A novel multifunctional O-methyltransferase implicated in a dual methylation pathway associated with lignin biosynthesis in loblolly pine

S-adenosyl-l-methionine (SAM)-dependent O-methyltransferases (OMTs) catalyze the methylation of hydroxycinnamic acid derivatives for the synthesis of methylated plant polyphenolics, including lignin. The distinction in the extent of methylation of lignins in angiosperms and gymnosperms, mediated by substrate-specific OMTs, represents one of the fundamental differences in lignin biosynthesis between these two classes of plants. In angiosperms, two types of structurally and functionally distinct lignin pathway OMTs, caffeic acid 3-O-methyltransferases (CAOMTs) and caffeoyl CoA 3-O-methyltransferases (CCoAOMTs), have been reported and extensively studied. However, little is known about lignin pathway OMTs in gymnosperms. We report here the first cloning of a loblolly pine (Pinus taeda) xylem cDNA encoding a multifunctional enzyme, SAM:hydroxycinnamic Acids/hydroxycinnamoyl CoA Esters OMT (AEOMT). The deduced protein sequence of AEOMT is partially similar to, but clearly distinguishable from, that of CAOMTs and does not exhibit any significant similarity with CCoAOMT protein sequences. However, functionally, yeast-expressed AEOMT enzyme catalyzed the methylation of CAOMT substrates, caffeic and 5-hydroxyferulic acids, as well as CCoAOMT substrates, caffeoyl CoA and 5-hydroxyferuloyl CoA esters, with similar specific activities and was completely inactive with substrates associated with flavonoid synthesis. The lignin-related substrates were also efficiently methylated in crude extracts of loblolly pine secondary xylem. Our results support the notion that, in the context of amino acid sequence and biochemical function, AEOMT represents a novel SAM-dependent OMT, with both CAOMT and CCoAOMT activities and thus the potential to mediate a dual methylation pathway in lignin biosynthesis in loblolly pine xylem.

Synthesis of medium-chain-length polyhydroxyalkanoates in Arabidopsis thaliana using intermediates of peroxisomal fatty acid β-oxidation

Polyhydroxyalkanoate (PHA) is a family of polymers composed primarily of R-3-hydroxyalkanoic acids. These polymers have properties of biodegradable thermoplastics and elastomers. Medium-chain-length PHAs (MCL-PHAs) are synthesized in bacteria by using intermediates of the β-oxidation of alkanoic acids. To assess the feasibility of producing MCL-PHAs in plants, Arabidopsis thaliana was transformed with the PhaC1 synthase from Pseudomonas aeruginosa modified for peroxisome targeting by addition of the carboxyl 34 amino acids from the Brassica napus isocitrate lyase. Immunocytochemistry demonstrated that the modified PHA synthase was appropriately targeted to leaf-type peroxisomes in light-grown plants and glyoxysomes in dark-grown plants. Plants expressing the PHA synthase accumulated electron-lucent inclusions in the glyoxysomes and leaf-type peroxisomes, as well as in the vacuole. These inclusions were similar to bacterial PHA inclusions. Analysis of plant extracts by GC and mass spectrometry demonstrated the presence of MCL-PHA in transgenic plants to approximately 4 mg per g of dry weight. The plant PHA contained saturated and unsaturated 3-hydroxyalkanoic acids ranging from six to 16 carbons with 41% of the monomers being 3-hydroxyoctanoic acid and 3-hydroxyoctenoic acid. These results indicate that the β-oxidation of plant fatty acids can generate a broad range of R-3-hydroxyacyl-CoA intermediates that can be used to synthesize MCL-PHAs.

Human fatty acid synthase: Assembling recombinant halves of the fatty acid synthase subunit protein reconstitutes enzyme activity

Our model of the native fatty acid synthase (FAS) depicts it as a dimer of two identical multifunctional proteins (Mr ≈ 272,000) arranged in an antiparallel configuration so that the active Cys-SH of the β-ketoacyl synthase of one subunit (where the acyl group is attached) is juxtaposed within 2 Å of the pantetheinyl-SH of the second subunit (where the malonyl group is bound). This arrangement generates two active centers for fatty acid synthesis and predicts that if we have two appropriate halves of the monomer, we should be able to reconstitute an active fatty acid-synthesizing site. We cloned, expressed, and purified catalytically active thioredoxin (TRX) fusion proteins of the NH2-terminal half of the human FAS subunit protein (TRX-hFAS-dI; residues 1–1,297; Mr ≈ 166) and of the C-terminal half (TRX-hFAS-dII-III; residues 1,296–2,504; Mr ≈ 155). Adding equivalent amounts of TRX-hFAS-dI and TRX-hFAS-dII-III to a reaction mixture containing acetyl-CoA, malonyl-CoA, and NADPH resulted in the synthesis of long-chain fatty acids. The rate of synthesis was dependent upon the presence of both recombinant proteins and reached a constant level when they were present in equivalent amounts, indicating that the reconstitution of an active fatty acid-synthesizing site required the presence of every partial activity associated with the subunit protein. Analyses of the product acids revealed myristate to be the most abundant with small amounts of palmitate and stearate, possibly because of the way the fused recombinant proteins interacted with each other so that the thioesterase hydrolyzed the acyl group in its myristoyl state. The successful reconstitution of the human FAS activity from its domain I and domains II and III fully supports our model for the structure–function relationship of FAS in animal tissues.

Cholesterol feeding reduces nuclear forms of sterol regulatory element binding proteins in hamster liver

Cholesterol feeding reduces the mRNAs encoding multiple enzymes in the cholesterol biosynthetic pathway and the low density lipoprotein receptor in livers of hamsters. Here we show that cholesterol feeding also reduces the levels of the nuclear NH2-terminal domains of sterol regulatory element binding proteins (SREBPs), which activate transcription of sterol-regulated genes. We show that livers of hamsters, like those of mice and humans, predominantly produce SREBP-2 and the 1c isoform of SREBP-1. Both are produced as membrane-bound precursors that must be proteolyzed to release the transcriptionally active NH2-terminal domains. Diets containing 0.1% to 1.0% cholesterol decreased the amount of nuclear SREBP-1c without affecting the amount of the membrane precursor or its mRNA, suggesting that cholesterol inhibits the proteolytic processing of SREBP-1 in liver as it does in cultured cells. Cholesterol also appeared to reduce the proteolytic processing of SREBP-2. In addition, at high levels of dietary cholesterol the mRNA encoding SREBP-2 declined and the amount of the precursor also fell, suggesting that cholesterol accumulation also may inhibit transcription of the SREBP-2 gene. The high-cholesterol diets reduced the amount of low density lipoprotein receptor mRNA by 30% and produced a more profound 70–90% reduction in mRNAs encoding 3-hydroxy-3-methylglutaryl CoA synthase and reductase. Treatment with lovastatin and Colestipol, which increases hepatic demands for cholesterol, increased the amount of SREBP-2 mRNA as well as the precursor and nuclear forms of the protein. This treatment caused a reciprocal decline in SREBP-1c mRNA and protein. Considered together, these data suggest that SREBPs play important roles in controlling transcription of sterol-regulated genes in liver, as they do in cultured cells.

Der3p/Hrd1p Is Required for Endoplasmic Reticulum-associated Degradation of Misfolded Lumenal and Integral Membrane Proteins

We have studied components of the endoplasmic reticulum (ER) proofreading and degradation system in the yeast Saccharomyces cerevisiae. Using a der3–1 mutant defective in the degradation of a mutated lumenal protein, carboxypeptidase yscY (CPY*), a gene was cloned which encodes a 64-kDa protein of the ER membrane. Der3p was found to be identical with Hrd1p, a protein identified to be necessary for degradation of HMG-CoA reductase. Der3p contains five putative transmembrane domains and a long hydrophilic C-terminal tail containing a RING-H2 finger domain which is oriented to the ER lumen. Deletion of DER3 leads to an accumulation of CPY* inside the ER due to a complete block of its degradation. In addition, a DER3 null mutant allele suppresses the temperature-dependent growth phenotype of a mutant carrying the sec61–2 allele. This is accompanied by the stabilization of the Sec61–2 mutant protein. In contrast, overproduction of Der3p is lethal in a sec61–2 strain at the permissive temperature of 25°C. A mutant Der3p lacking 114 amino acids of the lumenal tail including the RING-H2 finger domain is unable to mediate degradation of CPY* and Sec61–2p. We propose that Der3p acts prior to retrograde transport of ER membrane and lumenal proteins to the cytoplasm where they are subject to degradation via the ubiquitin-proteasome system. Interestingly, in ubc6-ubc7 double mutants, CPY* accumulates in the ER, indicating the necessity of an intact cytoplasmic proteolysis machinery for retrograde transport of CPY*. Der3p might serve as a component programming the translocon for retrograde transport of ER proteins, or it might be involved in recognition through its lumenal RING-H2 motif of proteins of the ER that are destined for degradation.

Ste6p Mutants Defective in Exit from the Endoplasmic Reticulum (ER) Reveal Aspects of an ER Quality Control Pathway in Saccharomyces cerevisiae

We are studying the intracellular trafficking of the multispanning membrane protein Ste6p, the a-factor transporter in Saccharomyces cerevisiae and a member of the ATP-binding cassette superfamily of proteins. In the present study, we have used Ste6p as model for studying the process of endoplasmic reticulum (ER) quality control, about which relatively little is known in yeast. We have identified three mutant forms of Ste6p that are aberrantly ER retained, as determined by immunofluorescence and subcellular fractionation. By pulse-chase metabolic labeling, we demonstrate that these mutants define two distinct classes. The single member of Class I, Ste6–166p, is highly unstable. We show that its degradation involves the ubiquitin–proteasome system, as indicated by its in vivo stabilization in certain ubiquitin–proteasome mutants or when cells are treated with the proteasome inhibitor drug MG132. The two Class II mutant proteins, Ste6–13p and Ste6–90p, are hyperstable relative to wild-type Ste6p and accumulate in the ER membrane. This represents the first report of a single protein in yeast for which distinct mutant forms can be channeled to different outcomes by the ER quality control system. We propose that these two classes of ER-retained Ste6p mutants may define distinct checkpoint steps in a linear pathway of ER quality control in yeast. In addition, a screen for high-copy suppressors of the mating defect of one of the ER-retained ste6 mutants has identified a proteasome subunit, Hrd2p/p97, previously implicated in the regulated degradation of wild-type hydroxymethylglutaryl-CoA reductase in the ER membrane.

The Role of the 3-Hydroxy 3-Methylglutaryl Coenzyme A Reductase Cytosolic Domain in Karmellae Biogenesis

In all cells examined, specific endoplasmic reticulum (ER) membrane arrays are induced in response to increased levels of the ER membrane protein 3-hydroxy 3-methylglutaryl coenzyme A (HMG-CoA) reductase. In yeast, expression of Hmg1p, one of two yeast HMG-CoA reductase isozymes, induces assembly of nuclear-associated ER stacks called karmellae. Understanding the features of HMG-CoA reductase that signal karmellae biogenesis would provide useful insights into the regulation of membrane biogenesis. The HMG-CoA reductase protein consists of two domains, a multitopic membrane domain and a cytosolic catalytic domain. Previous studies had indicated that the HMG-CoA reductase membrane domain was exclusively responsible for generation of ER membrane proliferations. Surprisingly, we discovered that this conclusion was incorrect: sequences at the carboxyl terminus of HMG-CoA reductase can profoundly affect karmellae biogenesis. Specifically, truncations of Hmg1p that removed or shortened the carboxyl terminus were unable to induce karmellae assembly. This result indicated that the membrane domain of Hmg1p was not sufficient to signal for karmellae assembly. Using β-galactosidase fusions, we demonstrated that the carboxyl terminus was unlikely to simply serve as an oligomerization domain. Our working hypothesis is that a truncated or misfolded cytosolic domain prevents proper signaling for karmellae by interfering with the required tertiary structure of the membrane domain.

Three-dimensional structure of human electron transfer flavoprotein to 2.1-Å resolution

Mammalian electron transfer flavoproteins (ETF) are heterodimers containing a single equivalent of flavin adenine dinucleotide (FAD). They function as electron shuttles between primary flavoprotein dehydrogenases involved in mitochondrial fatty acid and amino acid catabolism and the membrane-bound electron transfer flavoprotein ubiquinone oxidoreductase. The structure of human ETF solved to 2.1-Å resolution reveals that the ETF molecule is comprised of three distinct domains: two domains are contributed by the α subunit and the third domain is made up entirely by the β subunit. The N-terminal portion of the α subunit and the majority of the β subunit have identical polypeptide folds, in the absence of any sequence homology. FAD lies in a cleft between the two subunits, with most of the FAD molecule residing in the C-terminal portion of the α subunit. Alignment of all the known sequences for the ETF α subunits together with the putative FixB gene product shows that the residues directly involved in FAD binding are conserved. A hydrogen bond is formed between the N5 of the FAD isoalloxazine ring and the hydroxyl side chain of αT266, suggesting why the pathogenic mutation, αT266M, affects ETF activity in patients with glutaric acidemia type II. Hydrogen bonds between the 4′-hydroxyl of the ribityl chain of FAD and N1 of the isoalloxazine ring, and between αH286 and the C2-carbonyl oxygen of the isoalloxazine ring, may play a role in the stabilization of the anionic semiquinone. With the known structure of medium chain acyl-CoA dehydrogenase, we hypothesize a possible structure for docking the two proteins.

Cloning and expression of the multifunctional human fatty acid synthase and its subdomains in Escherichia coli

We engineered a full-length (8.3-kbp) cDNA coding for fatty acid synthase (FAS; EC 2.3.1.85) from the human brain FAS cDNA clones we characterized previously. In the process of accomplishing this task, we developed a novel PCR procedure, recombinant PCR, which is very useful in joining two overlapping DNA fragments that do not have a common or unique restriction site. The full-length cDNA was cloned in pMAL-c2 for heterologous expression in Escherichia coli as a maltose-binding protein fusion. The recombinant protein was purified by using amylose-resin affinity and hydroxylapatite chromatography. As expected from the coding capacity of the cDNA expressed, the chimeric recombinant protein has a molecular weight of 310,000 and reacts with antibodies against both human FAS and maltose-binding protein. The maltose-binding protein-human FAS (MBP-hFAS) catalyzed palmitate synthesis from acetyl-CoA, malonyl-CoA, and NADPH and exhibited all of the partial activities of FAS at levels comparable with those of the native human enzyme purified from HepG2 cells. Like the native HepG2 FAS, the products of MBP-hFAS are mainly palmitic acid (>90%) and minimal amounts of stearic and arachidic acids. Similarly, a human FAS cDNA encoding domain I (β-ketoacyl synthase, acetyl-CoA and malonyl-CoA transacylases, and β-hydroxyacyl dehydratase) was cloned and expressed in E. coli using pMAL-c2. The expressed fusion protein, MBP-hFAS domain I, was purified to apparent homogeneity (Mr 190,000) and exhibited the activities of the acetyl/malonyl transacylases and the β-hydroxyacyl dehydratase. In addition, a human FAS cDNA encoding domains II and III (enoyl and β-ketoacyl reductases, acyl carrier protein, and thioesterase) was cloned in pET-32b(+) and expressed in E. coli as a fusion protein with thioredoxin and six in-frame histidine residues. The recombinant fusion protein, thioredoxin-human FAS domains II and III, that was purified from E. coli had a molecular weight of 159,000 and exhibited the activities of the enoyl and β-ketoacyl reductases and the thioesterase. Both the MBP and the thioredoxin-His-tags do not appear to interfere with the catalytic activity of human FAS or its partial activities.

UV light selectively coinduces supply pathways from primary metabolism and flavonoid secondary product formation in parsley

The UV light-induced synthesis of UV-protective flavonoids diverts substantial amounts of substrates from primary metabolism into secondary product formation and thus causes major perturbations of the cellular homeostasis. Results from this study show that the mRNAs encoding representative enzymes from various supply pathways are coinduced in UV-irradiated parsley cells (Petroselinum crispum) with two mRNAs of flavonoid glycoside biosynthesis, encoding phenylalanine ammonia-lyase and chalcone synthase. Strong induction was observed for mRNAs encoding glucose 6-phosphate dehydrogenase (carbohydrate metabolism, providing substrates for the shikimate pathway), 3-deoxyarabinoheptulosonate 7-phosphate synthase (shikimate pathway, yielding phenylalanine), and acyl-CoA oxidase (fatty acid degradation, yielding acetyl-CoA), and moderate induction for an mRNA encoding S-adenosyl-homocysteine hydrolase (activated methyl cycle, yielding S-adenosyl-methionine for B-ring methylation). Ten arbitrarily selected mRNAs representing various unrelated metabolic activities remained unaffected. Comparative analysis of acyl-CoA oxidase and chalcone synthase with respect to mRNA expression modes and gene promoter structure and function revealed close similarities. These results indicate a fine-tuned regulatory network integrating those functionally related pathways of primary and secondary metabolism that are specifically required for protective adaptation to UV irradiation. Although the response of parsley cells to UV light is considerably broader than previously assumed, it contrasts greatly with the extensive metabolic reprogramming observed previously in elicitor-treated or fungus-infected cells.

Expression of sterol regulatory element-binding protein 1c (SREBP-1c) mRNA in rat hepatoma cells requires endogenous LXR ligands

The current paper describes a line of cultured rat hepatoma cells (McA-RH7777 cells) that mimics the behavior of rat liver by producing an excess of mRNA for sterol regulatory element-binding protein 1c (SREBP-1c) as opposed to SREBP-1a. These two transcripts are derived from a single gene by use of alternative promoters that are separated by many kilobases in the genome. The high level of SREBP-1c mRNA is abolished when cholesterol synthesis is blocked by compactin, an inhibitor of 3-hydroxy-3-methylglutaryl CoA (HMG CoA) reductase that inhibits cholesterol synthesis. Levels of SREBP-1c mRNA are restored by mevalonate, the product of the HMG CoA reductase reaction, and by ligands for the nuclear hormone receptor LXR, including 22(R)-hydroxycholesterol and T0901317. These data suggest that transcription of the SREBP-1c gene in hepatocytes requires tonic activation of LXR by an oxysterol intermediate in the cholesterol biosynthetic pathway. Reduction of this intermediate lowers SREBP-1c levels, and this in turn is predicted to lower the rates of fatty acid biosynthesis in liver.

Molecular cloning of a cytochrome P450 taxane 10β-hydroxylase cDNA from Taxus and functional expression in yeast

The early steps in the biosynthesis of Taxol involve the cyclization of geranylgeranyl diphosphate to taxa-4(5),11(12)-diene followed by cytochrome P450-mediated hydroxylation at C5, acetylation of this intermediate, and a second cytochrome P450-dependent hydroxylation at C10 to yield taxadien-5α-acetoxy-10β-ol. Subsequent steps of the pathway involve additional cytochrome P450 catalyzed oxygenations and CoA-dependent acylations. The limited feasibility of reverse genetic cloning of cytochrome P450 oxygenases led to the use of Taxus cell cultures induced for Taxol production and the development of an approach based on differential display of mRNA-reverse transcription-PCR, which ultimately provided full-length forms of 13 unique but closely related cytochrome P450 sequences. Functional expression of these enzymes in yeast was monitored by in situ spectrophotometry coupled to in vivo screening of oxygenase activity by feeding taxoid substrates. This strategy yielded a family of taxoid-metabolizing enzymes and revealed the taxane 10β-hydroxylase as a 1494-bp cDNA that encodes a 498-residue cytochrome P450 capable of transforming taxadienyl acetate to the 10β-hydroxy derivative; the identity of this latter pathway intermediate was confirmed by chromatographic and spectrometric means. The 10β-hydroxylase represents the initial cytochrome P450 gene of Taxol biosynthesis to be isolated by an approach that should provide access to the remaining oxygenases of the pathway.

Choline acetyltransferase mutations cause myasthenic syndrome associated with episodic apnea in humans

Choline acetyltransferase (ChAT; EC 2.3.1.6) catalyzes the reversible synthesis of acetylcholine (ACh) from acetyl CoA and choline at cholinergic synapses. Mutations in genes encoding ChAT affecting motility exist in Caenorhabditis elegans and Drosophila, but no CHAT mutations have been observed in humans to date. Here we report that mutations in CHAT cause a congenital myasthenic syndrome associated with frequently fatal episodes of apnea (CMS-EA). Studies of the neuromuscular junction in this disease show a stimulation-dependent decrease of the amplitude of the miniature endplate potential and no deficiency of the ACh receptor. These findings point to a defect in ACh resynthesis or vesicular filling and to CHAT as one of the candidate genes. Direct sequencing of CHAT reveals 10 recessive mutations in five patients with CMS-EA. One mutation (523insCC) is a frameshifting null mutation. Three mutations (I305T, R420C, and E441K) markedly reduce ChAT expression in COS cells. Kinetic studies of nine bacterially expressed ChAT mutants demonstrate that one mutant (E441K) lacks catalytic activity, and eight mutants (L210P, P211A, I305T, R420C, R482G, S498L, V506L, and R560H) have significantly impaired catalytic efficiencies.

Study of the 3-Hydroxy Eicosanoyl-Coenzyme A Dehydratase and (E)-2,3 Enoyl-Coenzyme A Reductase Involved in Acyl-Coenzyme A Elongation in Etiolated Leek Seedlings1

(R,S)-[1-14C]3-Hydroxy eicosanoyl-coenzyme A (CoA) has been chemically synthesized to study the 3-hydroxy acyl-CoA dehydratase involved in the acyl-CoA elongase of etiolated leek (Allium porrum L.) seedling microsomes. 3-Hydroxy eicosanoyl-CoA (3-OH C20:0-CoA) dehydration led to the formation of (E)-2,3 eicosanoyl-CoA, which has been characterized. Our kinetic studies have determined the optimal conditions of the dehydration and also resolved the stereospecificity requirement of the dehydratase for (R)-3-OH C20:0-CoA. Isotopic dilution experiments showed that 3-hydroxy acyl-CoA dehydratase had a marked preference for (R)-3-OH C20:0-CoA. Moreover, the very-long-chain synthesis using (R)-3-OH C20:0-CoA isomer and [2-14C]malonyl-CoA was higher than that using the (S) isomer, whatever the malonyl-CoA and the 3-OH C20:0-CoA concentrations. We have also used [1-14C]3-OH C20:0-CoA to investigate the reductant requirement of the enoyl-CoA reductase of the acyl-CoA elongase complex. In the presence of NADPH, [1-14C]3-OH C20:0-CoA conversion was stimulated. Aside from the product of dehydration, i.e. (E)-2,3 eicosanoyl-CoA, we detected eicosanoyl-CoA resulting from the reduction of (E)-2,3 eicosanoyl-CoA. When we replaced NADPH with NADH, the eicosanoyl-CoA was 8- to 10-fold less abundant. Finally, in the presence of malonyl-CoA and NADPH or NADH, [1-14C]3-OH C20:0-CoA led to the synthesis of very-long-chain fatty acids. This synthesis was measured using [1-14C]3-OH C20:0-CoA and malonyl-CoA or (E)-2,3 eicosanoyl-CoA and [2-14C]malonyl-CoA. In both conditions and in the presence of NADPH, the acyl-CoA elongation activity was about 60 nmol mg−1 h−1, which is the highest ever reported for a plant system.

A Multisubunit Acetyl Coenzyme A Carboxylase from Soybean1

A multisubunit form of acetyl coenzyme A (CoA) carboxylase (ACCase) from soybean (Glycine max) was characterized. The enzyme catalyzes the formation of malonyl CoA from acetyl CoA, a rate-limiting step in fatty acid biosynthesis. The four known components that constitute plastid ACCase are biotin carboxylase (BC), biotin carboxyl carrier protein (BCCP), and the α- and β-subunits of carboxyltransferase (α- and β-CT). At least three different cDNAs were isolated from germinating soybean seeds that encode BC, two that encode BCCP, and four that encode α-CT. Whereas BC, BCCP, and α-CT are products of nuclear genes, the DNA that encodes soybean β-CT is located in chloroplasts. Translation products from cDNAs for BC, BCCP, and α-CT were imported into isolated pea (Pisum sativum) chloroplasts and became integrated into ACCase. Edman microsequence analysis of the subunits after import permitted the identification of the amino-terminal sequence of the mature protein after removal of the transit sequences. Antibodies specific for each of the chloroplast ACCase subunits were generated against products from the cDNAs expressed in bacteria. The antibodies permitted components of ACCase to be followed during fractionation of the chloroplast stroma. Even in the presence of 0.5 m KCl, a complex that contained BC plus BCCP emerged from Sephacryl 400 with an apparent molecular mass greater than about 800 kD. A second complex, which contained α- and β-CT, was also recovered from the column, and it had an apparent molecular mass of greater than about 600 kD. By mixing the two complexes together at appropriate ratios, ACCase enzymatic activity was restored. Even higher ACCase activities were recovered by mixing complexes from pea and soybean. The results demonstrate that the active form of ACCase can be reassembled and that it could form a high-molecular-mass complex.