperature and the cavity is not hydrated. This confirms that the disulphide bridges are essential for the existence of the cavity, whereas its plasticity depends both on the hydrophobic chain lining the cavity and on the C-terminal flexibility. A high temperature (500K) MD simulation confirms the stability of the secondary structure elements and the flexibility of the loops and of the C-terminal segment. Two important structural transitions during this simulation are discussed and possible routes for the insertion and release of hydrophobic ligands are suggested.     

Three-dimensional solution structure of the B domain of staphylococcal protein A: comparisons of the solution and crystal structures.

The three-dimensional solution structure of the recombinant B domain (FB) of staphylococcal protein A, which specifically binds to the Fc portion of immunoglobulin G, was determined by NMR spectroscopy and hybrid distance geometry-dynamical simulated annealing calculations. On the basis of 692 experimental constraints including 587 distance constraints obtained from the nuclear Overhauser effect (NOE), 57 torsion angle (phi, chi 1) constraints, and 48 constraints associated with 24 hydrogen bonds, a total of 10 converged structures of FB were obtained. The atomic root mean square difference among the 10 converged structures is 0.52 +/- 0.10 A for the backbone atoms and 0.98 +/- 0.08 A for all heavy atoms (excluding the N-terminal segment from Thr1 to Glu9 and the C-terminal segment from Gln56 to Ala60, which are partially disordered). FB is composed of a bundle of three alpha-helices, i.e., helix I (Gln10-His19), helix II (Glu25-Asp37), and helix III (Ser42-Ala55). Helix II and helix III are antiparallel to each other, whereas the long axis of helix I is tilted at an angle of about 30 degrees with respect to those of helix II and helix III. Most of the hydrophobic residues of FB are buried in the interior of the bundle of the three helices. It is suggested that the buried hydrophobic residues form a hydrophobic core, contributing to the stability of FB.(ABSTRACT TRUNCATED AT 250 WORDS)     

Automated protein fold determination using a minimal NMR constraint strategy

Determination of precise and accurate protein structures by NMR generally requires weeks or even months to acquire and interpret all the necessary NMR data. However, even medium-accuracy fold information can often provide key clues about protein evolution and biochemical function(s). In this article we describe a largely automatic strategy for rapid determination of medium-accuracy protein backbone structures. Our strategy derives from ideas originally introduced by other groups for determining medium-accuracy NMR structures of large proteins using deuterated, (13)C-, (15)N-enriched protein samples with selective protonation of side-chain methyl groups ((13)CH(3)). Data collection includes acquiring NMR spectra for automatically determining assignments of backbone and side-chain (15)N, H(N) resonances, and side-chain (13)CH(3) methyl resonances. These assignments are determined automatically by the program AutoAssign using backbone triple resonance NMR data, together with Spin System Type Assignment Constraints (STACs) derived from side-chain triple-resonance experiments. The program AutoStructure then derives conformational constraints using these chemical shifts, amide (1)H/(2)H exchange, nuclear Overhauser effect spectroscopy (NOESY), and residual dipolar coupling data. The total time required for collecting such NMR data can potentially be as short as a few days. Here we demonstrate an integrated set of NMR software which can process these NMR spectra, carry out resonance assignments, interpret NOESY data, and generate medium-accuracy structures within a few days. The feasibility of this combined data collection and analysis strategy starting from raw NMR time domain data was illustrated by automatic analysis of a medium accuracy structure of the Z domain of Staphylococcal protein A.     

Solution structure and dynamics of a designed monomeric variant of the lambda Cro repressor.

The solution structure of a monomeric variant of the lambda Cro repressor has been determined by multidimensional NMR. Cro K56[DGEVK] differs from wild-type Cro by the insertion of five amino acids at the center of the dimer interface. 1H and 15N resonances for 70 of the 71 residues have been assigned. Thirty-two structures were calculated by hybrid distance geometry/simulated annealing methods using 463 NOE-distance restraints, 26 hydrogen-bond, and 39 dihedral-angle restraints. The root-mean-square deviation (RMSD) from the average structure for atoms in residues 3-60 is 1.03 +/- 0.44 A for the peptide backbone and 1.6 +/- 0.73 A for all nonhydrogen atoms. The overall structure conforms very well to the original design. Although the five inserted residues form a beta hairpin as expected, this engineered turn as well as other turns in the structure are not well defined by the NMR data. Dynamics studies of backbone amides reveal T1/T2 ratios of residues in the alpha2-alpha3, beta2-beta3, and engineered turn that are reflective of chemical exchange or internal motion. The solution structure and dynamics are discussed in light of the conformational variation that has been observed in other Cro structures, and the importance of flexibility in DNA recognition.     

The influence of stereospecific assignments on the determination of three-dimensional structures of proteins by nuclear magnetic resonance spectroscopy. Application to the sea anemone protein BDS-I

The influence of the stereospecific assignments of beta-methylene protons and the classification of chi 1 torsion angles on the definition of the three-dimensional structures of proteins determined from NMR data is investigated using the sea anemone protein BDS-I (43 residues) as a model system. Two sets of structures are computed. The first set comprises 42 converged structures (denoted STEREO structures) calculated on the basis of the complete list of restraints derived from the NMR data, consisting of 489 interproton and 24 hydrogen bonding distance restraints, supplemented by 23 phi backbone and 21 chi 1 side chain torsion angle restraints. The second set comprises 31 converged structures (denoted NOSTEREO structures) calculated from a reduced data set in which those restraints arising from stereospecific assignments, and the corresponding chi 1 torsion angle restraints, are explicitly omitted. The results show that the inclusion of the stereospecific restraints leads to a significant improvement in the definition of the structure of BDS-I, both with respect to the backbone and the detailed arrangement of the side chains. Average atomic rms differences between the individual structures and the mean structures for the backbone atoms are 0.67 +/- 0.12 A and 0.93 +/- 0.16 A for the STEREO and NOSTEREO structures, respectively; the corresponding values for all atoms are 0.90 +/- 0.17 A and 1.17 +/- 0.17 A, respectively. In addition, while the overall fold remains unchanged, there is a small but significant atomic displacement between the two sets of structures.     

Solution structure of the PDZ2 domain from human phosphatase hPTP1E and its interactions with C-terminal peptides from the Fas receptor

The solution structure of the second PDZ domain (PDZ2) from human phosphatase hPTP1E has been determined using 2D and 3D heteronuclear NMR experiments. The binding of peptides derived from the C-terminus of the Fas receptor to PDZ2 was studied via changes in backbone peptide and protein resonances. The structure is based on a total of 1387 nonredundant experimental NMR restraints including 1261 interproton distance restraints, 45 backbone hydrogen bonds, and 81 torsion angle restraints. Analysis of 30 lowest-energy structures resulted in rmsd values of 0.41 +/- 0.09 A for backbone atoms (N, Calpha, C') and 1.08 +/- 0.10 A for all heavy atoms, excluding the disordered N- and C-termini. The hPTP1E PDZ2 structure is similar to known PDZ domain structures but contains two unique structural features. In the peptide binding domain, the first glycine of the GLGF motif is replaced by a serine. This serine appears to replace a bound water observed in PDZ crystal structures that hydrogen bonds to the bound peptide's C-terminus. The hPTP1E PDZ2 structure also contains an unusually large loop following strand beta2 and proximal to the peptide binding site. This well-ordered loop folds back against the PDZ domain and contains several residues that undergo large amide chemical shift changes upon peptide binding. Direct observation of peptide resonances demonstrates that as many as six Fas peptide residues interact with the PDZ2 domain.     

Determination of the structure of the DNA binding domain of gamma delta resolvase in solution.

The DNA binding domain (DBD) of gamma delta resolvase (residues 141-183) is responsible for the interaction of this site-specific DNA recombinase with consensus site DNA within the gamma delta transposable element in Escherichia coli. Based on chemical-shift comparisons, the proteolytically isolated DBD displays side-chain interactions within a hydrophobic core that are highly similar to those of this domain when part of the intact enzyme (Liu T, Liu DJ, DeRose EF, Mullen GP, 1993, J Biol Chem 268:16309-16315). The structure of the DBD in solution has been determined using restraints obtained from 2-dimensional proton NMR data and is represented by 17 conformers. Experimental restraints included 458 distances based on analysis of nuclear Overhauser effect connectivities, 17 phi and chi 1 torsion angles based on analysis of couplings, and 17 backbone hydrogen bonds determined from NH exchange data. With respect to the computed average structure, these conformers display an RMS deviation of 0.67 A for the heavy backbone atoms and 1.49 A for all heavy atoms within residues 149-180. The DBD consists of 3 alpha-helices comprising residues D149-Q157, S162-T167, and R172-N183. Helix-2 and helix-3 form a backbone fold, which is similar to the canonical helix-turn-helix motif. The conformation of the NH2-terminal residues, G141-R148, appears flexible in solution. A hydrophobic core is formed by side chains donated by essentially all hydrophobic residues within the helices and turns. Helix-1 and helix-3 cross with a right-handed folding topology. The structure is consistent with a mechanism of DNA binding in which contacts are made by the hydrophilic face of helix-3 in the major groove and the amino-terminal arm in the minor groove. This structure represents an important step toward analysis of the mechanism of DNA interaction by gamma delta resolvase and provides initial structure-function comparisons among the divergent DBDs of related resolvases and invertases.     

Improvement of hydrogen bond geometry in protein NMR structures by residual dipolar couplings – an assessment of the interrelation of NMR restraints

We have examined how the hydrogen bond geometry in three different proteins is affected when structural restraints based on measurements of residual dipolar couplings are included in the structure calculations. The study shows, that including restraints based solely on (1)H(N)-(15)N residual dipolar couplings has pronounced impact on the backbone rmsd and Ramachandran plot but does not improve the hydrogen bond geometry. In the case of chymotrypsin inhibitor 2 the addition of (13)CO-(13)C(alpha) and (15)N-(13)CO one bond dipolar couplings as restraints in the structure calculations improved the hydrogen bond geometry to a quality comparable to that obtained in the 1.8 A resolution X-ray structure of this protein. A systematic restraint study was performed, in which four types of restraints, residual dipolar couplings, hydrogen bonds, TALOS angles and NOEs, were allowed in two states. This study revealed the importance of using several types of residual dipolar couplings to get good hydrogen bond geometry. The study also showed that using a small set of NOEs derived only from the amide protons, together with a full set of residual dipolar couplings resulted in structures of very high quality. When reducing the NOE set, it is mainly the side-chain to side-chain NOEs that are removed. Despite of this the effect on the side-chain packing is very small when a reduced NOE set is used, which implies that the over all fold of a protein structure is mainly determined by correct folding of the backbone.     

An approach to global fold determination using limited NMR data from larger proteins selectively protonated at specific residue types

Summary A combination of calculation and experiment is used to demonstrate that the global fold of larger proteins can be rapidly determined using limited NMR data. The approach involves a combination of heteronuclear triple resonance NMR experiments with protonation of selected residue types in an otherwise completely deuterated protein. This method of labelling produces proteins with -specific deuteration in the protonated residues, and the results suggest that this will improve the sensitivity of experiments involving correlation of side-chain (¹H and ¹³C) and backbone (¹H and ¹⁵N) amide resonances. It will allow the rapid assignment of backbone resonances with high sensitivity and the determination of a reasonable structural model of a protein based on limited NOE restraints, an application that is of increasing importance as data from the large number of genome sequencing projects accumulates. The method that we propose should also be of utility in extending the use of NMR spectroscopy to determine the structures of larger proteins.The first two authors contributed equally to this work.     

Crystal structure of peach Pru p 3, the prototypic member of the family of plant non-specific lipid transfer protein pan-allergens

This study describes the three-dimensional crystal structure of a non-specific lipid transport protein (ns-LTP) from Rosaceae. Whilst ns-LTPs from species other than Rosaceae, such as nuts, cereals, grape, oranges and vegetables are also responsible for plant food allergies, this is less frequent compared with ns-LTPs from Rosaceae in the Mediterranean area. In this heterologously expressed peach Pru p3, a ligand is present inside the central cavity of the protein, presumably a fatty acid that was present or produced in the culture medium of the expression organism Escherichia coli. Moreover, the two molecules of ns-LTP present in the asymmetric unit bind this ligand in a different way, suggesting a significant degree of plasticity for the peach ns-LTP binding cavity, despite the presence of four disulphide bridges. Two molecules are present in the asymmetric unit: molecule A is a fully liganded protein, while molecule B apparently represents a partially liganded state. Also, molecular dynamics simulation, along with other evidence, suggests that these two molecular conformations represent different states in solution. Comparison of the 3D models of different ns-LTPs justifies the evidence of a high degree of conservation of the putative IgE binding epitopes among proteins of the Rosaceae family and the presence of significant amino acid replacements in correspondence of the same regions in ns-LTPs of botanical species unrelated to Rosaceae.