Mode of Action of cGMP-dependent Protein Kinase-specific Inhibitors Probed by Photoaffinity Cross-linking Mass Spectrometry

The inhibitor peptide DT-2 (YGRKKRRQRRRPPLRKKKKKH) is the most potent and selective inhibitor of the cGMP-dependent protein kinase (PKG) known today. DT-2 is a construct of a PKG tight binding sequence (W45, LRKKKKKH, K_I = 0.8 μm) and a membrane translocating sequence (DT-6, YGRKKRRQRRRPP, K_I = 1.1 μm), that combined strongly inhibits PKG catalyzed phosphorylation (K_I = 12.5 nm) with ∼1000-fold selectivity toward PKG over protein kinase A, the closest relative of PKG. However, the molecular mechanism behind this inhibition is not entirely understood. Using a combination of photoaffinity labeling, stable isotope labeling, and mass spectrometry, we have located the binding sites of PKG-specific substrate and inhibitor peptides. Covalent linkage of a PKG-specific substrate analogue was localized in the catalytic core on residues 356–372, also known as the glycine-rich loop, essential for ATP binding. By analogy, the individual inhibitor peptides W45 and DT-6 were also found to cross-link near the glycine-rich loop, suggesting these are both substrate competitive inhibitors. A bifunctional photoreactive analogue of DT-2 was found to generate dimers of PKG. This cross-linking induced covalent PKG dimerization was not observed for an N-terminal deletion mutant of PKG, which lacks the dimerization domain. In addition, non-covalent mass spectrometry was used to determine binding stoichiometry and binding order of the inhibitor peptides. Dimeric PKG binds two W45 and DT-6 peptides, whereas only one DT-2 molecule was observed to bind to the dimeric PKG. Taken together, these findings imply that (i) the two individual components making up DT-2 are both targeted against the substrate-binding site and (ii) binding of a single DT-2 molecule inactivates both PKG monomers simultaneously, which is an indication that (iii) in cGMP-activated PKG the catalytic centers of both subunits may be in each other''s proximity.Among the superfamily of protein kinases the two cyclic nucleotide-regulated protein kinases, cAMP-dependent protein kinase and cGMP-dependent protein kinase, form a closely related subfamily of serine/threonine protein kinases (1–4). Both proteins share several structural elements, such as the N-terminal dimerization domain, an autoinhibition site, two in-tandem cyclic nucleotide-binding sites, and a highly conserved catalytic core (Fig. 1, A and B). Despite these similarities, these two enzymes display differences, which account for their unique properties. Whereas PKA² is nearly ubiquitous, PKG is primarily found in the lung, cerebellum, and smooth muscles (5, 6). From a structural point of view these cyclic nucleotide-dependent protein kinases differ as well. The holoenzyme of PKA is a tetramer composed of two regulatory and two catalytic subunits. The catalytic subunits are non-covalently attached to the regulatory subunit dimer. Upon interaction with cAMP, the catalytic subunits dissociate from the holoenzyme and are free to catalyze heterophosphorylation (Fig. 1C). The mammalian type I PKGs are homodimeric cytosolic proteins containing two identical polypeptides of ∼76 kDa. Alternative mRNA splicing produces type Iα and type Iβ PKG, which are identical proteins apart from their first ∼100 N-terminal residues (7). Each PKG subunit is composed of a regulatory and a catalytic domain on a single polypeptide chain. Consequently, when cGMP activates PKG, the catalytic and regulatory components remain physically attached (Fig. 1D). Within the catalytic domain PKA and PKG share a strong primary sequence homology (8). Not surprisingly, these enzymes also exhibit overlapping substrate specificities, a feature that often interferes with efforts to elucidate their distinct biological pathways. Peptide substrates with a primary amino acid sequence motif RRX(S/T)X are in general recognized by both PKA and PKG (9). Besides this strong overlapping substrate specificity, several studies report on subtle differences in determinants that discriminate for PKA and PKG substrate specificity (10–16). To specifically discriminate between PKG and PKA activity in biological assays a highly specific PKG peptide inhibitor was developed (17). This peptide, YGRKKRRQRRRPPLRKKKKKH (DT-2), is the most potent and selective PKG inhibitor known today. Recently, the validity of DT-2 as a superior inhibitor of PKG in terms of potency, selectivity, and membrane permeability has been demonstrated (18–24). The inhibitor is a construct of a substrate competitive sequence, LRKKKKKH (W45), derived from a library screen that selected for tight PKG binding sequences, with a significant specificity toward PKG over PKA, and a membrane translocating signal peptide, YGRKKRRQRRRPP (DT-6). DT-2 strongly inhibits PKG-catalyzed phosphorylation (K_i = 12.5 nm), however, the molecular nature of DT-2 inhibition is not entirely understood (25). Because high resolution structural data are not available for PKG, one of our goals is to elucidate binding sites for PKG-specific substrates and inhibitors in more detail using a combination of mass spectrometric techniques and photoaffinity labeling. To further delineate the nature of inhibition we have developed photoaffinity analogues of DT-2 and related inhibitory peptides, as well as a high affinity peptide substrate. The method of photoaffinity labeling enables the direct probing of target proteins through a covalent bond, which is photochemically introduced between a ligand and its specific receptor (26). In combination with modern mass spectrometric techniques this is a powerful approach for the characterization of peptide-protein interactions (27). Substrate and inhibitor peptides containing photoactivatable analogues of phenylalanine, 4-benzoyl-l-phenylalanine (Phe(Bz)) or 4′-(3-(trifluoromethyl)-3H-diazirin-3-yl)-l-phenylalanine (Phe(Tmd)) were synthesized and used to locate their substrate/inhibitor-binding sites on PKG. These measurements indicate that the substrate peptide resides near the glycine-rich loop within the catalytic domain and that the inhibitor peptides are directed similarly toward this substrate-binding site, thereby acting as competitive inhibitors. In addition, nanoflow electrospray ionization time of flight mass spectrometry (ESI-TOF-MS) was performed to study the interaction between DT-2 and PKG in more detail. ESI-MS has proven to be a useful tool to analyze the non-covalent interaction of proteins with ligands, oligonucleotides, peptides, or other proteins (28–31). Using this technique, important information on conformational changes (32–35), measurement of relative dissociation constants (36, 37), and sequential binding order and cooperativity (38, 39) can be obtained. ESI-MS confirms that PKG is primarily a homodimer and is able to bind four cGMP molecules. Binding of DT-2 was strongly enhanced in the presence of cGMP. Surprising is the observation that only one DT-2 molecule binds to dimeric PKG. The information derived from these measurements allows for molecular modeling and structural refinements of the next generation of PKG-selective inhibitors.Open in a separate windowFIGURE 1.Linear arrangement of the functional domains of the regulatory and catalytic subunit of PKA (A) and PKG (B) type I and schematic representation of the current working models of the activation process of PKA (C) and PKG (D) type 1. Binding of cAMP to the PKA induces a conformational change that results in the dissociation of the catalytic subunits. Binding of cGMP to PKG also induces a conformational change, which exposes the catalytic domains, but both catalytic domains remain near each other via the N-terminal dimerization domain. (Images adapted from Scholten et al. (4).)

TABLE 1

Inhibition contants (K_I) of PKA- or PKG-specific peptide inhibitors and the PKA/PKG specificity index

Critical Factors Determining Dimerization of Human Antizyme Inhibitor

Ornithine decarboxylase (ODC) is the first enzyme involved in polyamine biosynthesis, and it catalyzes the decarboxylation of ornithine to putrescine. ODC is a dimeric enzyme, whereas antizyme inhibitor (AZI), a positive regulator of ODC that is homologous to ODC, exists predominantly as a monomer and lacks decarboxylase activity. The goal of this paper was to identify the essential amino acid residues that determine the dimerization of AZI. The nonconserved amino acid residues in the putative dimer interface of AZI (Ser-277, Ser-331, Glu-332, and Asp-389) were substituted with the corresponding residues in the putative dimer interface of ODC (Arg-277, Tyr-331, Asp-332, and Tyr-389, respectively). Analytical ultracentrifugation analysis was used to determine the size distribution of these AZI mutants. The size-distribution analysis data suggest that residue 331 may play a major role in the dimerization of AZI. Mutating Ser-331 to Tyr in AZI (AZI-S331Y) caused a shift from a monomer configuration to a dimer. Furthermore, in comparison with the single mutant AZI-S331Y, the AZI-S331Y/D389Y double mutant displayed a further reduction in the monomer-dimer K_d, suggesting that residue 389 is also crucial for AZI dimerization. Analysis of the triple mutant AZI-S331Y/D389Y/S277R showed that it formed a stable dimer (K_d value = 1.3 μm). Finally, a quadruple mutant, S331Y/D389Y/S277R/E332D, behaved as a dimer with a K_d value of ∼0.1 μm, which is very close to that of the human ODC enzyme. The quadruple mutant, although forming a dimer, could still be disrupted by antizyme (AZ), further forming a heterodimer, and it could rescue the AZ-inhibited ODC activity, suggesting that the AZ-binding ability of the AZI dimer was retained.Polyamines (putrescine, spermidine, and spermine) have been shown to have both structural and regulatory roles in protein and nucleic acid biosynthesis and function (1–3). Ornithine decarboxylase (ODC,³ EC 4.1.1.17) is a central regulator of cellular polyamine synthesis (reviewed in Refs. 1, 4, 5). This enzyme catalyzes the pyridoxal 5-phosphate (PLP)-dependent decarboxylation of ornithine to putrescine, and it is the first and rate-limiting enzyme in polyamine biosynthesis (2, 3, 6, 7). ODC and polyamines play important roles in a number of biological functions, including embryonic development, cell cycle, proliferation, differentiation, and apoptosis (8–15). They also have been associated with human diseases and a variety of cancers (16–26). Because the regulation of ODC and polyamine content is critical to cell proliferation (11), as well as in the origin and progression of neoplastic diseases (23, 24), ODC has been identified as an oncogenic enzyme, and the inhibitors of ODC and the polyamine pathway are important targets for therapeutic intervention in many cancers (6, 11).ODC is ubiquitously found in organisms ranging from bacteria to humans. It contains 461 amino acid residues in each monomer and is a 106-kDa homodimer with molecular 2-fold symmetry (27, 28). Importantly, ODC activity requires the formation of a dimer (29–31). X-ray structures of the ODC enzyme reveal that this dimer contains two active sites, both of which are formed at the interface between the N-terminal domain of one monomer, which provides residues involved in PLP interactions, and the C-terminal domain of the other subunit, which provides the residues that interact with substrate (27, 32–41).ODC undergoes a unique ubiquitin-independent proteasomal degradation via a direct interaction with the regulatory protein antizyme (AZ). Binding of AZ promotes the dissociation of the ODC homodimers and targets ODC for degradation by the 26 S proteasome (42–46). Current models of antizyme function indicate that increased polyamine levels promote the fidelity of the AZ mRNA translational frameshift, leading to increased concentrations of AZ (47). The AZ monomer selectively binds to dimeric ODC, thereby inactivating ODC by forming inactive AZ-ODC heterodimers (44, 48–50). AZ acts as a regulator of polyamine metabolism that inhibits ODC activity and polyamine transport, thus restricting polyamine levels (4, 5, 51, 52). When antizymes are overexpressed, they inhibit ODC and promote ubiquitin-independent proteolytic degradation of ODC. Because elevated ODC activity is associated with most forms of human malignancies (1), it has been suggested that antizymes may function as tumor suppressors.In contrast to the extensive studies on the oncogene ODC, the endogenous antizyme inhibitor (AZI) is less well understood. AZI is homologous to the enzyme ODC. It is a 448-amino acid protein with a molecular mass of 50 kDa. However, despite the homology between these proteins, AZI does not possess any decarboxylase activity. It binds to antizyme more tightly than does ODC and releases ODC from the ODC-antizyme complex (53, 54). Both the AZI and AZ proteins display rapid ubiquitin-dependent turnover within a few minutes to 1 h in vivo (5). However, AZ binding actually stabilizes AZI by inhibiting its ubiquitination (55).AZI, which inactivates all members of the AZ family (53, 56), restores ODC activity (54), and prevents the proteolytic degradation of ODC, may play a role in tumor progression. It has been reported that down-regulation of AZI is associated with the inhibition of cell proliferation and reduced ODC activity, presumably through the modulation of AZ function (57). Moreover, overexpression of AZI has been shown to increase cell proliferation and promote cell transformation (58–60). Furthermore, AZI is capable of direct interaction with cyclin D1, preventing its degradation, and this effect is at least partially independent of AZ function (60, 61). These results demonstrate a role for AZI in the positive regulation of cell proliferation and tumorigenesis.It is now known that ODC exists as a dimer and that AZI may exist as a monomer physiologically (62). Fig. 1 shows the dimeric structures of ODC (Fig. 1A) and AZI (Fig. 1B). Although structural studies indicate that both ODC and AZI crystallize as dimers, the dimeric AZI structure has fewer interactions at the dimer interface, a smaller buried surface area, and a lack of symmetry of the interactions between residues from the two monomers, suggesting that the AZI dimer may be nonphysiological (62). In this study, we identify the critical amino acid residues governing the difference in dimer formation between ODC and AZI. Our preliminary studies using analytical ultracentrifugation indicated that ODC exists as a dimer, whereas AZI exists in a concentration-dependent monomer-dimer equilibrium. Multiple sequence alignments of ODC and AZI from various species have shown that residues 277, 331, 332, and 389 are not conserved between ODC and AZI (Open in a separate windowFIGURE 1.Crystal structure and the amino acid residues at the dimer interface of human ornithine decarboxylase (hODC) and mouse antizyme inhibitor (mAZI). A, homodimeric structure of human ODC with the cofactor PLP analog, LLP (Protein Data Bank code 1D7K). B, putative dimeric structure of mouse AZI (Protein Data Bank code 3BTN). The amino acid residues in the dimer interface are shown as a ball-and-stick model. The putative AZ-binding site is colored in cyan. This figure was generated using PyMOL (DeLano Scientific LLC, San Carlos, CA).

TABLE 1

Amino acid residues at the dimer interface of human ODC and AZI

Disease Mutations in the Human Mitochondrial DNA Polymerase Thumb Subdomain Impart Severe Defects in Mitochondrial DNA Replication

Forty-five different point mutations in POLG, the gene encoding the catalytic subunit of the human mitochondrial DNA polymerase (pol γ), cause the early onset mitochondrial DNA depletion disorder, Alpers syndrome. Sequence analysis of the C-terminal polymerase region of pol γ revealed a cluster of four Alpers mutations at highly conserved residues in the thumb subdomain (G848S, c.2542g→a; T851A, c.2551a→g; R852C, c.2554c→t; R853Q, c.2558g→a) and two Alpers mutations at less conserved positions in the adjacent palm subdomain (Q879H, c.2637g→t and T885S, c.2653a→t). Biochemical characterization of purified, recombinant forms of pol γ revealed that Alpers mutations in the thumb subdomain reduced polymerase activity more than 99% relative to the wild-type enzyme, whereas the palm subdomain mutations retained 50–70% wild-type polymerase activity. All six mutant enzymes retained physical and functional interaction with the pol γ accessory subunit (p55), and none of the six mutants exhibited defects in misinsertion fidelity in vitro. However, differential DNA binding by these mutants suggests a possible orientation of the DNA with respect to the polymerase during catalysis. To our knowledge this study represents the first structure-function analysis of the thumb subdomain in pol γ and examines the consequences of mitochondrial disease mutations in this region.As the only DNA polymerase found in animal cell mitochondria, DNA polymerase γ (pol γ)³ bears sole responsibility for DNA synthesis in all replication and repair transactions involving mitochondrial DNA (1, 2). Mammalian cell pol γ is a heterotrimeric complex composed of one catalytic subunit of 140 kDa (p140) and two 55-kDa accessory subunits (p55) that form a dimer (3). The catalytic subunit contains an N-terminal exonuclease domain connected by a linker region to a C-terminal polymerase domain. Whereas the exonuclease domain contains essential motifs I, II, and III for its activity, the polymerase domain comprising the thumb, palm, and finger subdomains contains motifs A, B, and C that are crucial for polymerase activity. The catalytic subunit is a family A DNA polymerase that includes bacterial pol I and T7 DNA polymerase and possesses DNA polymerase, 3′ → 5′ exonuclease, and 5′-deoxyribose phosphate lyase activities (for review, see Refs. 1 and 2). The 55-kDa accessory subunit (p55) confers processive DNA synthesis and tight binding of the pol γ complex to DNA (4, 5).Depletion of mtDNA as well as the accumulation of deletions and point mutations in mtDNA have been observed in several mitochondrial disorders (for review, see Ref. 6). mtDNA depletion syndromes are caused by defects in nuclear genes responsible for replication and maintenance of the mitochondrial genome (7). Mutation of POLG, the gene encoding the catalytic subunit of pol γ, is frequently involved in disorders linked to mutagenesis of mtDNA (8, 9). Presently, more than 150 point mutations in POLG are linked with a wide variety of mitochondrial diseases, including the autosomal dominant (ad) and recessive forms of progressive external ophthalmoplegia (PEO), Alpers syndrome, parkinsonism, ataxia-neuropathy syndromes, and male infertility (tools.niehs.nih.gov/polg) (9).Alpers syndrome, a hepatocerebral mtDNA depletion disorder, and myocerebrohepatopathy are rare heritable autosomal recessive diseases primarily affecting young children (10–12). These diseases generally manifest during the first few weeks to years of life, and symptoms gradually develop in a stepwise manner eventually leading to death. Alpers syndrome is characterized by refractory seizures, psychomotor regression, and hepatic failure (11, 12). Mutation of POLG was first linked to Alpers syndrome in 2004 (13), and to date 45 different point mutations in POLG (18 localized to the polymerase domain) are associated with Alpers syndrome (9, 14, 15). However, only two Alpers mutations (A467T and W748S, both in the linker region) have been biochemically characterized (16, 17).During the initial cloning and sequencing of the human, Drosophila, and chicken pol γ genes, we noted a highly conserved region N-terminal to motif A in the polymerase domain that was specific to pol γ (18). This region corresponds to part of the thumb subdomain that tracks DNA into the active site of both Escherichia coli pol I and T7 DNA polymerase (19–21). A high concentration of disease mutations, many associated with Alpers syndrome, is found in the thumb subdomain.Here we investigated six mitochondrial disease mutations clustered in the N-terminal portion of the polymerase domain of the enzyme (Fig. 1A). Four mutations (G848S, c.2542g→a; T851A, c.2551a→g; R852C, c.2554c→t; R853Q, c.2558g→a) reside in the thumb subdomain and two (Q879H, c.2637g→t and T885S, c.2653a→t) are located in the palm subdomain. These mutations are associated with Alpers, PEO, mitochondrial myopathy, encephalopathy, lactic acidosis, and stroke-like episodes (MELAS), ataxia-neuropathy syndrome, Leigh syndrome, and myocerebrohepatopathy (

The Atrazine Catabolism Genes atzABC Are Widespread and Highly Conserved

Pseudomonas strain ADP metabolizes the herbicide atrazine via three enzymatic steps, encoded by the genes atzABC, to yield cyanuric acid, a nitrogen source for many bacteria. Here, we show that five geographically distinct atrazine-degrading bacteria contain genes homologous to atzA, -B, and -C. The sequence identities of the atz genes from different atrazine-degrading bacteria were greater than 99% in all pairwise comparisons. This differs from bacterial genes involved in the catabolism of other chlorinated compounds, for which the average sequence identity in pairwise comparisons of the known members of a class ranged from 25 to 56%. Our results indicate that globally distributed atrazine-catabolic genes are highly conserved in diverse genera of bacteria.Atrazine [2-chloro-4-(ethylamino)-6-(isopropylamino)- 1,3,5-triazine] is a herbicide used for controlling broad-leaf and grassy weeds and is relatively persistent in soils (51). Atrazine and other s-triazine compounds have been detected in ground and surface waters at levels exceeding the Environmental Protection Agency’s maximum contaminant level of 3 ppb (30).Microbial populations exposed to synthetic chlorinated compounds, such as atrazine, often respond by producing enzymes that degrade these molecules. Most of our current understanding of the genes and enzymes involved in atrazine degradation derives from studies using Pseudomonas strain ADP, in which the first three enzymatic steps in atrazine degradation have been defined (6, 14, 15, 48). The genes atz A, -B, and -C, which encode these enzymes, have been cloned and sequenced. Atrazine chlorohydrolase (AtzA), hydroxyatrazine ethylaminohydrolase (AtzB), and N-isopropylammelide isopropylaminohydrolase (AtzC) sequentially convert atrazine to cyanuric acid (6, 14, 15, 48) (Fig. ​(Fig.1).1). Cyanuric acid and related compounds are catabolized by many soil bacteria (10, 11, 17, 24, 26, 61), and by Pseudomonas sp. ADP, to carbon dioxide and ammonia (35). This provides the evolutionary pressure for the atzA, -B, and -C genes to permit bacterial growth on the more than one billion pounds of atrazine that have been applied to soils globally (20). Here we used a knowledge of the atzA, -B, and -C gene sequences to investigate the presence of homologous genes in other atrazine-degrading bacteria. In this study, we report that five atrazine-degrading microorganisms, which were recently isolated from geographically separated sites exposed to atrazine, contained nearly identical atzA, -B, and -C genes. Open in a separate windowFIG. 1Pathway for atrazine catabolism to cyanuric acid in Pseudomonas sp. strain ADP.

Atrazine-catabolizing bacteria used in this study.

Until recently, attempts at isolating bacteria (18) or fungi (27) that completely degrade atrazine to carbon dioxide, ammonia, and chloride were unsuccessful. While several microorganisms were shown to dealkylate atrazine, they were unable to displace the chlorine atom (41, 54). Since 1994, several research groups have independently isolated atrazine-degrading bacteria that displaced the chlorine atom and mineralized atrazine (3, 7, 13, 35, 39, 46). Six of these bacterial cultures, listed in Table ​Table1,1, were studied here, and the Clavibacter strain had been investigated previously (13).

TABLE 1

Recently isolated atrazine-catabolizing bacteria

Chimeric Nitrogenase-like Enzymes of (Bacterio)chlorophyll Biosynthesis

Nitrogenase-like light-independent protochlorophyllide oxidoreductase (DPOR) is involved in chlorophyll biosynthesis. Bacteriochlorophyll formation additionally requires the structurally related chlorophyllide oxidoreductase (COR). During catalysis, homodimeric subunit BchL₂ or ChlL₂ of DPOR transfers electrons to the corresponding heterotetrameric catalytic subunit, (BchNB)₂ or (ChlNB)₂. Analogously, subunit BchX₂ of the COR enzymes delivers electrons to subunit (BchYZ)₂. Various chimeric DPOR enzymes formed between recombinant subunits (BchNB)₂ and BchL₂ from Chlorobaculum tepidum or (ChlNB)₂ and ChlL₂ from Prochlorococcus marinus and Thermosynechococcus elongatus were found to be enzymatically active, indicating a conserved docking surface for the interaction of both DPOR protein subunits. Biotin label transfer experiments revealed the interaction of P. marinus ChlL₂ with both subunits, ChlN and ChlB, of the (ChlNB)₂ tetramer. Based on these findings and on structural information from the homologous nitrogenase system, a site-directed mutagenesis approach yielded 10 DPOR mutants for the characterization of amino acid residues involved in protein-protein interaction. Surface-exposed residues Tyr¹²⁷ of subunit ChlL, Leu⁷⁰ and Val¹⁰⁷ of subunit ChlN, and Gly⁶⁶ of subunit ChlB were found essential for P. marinus DPOR activity. Next, the BchL₂ or ChlL₂ part of DPOR was exchanged with electron-transferring BchX₂ subunits of COR and NifH₂ of nitrogenase. Active chimeric DPOR was generated via a combination of BchX₂ from C. tepidum or Roseobacter denitrificans with (BchNB)₂ from C. tepidum. No DPOR activity was observed for the chimeric enzyme consisting of NifH₂ from Azotobacter vinelandii in combination with (BchNB)₂ from C. tepidum or (ChlNB)₂ from P. marinus and T. elongatus, respectively.Chlorophyll and bacteriochlorophyll biosynthesis, as well as nitrogen fixation, are essential biochemical processes developed early in the evolution of life (1). During biological fixation of nitrogen, nitrogenase catalyzes the reduction of atmospheric dinitrogen to ammonia (2). Enzyme systems homologous to nitrogenase play a crucial role in the formation of the chlorin and bacteriochlorin ring system of chlorophylls (Chl)2 and bacteriochlorophylls (Bchl) (3, 4) (Fig. 1a). For the synthesis of both Chl and Bchl, the stereospecific reduction of the C-17-C-18 double bond of ring D of protochlorophyllide (Pchlide) catalyzed by the nitrogenase-like enzyme light-independent (dark-operative) protochlorophyllide oxidoreductase (DPOR) results in the formation of chlorophyllide (Chlide) (Fig. 1a, left) (5, 6). DPOR enzymes consist of three protein subunits which are designated BchN, BchB and BchL in Bchl-synthesizing organisms and ChlN, ChlB and ChlL in Chl-synthesizing organisms. A second reduction step at ring B (C-7-C-8) unique to the synthesis of Bchl converts the chlorin Chlide into a bacteriochlorin ring structure to form bacteriochlorophyllide (Bchlide) (Fig. 1a, right, Bchlide). This reaction is catalyzed by another nitrogenase-like enzyme, termed chlorophyllide oxidoreductase (COR) (7). COR enzymes are composed of subunits BchY, BchZ, and BchX.Open in a separate windowFIGURE 1.Comparison of the three subunit enzymes DPOR, COR, and nitrogenase. a, during Chl and Bchl biosynthesis, ring D is stereospecifically reduced by the nitrogenase-like enzyme DPOR (subunit composition BchL₂/(BchNB)₂ or ChlL₂/(ChlNB)₂) leading to the chlorin Chlide. Subunits N, B, and L are named ChlN, ChlB, and ChlL in Chl-synthesizing organisms and BchN, BchB, and BchL in Bchl-synthesizing organisms. The synthesis of Bchl additionally requires the stereospecific B ring reduction by a second nitrogenase-like enzyme called COR, with the subunit composition BchX₂/(BchYZ)₂. COR catalyzes the formation of the bacteriochlorin Bchlide. Subunits Y, Z, and X of the COR enzyme are named BchY, BchZ, and BchX. b, the homologous nitrogenase complex has the subunit composition NifH₂/(NifD/NifK)₂. Rings A–E and the carbon atoms are designated according to IUPAC nomenclature (41). R is either a vinyl or an ethyl moiety. The position marked by an asterisk indicates either a vinyl or a hydroxyethyl moiety (42).All subunits share significant amino acid sequence homology to the corresponding subunits of nitrogenase, which are designated NifD, NifK, and NifH, respectively (1) (compare Fig. 1, a and b). Whereas subunits BchL or ChlL, BchX and NifH exhibit a sequence identity at the amino acid level of ∼33%, subunits BchN or ChlN, BchY, NifD, and BchB or ChlB, BchZ, and NifK, respectively, show lower sequence identities of ∼15% (1). For all enzymes a common oligomeric protein architecture has been proposed consisting of the heterotetrameric complexes (BchNB)₂ or (ChlNB)₂, (BchYZ)₂, and (NifD/NifK)₂, which are completed by a homodimeric protein subunit BchL₂ or ChlL₂, BchX₂, and NifH₂, respectively (compare Fig. 1, a and b) (3, 7, 8).Nitrogenase is a well characterized protein complex that catalyzes the reduction of nitrogen to ammonia in a reaction that requires at least 16 molecules of MgATP (2, 9, 10). During nitrogenase catalysis, subunit NifH₂ (Fe protein) associates with and dissociates from the (NifD/NifK)₂ complex (MoFe protein). Binding, hydrolysis of MgATP and structural rearrangements are coupled to sequential intersubunit electron transfer. For this purpose, NifH₂ contains an ATP-binding motif and an intersubunit [4Fe-4S] cluster coordinated by two cysteine residues from each NifH monomer (1, 11). Electrons from this [4Fe-4S] cluster are transferred via a [8Fe-7S] cluster (P-cluster) onto the [1Mo-7Fe-9S-X-homocitrate] cluster (MoFe cofactor). Both of the latter clusters are located on (NifD/NifK)₂, where dinitrogen is reduced to ammonia (10). Three-dimensional structures of NifH₂ in complex with (NifD/NifK)₂ revealed a detailed picture of the dynamic interaction of both subcomplexes (8, 12).Based on biochemical and bioinformatic approaches, it has been proposed that the initial steps of DPOR reaction strongly resemble nitrogenase catalysis. Key amino acid residues essential for DPOR function have been identified by mutagenesis of the enzyme from Chlorobaculum tepidum (formerly denoted as Chlorobium tepidum) (3). The catalytic mechanism of DPOR includes the electron transfer from a “plant-type” [2Fe-2S] ferredoxin onto the dimeric DPOR subunit, BchL₂, carrying an intersubunit [4Fe-4S] redox center coordinated by Cys⁹⁷ and Cys¹³¹ in C. tepidum. Analogous to nitrogenase, Lys¹⁰ in the phosphate-binding loop (P-loop) and Leu¹²⁶ in the switch II region of DPOR were found essential for DPOR catalysis. Moreover, it was shown that the BchL₂ protein from C. tepidum does not form a stable complex with the catalytic (BchNB)₂ subcomplex. Therefore, a transient interaction responsible for the electron transfer onto protein subunit (BchNB)₂ has been proposed (3).The subsequent [Fe-S] cluster-dependent catalysis and the specific substrate recognition at the active site located on subunit (BchNB)₂ are unrelated to nitrogenase. The (BchNB)₂ subcomplex was shown to carry a second [4Fe-4S] cluster, which was proposed to be ligated by Cys²¹, Cys⁴⁶, and Cys¹⁰³ of the BchN subunit and Cys⁹⁴ of subunit BchB (C. tepidum numbering) (3). No evidence for any type of additional cofactor was obtained from biochemical and EPR spectroscopic analyses (5, 13). Thus, despite the same common oligomeric architecture, the catalytic subunits (BchNB)₂ and (ChlNB)₂ clearly differ from the corresponding nitrogenase complex, as no molybdenum-containing cofactor or P-cluster equivalent is employed (5, 14). From these results it was concluded that electrons from the [4Fe-4S] cluster of (BchNB)₂ or (ChlNB)₂ are transferred directly onto the Pchlide substrate at the active site of DPOR.The second nitrogenase-like enzyme, COR, catalyzes the reduction of ring B of Chlide during the biosynthesis of Bchl (7). Therefore, an accurate discrimination of the ring systems of the individual substrates is required. COR subunits share an overall amino acid sequence identity of 15–22% for BchY and BchZ and 31–35% for subunit BchX when compared with the corresponding DPOR subunits (supplemental Figures S2–S4). In amino acid sequence alignments of BchX proteins with the closely related BchL or ChlL subunits of DPOR, both cysteinyl ligands responsible for [4Fe-4S] cluster formation and residues for ATP binding are conserved (1). Furthermore, all cysteinyl residues characterized as ligands for a catalytic [4Fe-4S] cluster in (BchNB)₂ or (ChlNB)₂ are conserved in the sequences of subunits BchY and BchZ of COR (7). These findings correspond to a recent EPR study in which a characteristic signal for a [4Fe-4S] cluster was obtained for the COR subunit BchX₂ as well as for subunit (BchYZ)₂ (15). These results indicate that the catalytic mechanism of COR strongly resembles DPOR catalysis. In vitro assays for nitrogenase as well as for DPOR and COR make use of the artificial electron donor dithionite in the presence of high concentrations of ATP (7, 16, 17).

TABLE 1

Amino acid sequence identities of the individual subunits of DPOR, COR, and nitrogenaseAmino acid sequences of the individual subunits of DPOR, COR, and nitrogenase employed in the present study (compareFig. 3A) were aligned by using the ClustalW method in MegAlign (DNASTAR), and sequence identities were calculated.

Molecular and Biochemical Characterization of the Protein Template Controlling Biosynthesis of the Lipopeptide Lichenysin

Lichenysins are surface-active lipopeptides with antibiotic properties produced nonribosomally by several strains of Bacillus licheniformis. Here, we report the cloning and sequencing of an entire 26.6-kb lichenysin biosynthesis operon from B. licheniformis ATCC 10716. Three large open reading frames coding for peptide synthetases, designated licA, licB (three modules each), and licC (one module), could be detected, followed by a gene, licTE, coding for a thioesterase-like protein. The domain structure of the seven identified modules, which resembles that of the surfactin synthetases SrfA-A to -C, showed two epimerization domains attached to the third and sixth modules. The substrate specificity of the first, fifth, and seventh recombinant adenylation domains of LicA to -C (cloned and expressed in Escherichia coli) was determined to be Gln, Asp, and Ile (with minor Val and Leu substitutions), respectively. Therefore, we suppose that the identified biosynthesis operon is responsible for the production of a lichenysin variant with the primary amino acid sequence l-Gln–l-Leu–d-Leu–l-Val–l-Asp–d-Leu–l-Ile, with minor Leu and Val substitutions at the seventh position.Many strains of Bacillus are known to produce lipopeptides with remarkable surface-active properties (11). The most prominent of these powerful lipopeptides is surfactin from Bacillus subtilis (1). Surfactin is an acylated cyclic heptapeptide that reduces the surface tension of water from 72 to 27 mN m⁻¹ even in a concentration below 0.05% and shows some antibacterial and antifungal activities (1). Some B. subtilis strains are also known to produce other, structurally related lipoheptapeptides (Table ​(Table1),1), like iturin (32, 34) and bacillomycin (3, 27, 30), or the lipodecapeptides fengycin (50) and plipastatin (29).

TABLE 1

Lipoheptapeptide antibiotics of Bacillus spp.

Peptidoglycan Fine Structure of the Radiotolerant Bacterium Deinococcus radiodurans Sark

Peptidoglycan from Deinococcus radiodurans was analyzed by high-performance liquid chromatography and mass spectrometry. The monomeric subunit was: N-acetylglucosamine–N-acetylmuramic acid–l-Ala–d-Glu-(γ)–l-Orn-[(δ)Gly-Gly]–d-Ala–d-Ala. Cross-linkage was mediated by (Gly)₂ bridges, and glycan strands were terminated in (1→6)anhydro-muramic acid residues. Structural relations with the phylogenetically close Thermus thermophilus are discussed.The gram-positive bacterium Deinococcus radiodurans is remarkable because of its extreme resistance to ionizing radiation (14). Phylogenetically the closest relatives of Deinococcus are the extreme thermophiles of the genus Thermus (4, 11). In 16S rRNA phylogenetic trees, the genera Thermus and Deinococcus group together as one of the older branches in bacterial evolution (11). Both microorganisms have complex cell envelopes with outer membranes, S-layers, and ornithine-Gly-containing mureins (7, 12, 19, 20, 22, 23). However, Deinococcus and Thermus differ in their response to the Gram reaction, having positive and negative reactions, respectively (4, 14). The murein structure for Thermus thermophilus HB8 has been recently elucidated (19). Here we report the murein structure of Deinococcus radiodurans with similar detail.D. radiodurans Sark (23) was used in the present study. Cultures were grown in Luria-Bertani medium (13) at 30°C with aeration. Murein was purified and subjected to amino acid and high-performance liquid chromatography (HPLC) analyses as previously described (6, 9, 10, 19). For further analysis muropeptides were purified, lyophilized, and desalted as reported elsewhere (6, 19). Purified muropeptides were subjected to plasma desorption linear time-of-flight mass spectrometry (PDMS) as described previously (1, 5, 16, 19). Positive and negative ion mass spectra were obtained on a short linear ²⁵²californium time-of-flight instrument (BioIon AB, Uppsala, Sweden). The acceleration voltage was between 17 and 19 kV, and spectra were accumulated for 1 to 10 million fission events. Calibration of the mass spectra was done in the positive ion mode with H⁺ and Na⁺ ions and in the negative ion mode with H⁻ and CN⁻ ions. Calculated m/z values are based on average masses.Amino acid analysis of muramidase (Cellosyl; Hoechst, Frankfurt am Main, Germany)-digested sacculi (50 μg) revealed Glu, Orn, Ala, and Gly as the only amino acids in the muramidase-solubilized material. Less than 3% of the total Orn remained in the muramidase-insoluble fraction, indicating an essentially complete solubilization of murein.Muramidase-digested murein samples (200 μg) were analyzed by HPLC as described in reference 19. The muropeptide pattern (Fig. ​(Fig.1)1) was relatively simple, with five dominating components (DR5 and DR10 to DR13 [Fig. 1]). The muropeptides resolved by HPLC were collected, desalted, and subjected to PDMS. The results are presented in Table ​Table11 compared with the m/z values calculated for best-matching muropeptides made up of N-acetylglucosamine (GlucNAc), N-acetylmuramic acid (MurNAc), and the amino acids detected in the murein. The more likely structures are shown in Fig. ​Fig.1.1. According to the m/z values, muropeptides DR1 to DR7 and DR9 were monomers; DR8, DR10, and DR11 were dimers; and DR12 and DR13 were trimers. The best-fitting structures for DR3 to DR8, DR11, and DR13 coincided with muropeptides previously characterized in T. thermophilus HB8 (19) and had identical retention times in comparative HPLC runs. The minor muropeptide DR7 (Fig. ​(Fig.1)1) was the only one detected with a d-Ala–d-Ala dipeptide and most likely represents the basic monomeric subunit. The composition of the major cross-linked species DR11 and DR13 confirmed that cross-linking is mediated by (Gly)₂ bridges, as proposed previously (20). Open in a separate windowFIG. 1HPLC muropeptide elution patterns of murein purified from D. radiodurans. Muramidase-digested murein samples were subjected to HPLC analysis, and the A₂₀₄ of the eluate was recorded. The most likely structures for each muroeptide as deduced by PDMS are shown. The position of residues in brackets is the most likely one as deduced from the structures of other muropeptides but could not be formally demonstrated. R = GlucNac–MurNac–l-Ala–d-Glu-(γ)→.

TABLE 1

Calculated and measured m/z values for the molecular ions of the major muropeptides from D. radiodurans

Biological Activity of Nerve Growth Factor Precursor Is Dependent upon Relative Levels of Its Receptors

Nerve growth factor (NGF) is produced as a precursor called pro-nerve growth factor (proNGF), which is secreted by many tissues and is the predominant form of NGF in the central nervous system. In Alzheimer disease brain, cholinergic neurons degenerate and can no longer transport NGF as efficiently, leading to an increase in untransported NGF in the target tissue. The protein that accumulates in the target tissue is proNGF, not the mature form. The role of this precursor is controversial, and both neurotrophic and apoptotic activities have been reported for recombinant proNGFs. Differences in the protein structures, protein expression systems, methods used for protein purification, and methods used for bioassay may affect the activity of these proteins. Here, we show that proNGF is neurotrophic regardless of mutations or tags, and no matter how it is purified or in which system it is expressed. However, although proNGF is neurotrophic under our assay conditions for primary sympathetic neurons and for pheochromocytoma (PC12) cells, it is apoptotic for unprimed PC12 cells when they are deprived of serum. The ratio of tropomyosin-related kinase A to p75 neurotrophin receptor is low in unprimed PC12 cells compared with primed PC12 cells and sympathetic neurons, altering the balance of proNGF-induced signaling to favor apoptosis. We conclude that the relative level of proNGF receptors determines whether this precursor exhibits neurotrophic or apoptotic activity.Nerve growth factor (NGF)³ regulates neuronal survival, neurite outgrowth, and differentiation in the peripheral and central nervous systems (1). The mature form of NGF forms a non-covalent homodimer and binds with high affinity (k_d ≈ 10⁻¹¹ m) to tropomyosin-related kinase A (TrkA) and with low affinity (k_d ≈ 10⁻⁹ m) to the common neurotrophin receptor p75^NTR (p75 neurotrophin receptor) (2). NGF promotes cell survival and growth in cells expressing TrkA through activation of the phosphatidylinositol 3-kinase/AKT pathway and the Ras/mitogen-activated protein kinase (MAPK) pathway (3, 4). p75^NTR plays diverse roles, ranging from cell survival to cell death depending on the cellular context in which it is expressed. Through activation of the NF-κB pathway, p75^NTR can contribute to cell survival in sensory neurons (5), it is involved in axonal growth via regulation of Rho activity (6), and it can interact with Trks to enhance neurotrophin affinity (at low concentration of ligand) and specificity of binding to Trks (7–9). High levels of p75^NTR expression can induce apoptosis when there are low levels of Trk or when Trk is absent (10, 11). Apoptosis occurs through increased ceramide production (12), activation of c-Jun N-terminal kinase (JNK1), and p53 (10, 13). p75^NTR requires a co-receptor called sortilin to induce cell death (14).NGF is produced as a precursor called pro-nerve growth factor (proNGF) (15). ProNGF is secreted by many tissues such as prostate cells, spermatids, hair follicles, oral mucosal keratinocytes, sympathetic neurons, cortical astrocytes, heart, and spleen (16–20). ProNGF is the predominant form of NGF in the central and peripheral nervous systems, whereas little or no mature NGF can be detected (21–24). In Alzheimer disease brain, retrograde transport from the cortex and hippocampus to basal forebrain cholinergic neurons is reduced as these neurons degenerate, with concomitant proNGF accumulation in the cortex and hippocampus (21, 23). This suggested that proNGF mediates biological activity besides its prodomain function of promoting protein folding and regulation of neurotrophin secretion (25–28). To study the role of proNGF protein in vitro, point mutations were inserted at the cleavage site used by furin, a proprotein convertase known to cleave proNGF (29), to minimize the conversion of proNGF to mature NGF. The resulting recombinant, cleavage-resistant proNGFs reportedly exhibit either apoptotic activity (30, 31) or neurotrophic activity (32, 33). These recombinant proteins differ in several ways (ProNGF(R−1G)ProNGFhisProNGFEProNGF123WT-NGFhisMutations−1 (R to G)−2 and −1 (RR to AA), 118 and 119 (RR to AA)−1 and +1 (RS to AA)−73 and −72 (RR to AA), −43 and −42 (KKRR to KAAR), −2 and −1 (KR to AA)None: cleavable proNGFTagNo tagHistidine tagNo tagNo tagHistidine tagExpression systemInsect cellsInsect cells, mammalian cellsBacteriaInsect cellsInsect cells, mammalian cellsPurificationNo purificationNickel columnRefolded from inclusion bodies, FPLCCation exchange chromatography, immunoaffinity chromatographyNickel columnOpen in a separate window     

Glycopeptide-preferring Polypeptide GalNAc Transferase 10 (ppGalNAc T10), Involved in Mucin-type O-Glycosylation, Has a Unique GalNAc-O-Ser/Thr-binding Site in Its Catalytic Domain Not Found in ppGalNAc T1 or T2

Mucin-type O-gly co sy la tion is initiated by a large family of UDP-GalNAc:polypeptide α-N-acetylgalactosaminyltransferases (ppGalNAc Ts) that transfer GalNAc from UDP-GalNAc to the Ser and Thr residues of polypeptide acceptors. Some members of the family prefer previously gly co sylated peptides (ppGalNAc T7 and T10), whereas others are inhibited by neighboring gly co sy la tion (ppGalNAc T1 and T2). Characterizing their peptide and glycopeptide substrate specificity is critical for understanding the biological role and significance of each isoform. Utilizing a series of random peptide and glycopeptide substrates, we have obtained the peptide and glycopeptide specificities of ppGalNAc T10 for comparison with ppGalNAc T1 and T2. For the glycopeptide substrates, ppGalNAc T10 exhibited a single large preference for Ser/Thr-O-GalNAc at the +1 (C-terminal) position relative to the Ser or Thr acceptor site. ppGalNAc T1 and T2 revealed no significant enhancements suggesting Ser/Thr-O-GalNAc was inhibitory at most positions for these isoforms. Against random peptide substrates, ppGalNAc T10 revealed no significant hydrophobic or hydrophilic residue enhancements, in contrast to what has been reported previously for ppGalNAc T1 and T2. Our results reveal that these transferases have unique peptide and glycopeptide preferences demonstrating their substrate diversity and their likely roles ranging from initiating transferases to filling-in transferases.Mucin-type O-glycosylation is a common post-translational modification of secreted and membrane-associated proteins. O-Glycan biosynthesis is initiated by the transfer of GalNAc from UDP-GalNAc to the hydroxyl groups of serine or threonine residues in a polypeptide, catalyzed by a family of polypeptide N-α-acetylgalactosaminyltransferases (ppGalNAc Ts).⁵ To date, 16 mammalian members have been reported in the literature (1–16) with a total of at least 20 members currently present in the human genome data base. Multiple members of the ppGalNAc T family have also been identified in Drosophila (9, 10, 14), Caenorhabditis elegans (3, 8), and single and multicellular organisms (17–20). Several members show close sequence orthologues across species suggesting that the ppGalNAc Ts are responsible for biologically significant functions that have been conserved during evolution. For example, in Drosophila four isoforms have close sequence orthologues to the mammalian transferases. Of the two that have been recently compared, nearly identical peptide substrate specificities have been observed between the fly and mammals, suggesting common but presently unknown functions preserved across these diverse species (21).Recently, several ppGalNAc T isoforms have been shown to be important for normal development or cellular processes. For example, inactive mutations in the fly PGANT35A (the T11 orthologue in mammals) are lethal because of the disruption of the tracheal tube structures (9, 10, 22), whereas mutations in PGANT3 alter epithelial cell adhesion in the Drosophila wing blade resulting in wing blistering (23). In humans, mutations in ppGalNAc T3 are associated with familial tumoral calcinosis, the result of the abnormal processing and secretion of the phosphaturic factor FGF23 (24, 25). Human ppGalNAc T14 has been suggested to modulate apoptotic signaling in tumor cells by its glycosylation of the proapoptotic receptors DLR4 and DLR5 (26), and very recently the specific O-glycosylation of the TGFB-II receptor (ActR-II) by the GalNTL1 has been shown to modulate its signaling in development (16).Historically, the major targets of the ppGalNAc Ts have been thought to be heavily O-glycosylated mucin domains of membrane and secreted glycoproteins. Such domains typically contain 15–30% Ser or Thr, which are highly (>50%) substituted by GalNAc. One question in the field is as follows. How is this high degree of peptide core glycosylation achieved and is it related to the large number of ppGalNAc isoforms, some of which may even have specific mucin domain preferences? Interestingly, some members of the ppGalNAc T family are known to prefer substrates that have been previously modified with O-linked GalNAc on nearby Ser/Thr residues, hence having so-called glycopeptide or filling-in activities, i.e. ppGalNAc T7 and T10 (8, 27–29). Others simply possess altered preferences against glycopeptide substrates, i.e. ppGalNAc T2 and T4 (30–33), or may be inhibited by neighboring glycosylation, i.e. ppGalNAc T1 and T2 (29, 34, 35). These latter transferases have been called early or initiating transferases, preferring nonglycosylated over-glycosylated substrates. Presently, little is known about which factors dictate the different peptide/glycopeptide specificities among the ppGalNAc Ts.The ppGalNAc Ts consist of an N-terminal catalytic domain tethered by a short linker to a C-terminal ricin-like lectin domain containing three recognizable carbohydrate-binding sites (36). Because ppGalNAc T7 and T10 prefer to transfer GalNAc to glycopeptide acceptors, it has been widely assumed that their C-terminal lectin domains would play significant roles in this activity, as has been demonstrated for other family members (27, 28, 32). Recently, Kubota et al. (37) solved the crystal structure of ppGalNAc T10 in complex with Ser-GalNAc specifically bound to its lectin domain. In this work (37), the authors further demonstrated that a T10 lectin domain mutant indeed had altered specificity against GalNAc-containing glycopeptide substrates when the acceptor Ser/Thr site was distal from the pre-existing glycopeptide GalNAc site. However, it was also observed that the lectin mutant still possessed relatively unaltered glycopeptide activity when the acceptor Ser/Thr site was directly N-terminal of a pre-existing glycopeptide GalNAc site. Kubota et al. (37) therefore concluded that for ppGalNAc T10, both its lectin and indeed its catalytic domain must contain distinct peptide GalNAc recognition sites. In support of this, Raman et al. (33) have shown that the complete removal of the ppGalNAc T10 lectin domain only slightly alters its specificity against distal glycopeptide substrates while showing no difference in its ability to glycosylate residues directly N-terminal of an existing site of glycosylation. Thus, it seems that the catalytic domain of ppGalNAc T10 may have specific requirements for a peptide O-linked GalNAc in at least the +1 position (toward the C terminus) of residues being glycosylated. As no systematic determination of the glycopeptide binding properties of the ppGalNAc Ts catalytic domain has been performed, it is unknown whether additional GalNAc peptide-binding sites exist in T10 or, for that matter, any of the other ppGalNAc Ts.We have recently reported the use of oriented random peptide substrates, GAGA(X)_nT(X)_nAGAGK (where X indicates randomized amino acid positions and n = 3 and 5) for determining the peptide substrate specificities of mammalian ppGalNAc T1, T2, and their fly orthologues (21, 38). In the present work, we extend this approach to the determination of the catalytic domain glycopeptide (Ser/Thr-O-GalNAc) substrate preferences for ppGalNAc T1, T2, and T10 employing two n = 4 oriented random glycopeptide libraries (21). Interestingly, ppGalNAc T10 displays few significant enhancements and specifically lacks the Pro residue enhancements observed for ppGalNAc T1 and T2. These findings further demonstrate the vast substrate diversity of the catalytic domains of the ppGalNAc T family of transferases.

TABLE 1

ppGalNAc transferase random substrates utilized in this workPVI, PVII, GP-I, and GP-II random (glyco)peptide substrates.

Protein Identification Using Top-Down Spectra

In the last two years, because of advances in protein separation and mass spectrometry, top-down mass spectrometry moved from analyzing single proteins to analyzing complex samples and identifying hundreds and even thousands of proteins. However, computational tools for database search of top-down spectra against protein databases are still in their infancy. We describe MS-Align+, a fast algorithm for top-down protein identification based on spectral alignment that enables searches for unexpected post-translational modifications. We also propose a method for evaluating statistical significance of top-down protein identifications and further benchmark various software tools on two top-down data sets from Saccharomyces cerevisiae and Salmonella typhimurium. We demonstrate that MS-Align+ significantly increases the number of identified spectra as compared with MASCOT and OMSSA on both data sets. Although MS-Align+ and ProSightPC have similar performance on the Salmonella typhimurium data set, MS-Align+ outperforms ProSightPC on the (more complex) Saccharomyces cerevisiae data set.In the past two decades, proteomics was dominated by bottom-up mass spectrometry that analyzes digested peptides rather than intact proteins. Bottom-up approaches, although powerful, do have limitations in analyzing protein species, e.g. various proteolytic forms of the same protein or various protein isoforms resulting from alternative splicing. Top-down mass spectrometry focuses on analyzing intact proteins and large peptides (1–10) and has advantages in localizing multiple post-translational modifications (PTMs)¹ in a coordinated fashion (e.g. combinatorial PTM code) and identifying multiple protein species (e.g. proteolytically processed protein species) (11). Until recently, most top-down studies were limited to single purified proteins (12–15). Top-down studies of protein mixtures were restricted by difficulties in separating and fragmenting intact proteins and a shortage of robust computational tools.In the last two years, because of advances in protein separation and top-down instrumentation, top-down mass spectrometry moved from analyzing single proteins to analyzing complex samples containing hundreds and even thousands of proteins (16–21). Because algorithms for interpreting top-down spectra are still in their infancy, many recent developments include computational innovations in protein identification.Because top-down spectra are complex, the first step in top-down spectral interpretation is usually spectral deconvolution, which converts a complex top-down spectrum to a list of monoisotopic masses (a deconvolved spectrum). Every protein (possibly with modifications) can be scored against a top-down deconvoluted spectrum, resulting in a Protein-Spectrum-Match (PrSM). The top-down protein identification problem is finding a protein in a database with the highest scoring PrSM for a top-down spectrum and further output the PrSM if it is statistically significant. There are several software tools for top-down protein identification (SoftwareIdentification of unexpected modificationsProteogenomics search against 6-frame translationSpeedEstimation of statistical significanceProSightPC+/−^a+Fast/Slow^b+PIITA+/−−Fast−UStag++Fast−MS-TopDown+−Slow−MS-Align+++Fast+Open in a separate window^a ProSightPC has various search modes that contribute to bridging the gap between blind and restrictive modes of MS/MS database search. It can identify truncated proteins by using biomarker search and identify unexpected modifications by using Δm mode and setting the error tolerance of precursor mass to a large value (e.g., 1999 Da). However, it is not designed for identifying truncated proteins with unexpected PTMs which are not represented in the “shotgun annotated” database.^b In its most advances mode, ProSightPC can search the annotated top-down database that contains various protein species. However, ProSightPC searches in this mode become an order of magnitude slower.

• ProSightPC—ProSightPC is the most commonly used tool for top-down protein identification (22, 23). ProSightPC searches spectra against a “shotgun annotated” protein database, which is generated by considering all expected PTMs. The “shotgun annotated” protein database is much larger than the original protein database. ProSightPC can identify some (but not all) proteins with unexpected PTMs using advanced search options, such as biomarker search and Δm mode, but it is not designed for identifying truncated proteins with unexpected PTMs that are not represented in the “shotgun annotated” database. ProSightPC is a fast tool that reports the statistical significance of PrSMs.
• PIITA—Unlike ProSightPC, PIITA (19) is a precursor independent method that uses only fragment ions for protein identification. It is capable of identifying protein species with unexpected PTMs on N- or C-termini, but it cannot directly identify protein species with PTMs on both N- and C-termini. PIITA is a fast tool that provides FIT scores and Δ scores rather than statistical significance estimates.
• USTag—Unique Sequence Tag (USTag) (17) generates long (6 amino acids or longer) peptide sequence tags to identify PrSMs. This approach, although fast, relies on long peptide sequence tags that may be difficult to obtain for some spectra. It also does not provide an estimate of the statistical significance of PrSMs.
• MS-TopDown—MS-TopDown (24) is based on spectral alignment (25). MS-TopDown allows one to match top-down spectra to proteins with unexpected PTMs, i.e. without knowing which PTMs are present in the sample. However, MS-TopDown is rather slow when searching against large proteomes and does not provide the statistical significance of PrSMs, making it difficult to select good PrSMs.
• In addition, MASCOT, SEQUEST, and OMSSA (16, 26, 27) have been used for top-down protein identification.

We describe MS-Align+, a fast software tool for top-down protein identification. MS-Align+ shares the spectral alignment approach with MS-TopDown, but greatly improves on speed, statistical analysis (providing E-values of PrSMs), and the number of identified PrSMs (e.g. by finding spectral alignments between spectra and truncated proteins). We benchmarked various tools for top-down protein identification on two data sets from Saccharomyces cerevisiae (SC) and Salmonella typhimurium (ST). We demonstrate that MS-Align+ significantly increase the number of identified spectra as compared with MASCOT and OMSSA on both data sets. Although MS-Align+ and ProSightPC have similar performance on the ST data set, MS-Align+ outperforms ProSightPC on the more complex SC data set.