A Novel Nuclear Member of the Thioredoxin Superfamily

Abstract

We describe the isolation and characterization of a cDNA encoding maize (Zea mays L.) nucleoredoxin (NRX), a novel nuclear protein that is a member of the thioredoxin (TRX) superfamily. NRX is composed of three TRX-like modules arranged as direct repeats of the classic TRX domain. The first and third modules contain the amino acid sequence WCPPC, which indicates the potential for TRX oxidoreductase activity, and insulin reduction assays indicate that at least the third module possesses TRX enzymatic activity. The carboxy terminus of NRX is a non-TRX module that possesses C residues in the proper sequence context to form a Zn finger. Immunolocalization preferentially to the nucleus within developing maize kernels suggests a potential for directed alteration of the reduction state of transcription factors as part of the events and pathways that regulate gene transcription.

The concept of redox regulation is becoming increasingly important at diverse levels of cellular function. Under oxidative stress, cells exposed to reactive oxygen species without appropriate buffering capacity could die. Moderation of the redox state is an important function in intracellular signaling (Nakamura et al., 1997). Furthermore, the binding of several transcription factors, such as NFB 66 and AP-1, has been shown to be markedly influenced by their respective redox states (Schenk et al., 1994; Hirota et al., 1997). TRXs are typically small, approximately 12-kD proteins that participate in cellular redox reactions. They are widely distributed, being present in some form in all organisms, from bacteria and yeast to animals and plants. Plants have multiple family members, including plant-specific forms that are found in the chloroplast (the TRX-f and TRX-m forms), as well as TRX-h, which is more animal-like in sequence and is found in the cytosol, ER, and mitochondria, as determined by cellular-fractionation studies (Marcus et al., 1991; Buchanan et al., 1994). Although multiple family members were thought to be unique to plants, recent reports document larger TRXs such as an 18-kD mammalian TRX and a novel 15.5-kD Escherichia coli TRX (Miranda-Vizuete et al., 1997; Spyrou et al., 1997).

The redox activities of TRXs are due to an active site that contains vicinal C residues that can exist in a reduced or mutually oxidized, disulfide bridge state. An active-site motif of WCGPC is conserved among the classic TRXs of E. coli and many other species. However, other active-site sequences exist, including a WCPPC motif that has been identified in TRX and TRX-like sequences within diverse organisms such as Arabidopsis, Caenorhabditis elegans, and Mus musculus (Rivera-Madrid et al., 1995; Kurooka et al., 1997). The classic TRX has its active site protruding from a highly organized, globular structure composed of five strands of β-sheets enclosed by four α-helices (Holmgren, 1985). Specific conserved residues suggest a conserved tertiary structure, even though the amino acid sequence among TRXs in general can vary from 26% to 67% sequence identity compared with E. coli TRX (Eklund et al., 1991).

TRX-like domains have been identified in many eukaryotic proteins through sequence similarity. A gene from a self-incompatibility locus in the grass Phalaris coerulescens has a functional TRX domain (Li et al., 1995). PDIs contain multiple TRX domains and are responsible for multiple disulfide rearrangements in protein folding, an activity not overtly present in single-module TRXs (Freedman et al., 1994). Human PDI, for example, has two TRX modules that can be identified by sequence homology. A third module of PDI lacks vicinal C residues but retains a TRX fold, as determined by NMR analysis (Kemmink et al., 1997). PDIs usually have the active site of WCGHC and typically exhibit ER-targeting and -retention signals. A mouse TRX-domain-containing protein found in the nucleus has a TRX domain with the WCPPC active-site motif (Kurooka et al., 1997) and vestiges of a second, inactive TRX-like domain. Thus, it appears that a major diversifying mechanism among TRX-like molecules is to link active and inactive TRX domains together, perhaps thereby increasing functional diversity or specificity.

Classic functional roles associated with TRXs include thiol redox regulation of enzymes. Chloroplast TRX members utilize Fd and Fd reductase in light-mediated reactions that target enzymes of the reductive pentose phosphate or Calvin cycle (Buchanan et al., 1994). Animal TRX, E. coli TRX, and plant TRX-h utilize NADPH with TRX reductase in redox regulation. Specific roles for TRX-h in plants include the reduction of purothionin, which is a small, disulfide-rich protein in wheat grains, and early signaling in seed germination (Buchanan et al., 1994). Recent reports continue to elucidate the role of TRX in the regulation of transcription factors through changes in the state of critical C residues located within their DNA-binding domains. These changes may be effected directly as in the case of NFκB or indirectly in AP-1 through Ref-1 (Hayashi et al., 1993; Hirota et al., 1997). Although TRXs lack a recognized nuclear-localization signal, they can be translocated from the cytoplasm to the nucleus in response to stress (Hirota et al., 1997; Nakamura et al., 1997).

We report here the isolation from maize of a cDNA encoding a novel, multiple TRX-domain protein that we call NRX because it is highly localized to the nucleus, is composed of multiple TRX-domain modules, and in its recombinant form possesses TRX enzymatic activity.

MATERIALS AND METHODS

Screening for Factors That Influence DNA Binding

A partially purified preparation of the ARF (Anaerobic Response Factor)-B2 DNA-binding activity (Ferl, 1990) was used for the production of a panel of hybridomas in anticipation of recovering antibodies that influence DNA-binding activities. Several hundred hybridoma cell supernatants were screened by assaying the effect of the supernatant on bandshifts involving ARF-B2 activity. Two supernatants that reduced DNA-binding activity were subsequently used to screen a λ-gt11 maize (Zea mays L.) suspension-cell library (Clontech, Palo Alto, CA) as previously described (Lu et al., 1992). One positive clone was purified to homogeneity and its phage eluate was subjected to a PCR reaction with λ-gt11 primers to recover the insert. A product approximately 860 bp in length was restricted with EcoRI and cloned into pCR2 18 to produce plasmid Z863 (Fig. 1).

Figure 1.

cDNA clones used for the contig assembly of maize NRX. A, Contig assembled by merging clones. The first clone, Z863, was identified by immunoscreening a λ-gt11 custom-made maize suspension-cell library. This fragment was cloned into pUC 18, and the insert was used to screen for longer clones. The clone C1300 in the vector pUC 18 provided the carboxyl end of the cDNA. The ZmD1 clone was cloned into pCR2 and represents the longest cDNA. The clone referred to as D5 was obtained through a modified 5′-rapid amplification of cDNA ends technique and was cloned into pCRBLUNT. The clone designated Z10 was generated through the use of engineered NdeI and BamHI sites. The NdeI site coincides with the first inframe ATG codon, and the BamHI site lies just downstream of the stop codon.

cDNA Clones

Since northern analysis suggested that Z863 was a partial cDNA clone, the EcoRI insert from Z863 was random-prime labeled (Boehringer Mannheim) as a probe for further screening. Several λ-gt11 clones were purified to homogeneity and their respective phage eluates were used to generate PCR products for cloning into pCR2 and pCRBLUNT (Invitrogen, Carlsbad, CA). This screen produced cDNA plasmids ZmD1 and C1300 (Fig. 1).

A modification of a 5′-rapid amplification of cDNA ends protocol (Jain et al., 1992) was used to clone the 5′ end of the cDNA clone. The following antisense nested primers were used: RF 271 5′ CCGAAGCAAAGACGACCT 3′ (nucleotide positions 362–345) and RF 272 5′ GAAAAGGGGACAGCCAAC 3′ (nucleotide positions 429–412). The first strand was prepared from total RNA isolated from the maize suspension culture. PCR amplification with the 3′ most of the nested primers (RF 272) was performed with 1 cycle at 95° for 80 s, 45° for 5 min, and 72° for 2 min followed by 3 cycles at 95° for 40 s, 48° for 1 min, and 72° for 2 min. This first round was filtered through an Ultrafree MC-regenerated cellulose membrane with a 30,000 nominal M_r limit (Millipore) cutoff to remove primers and dNTPs. The PCR amplification with the 5′-most primer (RF 271) was performed with the following specifications: 95° for 2 min and 30 cycles at 95° for 1 min, 58° for 2 min, and 72° for 2 min. It is noteworthy that only Vent polymerase (New England Biolabs) amplified a product that was subcloned into pCR2 to yield plasmid D5 (Fig. 1).

A composite cDNA clone Z10 was generated by merging the products of the 5′ extension protocol (D5) and the longest clone (ZmD1) by a modification of strand-overlap extension (Kim et al., 1996). First, D5 was used as the template to generate the amino end product of the clone. The 5′ sense oligo (primer 1) included an engineered NdeI and the sequence from nucleotide positions 119 to 136 of NRX, whereas the antisense oligo (primer 2) annealed near the 3′ end of D5, at nucleotide positions 344 to 325. Second, the template ZmD1 was used with a sense oligo (primer 3) at nucleotide positions 241 to 269 and an antisense oligo (primer 4) just downstream of the stop codon near nucleotide positions 1876 to 1858 and included an engineered BamHI site to generate the carboxy end of the clone. The overlap area was approximately 100 bp and facilitated amplification of a composite clone of 1757 bp when both products were subjected to a second round of PCR using the 5′-most sense oligo (primer 1) and the 3′-most antisense oligo (primer 4). The composite PCR product was immediately cloned into pCRBLUNT. This clone, representing a putative full-length clone for expression studies, was designated Z10 (Fig. 1).

Sequence Analysis

Automated dideoxy chain termination on both strands of these plasmids using Taq polymerase was performed on a DNA sequencer (model ABI 373, Applied Biosystems). Computer analyses on DNA sequences were performed with MacVector utilities (IBI, Eastman Kodak) and Geneworks (IntelliGenetics, Oxford Molecular Group, Oxford, UK). The final composite sequence was generated by the merging sequences from D5, the longest clone, ZmD1, and a shorter clone, C1300, that contained a poly(A⁺) tail (Fig. 1). Homology searches were done using BLAST software (Altschul et al., 1997).

Expression Studies in E. coli

For expression of the carboxy-terminal portion of NRX (ΔNRX), the EcoRI fragment of Z863 was subcloned into pETH 3A (McCarty et al., 1991) and, after sequence confirmation, the construct was transformed into BL21DE3 for protein expression. This construct was referred to as Z10–14 and includes the carboxy-terminal portion of the NRX protein underlined in Figure 2. The ΔNRX coding sequence starts 22 amino acids from the NdeI site in the pETH vector, so the resulting NRX is a fusion protein that includes the following deduced amino acids from the vector as an amino-terminal leader: MASMTGGQQMGRSSFPGSSNSG. This fusion construct was later moved as a NdeI-ClaI fragment to pET 15b.

Figure 2.

DNA and amino acid sequence of maize NRX. The merged cDNA clones resulted in a composite clone 2005 bp in length. The deduced amino acids are shown directly under the sequence. The amino acid sequence underlined represents the specific area of the maize NRX to which rabbit polyclonal antibodies were raised. A vertical bar between amino acids 486 and 487 indicates where the carboxy-terminal extension begins. Preliminary sequence information from genomic PCR fragments suggests the location of two introns, indicated by the darkened triangles. A potential polyadenylation is double-underlined.

For expression of the full-length NRX protein, the coding region of Z10 was cloned into pET 15b using engineered NdeI and BamHI sites, as described above.

The full-length and the truncated expression plasmids were established in BL21DE3, and expression of NRX was induced with isopropylthio-β-galactoside at a final concentration of 1 mm. After induction, the bacteria were subjected to several freeze/thaw cycles, treated with DNase, and centrifuged to recover the soluble protein fraction essentially according to standard protocols (Sambrook et al., 1989). Both of these constructs resulted in target-protein expression in the soluble and insoluble fractions. The soluble fraction was subjected to affinity purification on an immobilized Ni column. Inherent thrombin cleavage sites in the NRX sequence precluded thrombin cleavage to release the His tag. Consequently, the proteins purified from pET 15b retaining the His leader correlated with their predicted sizes of 30 and 65 kD for ΔNRX and NRX, respectively (data not shown).

Polyclonal Antibody Production and Clarification

The original ΔNRX supernatant fraction obtained from the construct Z10–14 was subjected to Mono-Q chromatography using a linear gradient from 0 to 1.0 m NaCl in 50 mm Tris-HCl at pH 8.0. Fractions containing the 30-kD ΔNRX protein were identified by electrophoresis and staining with Coomassie blue, and were then pooled and subjected to a second round of Mono-Q chromatography. Fractions containing ΔNRX were pooled and subjected to size chromatography on Sephadex-75 in 50 mm Tris HCl at pH 8.0. Once again, fractions containing ΔNRX were pooled and their purity was estimated at greater than 95% by Coomassie blue staining. These ΔNRX fractions were used for the production of polyclonal rabbit antibodies (Bioworld, Dublin, OH).

Clarified polyclonal antibodies were prepared by incubation of the rabbit serum with acetone powder of whole-cell proteins isolated from an induced BL21DE3 line transformed with pET 3Xa, which contains the T7 capsid protein (Harlow and Lane, 1988). The acetone powder preparation was used at 1% for 30 min at 4°C, spun at 10,000g for 10 min, and then the supernatant was saved as the clarified antibody stock.

Western Analysis

Two grams of frozen tissue of 1-week-old maize seedlings or 10-DAP kernels were pulverized in a mortar and pestle with liquid nitrogen, then resuspended in TBS, pH 7.6, with 1 mm EDTA and a protease inhibitor cocktail tablet (1 873 580, Boehringer Mannheim). After spinning at 14,000 rpm for 10 min, portions of the extract supernatant were analyzed on a standard 10% acrylamide SDS denaturing gel (Sambrook et al., 1989). The gel was subsequently electroblotted to a nitrocellulose membrane. After blocking in TBS with 7% dry milk powder the membrane was incubated with the clarified anti-ΔNRX polyclonal antibodies at a dilution of 1:30,000. The secondary antibody was used at a dilution of 1:7,000. After washing the membrane, the SuperSignal substrate system (Pierce) was used.

Test for TRX Activity

Both the full-length NRX protein and ΔNRX were assayed for TRX enzymatic activity using the insulin-disulfide reduction assay (Holmgren, 1979). The 100-μL reaction volume contained 84 mm sodium phosphate, pH 7.0, 2 mm EDTA, and 0.17 mm insulin (I 5523, Sigma). Spirulina TRX (T 3658, Sigma) served as a positive control and BSA as a negative control. The assay was initiated by the addition of 1 mm DTT. Measurements were performed at 650 nm at 5-min intervals on a spectrophotometer (model DU7400, Beckman). The cuvettes were standardized with water blanks prior to monitoring the insulin disulfide reductions.

Northern Analysis

For scrutiny of organ-specific expression of NRX mRNA, total RNA was extracted from maize suspension-cultured cells, immature kernels, and leaves, roots, and stems from 1-week-old seedlings using Trizol (Life Technologies). Approximately 7 μg of total RNA was size-fractionated in a 1.2% Mops/formaldehyde gel and transferred to Hybond-N⁺ membrane (Amersham) via capillary transfer in 20× SSC. The membrane was rinsed briefly in water, dried at 80° for 10 min, and UV cross-linked for 2.5 min. The membrane was hybridized to a random-prime-labeled probe prepared from an EcoRI insert from ZmD1.

Southern Analysis

Approximately 10 μg of genomic DNA isolated from maize suspension-cultured cells was digested with representative restriction enzymes identified in the cDNAs isolated in this study. The samples were subsequently resolved on a 1.5% agarose gel in 1× TBE and transferred to Hybond-N⁺ membrane via capillary transfer in 10× SSC. The membranes were rinsed briefly in 2× SSC, dried at 80°C for 10 min, and UV-cross-linked for 2.5 min. The same probe prepared for the northern analysis was also used in the Southern analysis.

Immunolocalization

Immature kernels (13 DAP) were embedded in paraffin medium (Paraplast Plus, Sigma) as previously described (Cheng et al., 1996). The lengthwise sections were cut at a 12-μm thickness with a rotary microtome and mounted onto slides (Probe On Plus, Fisher Scientific) overnight on a slide warmer set at 42°C. The paraffin medium was removed from the sections with xylene, and the sections were rehydrated with a graded-ethanol dilution series of 95%, 75%, 50%, and 25%. After a brief rinse in water, the sections were incubated in PBS prior to initiating immunogold staining as specified for the Histogold kit (Zymed, San Francisco, CA). The primary antibody was used at a dilution of 1:7000. A preimmune serum was included as a negative control. Application of anti-rabbit secondary IgG antibodies conjugated to gold particles was followed by silver enhancement to discern the specificity of this reaction. Initially, the primary antibody incubation was performed overnight, but a 4-h incubation interval proved satisfactory.

RESULTS

Identification of Maize NRX

The original NRX cDNA clone was recovered from the screening of an expression library with monoclonal antibodies raised against a partially purified DNA-binding activity. Rescreening the library by hybridization and using a modified 5′-rapid amplification of cDNA ends protocol produced a set of overlapping cDNA clones that spanned the entire coding region of NRX (Fig. 1).

The nucleotide sequence of the maize NRX cDNA composite is presented in Figure 2. The composite sequence is 2005 nucleotides in length, with a single open reading frame that begins with the first ATG codon at position 119 and ends with a stop codon at position 1825. There are 118 nucleotides of 5′ untranslated leader, and 180 nucleotides of 3′-noncoding sequence that contains a presumptive poly(A⁺) addition sequence and evidence of a poly(A⁺) tail. The integrity of this reconstructed contig was confirmed by preliminary genomic sequence data and the existence of a recent homologous Arabidopsis sequence (accession no. AC004473).

The open reading frame encodes a protein of 569 amino acids, whose deduced sequence is also presented in Figure 2. Density-plot-matrix analysis of the composite cDNA sequence compared against itself indicates that the bulk of the sequence is composed of a repetitive structure; 3 repeats corresponding to 162 amino acids are clearly indicated by the parallel diagonal lines accompanying the central identity diagonal of Figure 3. After the 3 repeated modules, there is a unique carboxy-terminal region corresponding to approximately 80 amino acids preceding the 3′ untranslated region and the poly(A⁺) tail.

Figure 3.

Protein dot matrix reveals the multiple repeat nature of NRX. A dot matrix from Genworks (Intelligenetics, Oxford Molecular Group, Oxford, UK) graphically identifies the regions of similarity repeated within the NRX sequence. The NRX amino acid sequence was plotted against itself. Note that the main diagonal was not eliminated and that the repetitive nature of the maize NRX is shown as parallel dots at intervals of 162 amino acids.    

A BLAST search (Altschul et al., 1997) of nonredundant proteins revealed homology of the repeated modules with a novel mouse NRX and roundworm TRX entries from genomic sequence data. Figure 4 is an alignment of the three maize NRX TRX-like tandem repeats designated a, a*, and a′, with E. coli TRX and a plant TRX-h for reference. The E. coli TRX active site WCGPC has been modified to a WCPPC motif in each of the aligned sequences, except for the second TRX-like module of NRX, where the central PP dipeptide sequence has been maintained but the vicinal C residues are lost. The alignment also indicates that all of the TRX-like modules from maize NRX, the mouse module, and the roundworm sequence possess an insertion of approximately 25 amino acids relative to the classic 12-kD TRX (Eklund et al., 1991). This insertion sequence is relatively conserved, with the sequences from mouse and maize maintaining as high a degree of similarity within the inserted sequence as is noted for the modules as a whole (Fig. 4).

Figure 4.

An alignment comparing the TRX-like modules of maize NRX to its closest homologs and prototype TRXs. The repetitive maize TRX-like modules within NRX are identified as a, a*, and a′, respectively. The mouse NRX sequence includes only amino acids 151 to 322 for clarity. Introducing gaps to maximize alignments readily identifies the large expansion within mouse NRX, the C. elegans entry, and maize NRX. The assigned accession numbers are as follows: E. coli TRX, M54881; Arabidopsis TRX-h, Z35474; C. elegans, Z48795; mouse NRX, X92750; and maize NRX, U90944. Identical residues that occur in at least four entries are in bold type and are underlined. Secondary structures from the TRX-fold substructure are depicted above the TRX sequence as per the method of Martin (1995).

There is a unique extension at the carboxyl end of the third TRX-like module of maize NRX (Fig. 2). This extension has clusters of basic residues potentially suggestive of a bipartite nuclear-localization signal, and C residues that potentially form a Zn finger.

Tissue-Specific Distribution of NRX

The clarified antibodies produced against maize NRX were used in a western analysis of the tissue and organ-specific distribution of NRX. The predominant cross-reacting proteins appeared to be approximately 69 kD in size (Fig. 5), which roughly correlates with the size of 65.5 kD for the rNRX predicted from the cDNA sequence (MacVector). NRX cross-reacting bands were abundant in extracts from kernels and suspension cells, and less-detectable NRX amounts were observed in roots and leaves. The smaller cross-reacting bands in the gel blots likely represent proteolytic cleavage products, since recombinant, bacterially expressed NRX occasionally shows multiple bands as well (data not shown). It remains possible at this point that related TRXs such as TRX-h may cross-react with NRX antibodies. However, given the strict subcellular localization noted below, this explanation for the smaller bands is considered unlikely. The presence of a cross-reacting band larger than the rNRX may merely represent modified NRX molecules, as there are potential glycosylation sites in this clone. Preliminary data suggests that NRX may also be phosphorylated.

Figure 5.

Western analysis. Approximately 5 μg of a crude protein extract was loaded onto a 10% SDS-PAGE gel to resolve the total proteins isolated from: lane 1, 10-DAP kernels; lane 2, suspension-cell culture; lane 3, epicotyl; and lane 4, roots. Lane 5 represents approximately 0.025 μg of full-length recombinant NRX.

NRX Is Localized to the Nucleus

Figure 6 shows the cellular localization of maize NRX within the kernel, using the same clarified polyclonal antibodies used in the western analysis. The antibodies to NRX clearly cross-reacted with antigens present within the nuclei of cells in the scutellum of kernels harvested 13 DAP (Fig. 6B). Preimmune serum used as the primary antibody showed no staining (Fig. 6A).

Figure 6.

NRX is present in nuclei, as detected by immunolocalization. A, Longitudinal section through the scutellum of a 13-DAP kernel challenged with only the preimmune serum. No detection of cross-reactivity is evident. B, Nuclei within the scutellum of a 13-DAP kernel cross-reacting with clarified polyclonal antibodies raised against the carboxyl end of maize NRX.

NRX Copy Number and Gene Expression

Hybridization analysis of multiple restriction digest profiles of maize genomic DNA resulted primarily in single bands indicative of a single-copy gene (Fig. 7). A few faintly hybridizing bands are visible in several lanes, and the SacI digest does show multiple hybridizing bands. However, taken together, these data are consistent with the interpretation that NRX is a single-copy gene, but that there may be a few related sequences in the genome of maize.

Figure 7.

Southern analysis. Approximately 10 μg of genomic DNA isolated from maize suspension-cultured cells was digested with representative restriction enzymes to examine the relative complexity of the NRX gene. Most of the digests resulted in a single prominent band, indicative of a single gene family, but the SacI digest reveals several bands.

Northern analysis demonstrated that a single 2-kb transcript hybridizes with the NRX cDNA insert probe (Fig. 8), indicating that the composite cDNA sequence is likely nearly full-length. The RNA blot also demonstrates that the mRNA for NRX is abundant in kernels and the cell-suspension culture, consistent with the high levels of NRX protein detected by western analysis.

Figure 8.

RNA analysis of NRX gene expression. Approximately 7 μg of total RNA isolated from maize suspension-cultured cells, immature kernels, and from the leaves and roots of 1-week-old maize seedlings was size-fractionated in a 1.2% agarose Mops/formaldehyde gel. The RNA was then transferred to Hybond-N⁺ membrane in 20× SSC. After UV cross-linking, the membrane was probed with the largest EcoRI insert fragment from ZmD1. This probe identified a transcript size of about 2 kb. The transcript appeared to be abundant in cultures as well as kernels. The mRNA was less abundant in leaves and roots.

TRX Activity of NRX

Recombinant full-length NRX and the truncated ΔNRX (containing the intact TRX a′ module and the carboxy-terminal extension) were assayed for TRX enzymatic activity by monitoring the reduction of insulin. Spirulina TRX was included as a positive control, whereas BSA and DTT alone were negative controls (Fig. 9). Both the full-length NRX (at 0.13 μm) and the truncated ΔNRX (at 0.26 μm) demonstrated significant TRX activity above background, and the activity was similar to that of Spirulina TRX (used at 0.50 μm) if adjusted for the concentration of input protein. There was an obvious lag for both forms and a corresponding slower rate of precipitation of NRX, as expected since the concentrations were lower than the control TRX. This phenomenon was noted by Holmgren in his classic assay (Holmgren, 1979).

Figure 9.

. TRX activity shown by disulfide reduction of insulin. Approximately 8 μg of the truncated NRX, the full-length NRX, and a representative prokaryotic TRX (Spirulina) were assayed for TRX activity based on ability to reduce the disulfide bonds of bovine insulin (Holmgren, 1979). TRX accelerated the reaction more dramatically than the two NRX proteins, yet the longer delay and the corresponding slower rate of precipitation is consistent with assays at lower concentrations. Note that the nonenzymatic breakage with only DTT had an extremely long lag phase, but that the same amount of precipitate was observed after leaving the assay cuvettes for an extended period (data not shown).

DISCUSSION

Maize NRX is clearly a unique member of the growing superfamily of TRX proteins. Structural and sequence similarity with the mouse NRX suggests the name classification of NRX, which was first coined by Kurooka et al. (1997). The presence of three TRX-like modules within maize NRX is consistent with members of the TRX superfamily, such as PDIs, that appear to utilize multiple TRX-like modules as a major mechanism for structural and functional diversification. To our knowledge, maize NRX is the first member of the superfamily to be recognized as having three TRX-like modules in tandem, and the first such member of the superfamily to have a putative Zn-finger module at its carboxy terminus.

A model for the structure of maize NRX is presented in Figure 10. The first and third TRX-like domains (a and a′) are most clearly TRX like, sharing 35% identity with each other and possessing the vicinal C active site. The middle domain (a*) lacks the vicinal C active site, but shares 35% identity with a and a′. The a′ domain has demonstrable oxidoreductase enzymatic activity, consistent with the structural prediction of its TRX homology. Although the entire NRX has activity as well, independent activity of the a domain has yet to be tested, and the a* domain is predicted to be inactive with regard to oxidoreductase activity. The carboxy terminus of the protein (z) consists of a putative Zn finger, but NRX has not yet been shown to possess DNA or metal-binding capacity.

Figure 10.

Model for the maize NRX protein. The three TRX-like modules are arranged in tandem and labeled as a, a*, and a′. The last module containing the putative Zn finger is labeled “z.” Limited homology with nuclear factors corresponding to specific amino acid regions within the maize NRX are shown in boxes, and the asterisk in HTF 10 represents additional amino acids that were omitted for clarity. The following accession numbers were used: v-erb-A, P12891; CCAAT-box-binding transcription factor or nuclear factor (CTF/NF)-1B2, P17925; mastermind, M92914; HTF 10, Q05481; v-erb-A Zn finger, I57696; Hunch ZFN, P05064; and ZNF, F14840.

The structural organization of maize NRX addresses several issues of structural motifs within proteins carrying multiple TRX-like modules. The maize NRX probably arose through gene triplication, since all three TRX-like modules (a, a*, and a′) share approximately 35% sequence identity among themselves. Even though the a* module lacks the active site motif, it retains overall sequence similarity such as the double Ps in the residual active site and other conserved residues believed to be important for conferring TRX activity and secondary structure (Eklund et al., 1991). Therefore, the a* domain is still recognizable as a TRX-like module. PDIs, in apparent contrast, have two TRX-like modules (a and a′) that are separated by intervening segments (b and b′) that are not overtly homologous to TRX (Darby et al., 1996). However, recent structural studies indicate that the PDI-b domain retains structural similarity to a TRX fold, suggesting that all four PDI modules arose by gene duplication or shuffling of a common ancestral TRX (Kemmink et al., 1997). Therefore, multiplication and divergence of TRX-like modules appears to be a major mechanism for structural and perhaps functional diversification of the TRX superfamily of proteins.

The TRX fold represents a substructure of TRX, missing the N-terminal β-strand and α-helix. Analyses of protein classes exhibiting TRX folds notes that insertions typically are found between the β-2 strand and α-2 helix (Martin, 1995). Using the E. coli TRX sequence as a reference for secondary structures, the largest insertions in the maize NRX-alignment figure (Fig. 4) occur between the secondary structures β-2 and α-2 within the TRX-fold substructure. Sequence similarity and oxidoreductase activity for mouse and maize NRX is suggestive of such TRX folds. In addition, TRX-fold-containing proteins such as glutathione peroxidase and DsbA (the bacterial equivalent of PDI) have even larger insertions in this region, 40 and 76 amino acids, respectively (Martin, 1995), compared with the 23 to 24 amino acids found in maize NRX.

Intron capture has been invoked to explain the expanded TRX module in the roundworm TRX-like entry (Sahrawy et al., 1996). Although there is a intron at the end of this substantial insertion in the roundworm sequence entry (Z48795), preliminary genomic sequence information from the maize NRX does not show an intron here. Furthermore, scrutiny of the mouse and maize modules suggests conservation of the amino acids within the region of this insertion (see Fig. 4). This sequence conservation among vastly divergent eukaryotic species suggests that intron capture is an unlikely explanation for the presence of these insertions, as it is very unlikely that plants and animals would have retained enough sequence identity within introns such that their parallel capture would have produced the similarity of amino acid sequence within the insertion. It is more likely that the inserted sequence predates the radiation of plants and animals and is retained because of functional significance.

Several members of the TRX superfamily have additional modules at the carboxy terminus. The mouse NRX and PDIs have carboxyl extensions, but those extensions are not similar in sequence or potential structure to the Zn finger of maize NRX. Recently, however, a novel E. coli TRX sequence was reported to have an extension containing extra C residues in a proper sequence context to form a putative Zn finger, but this extension is located at the amino terminus (Miranda-Vizuete et al., 1997). The maize NRX carboxy terminus has clustering of basic residues suggestive of a bipartite nuclear localization motif, and sequence analysis (Lupas, 1996) hints at a potential for coil-coil interaction in this region. The strongest coil-coil potential exists near amino acids 471 to 493, which also has homology to a domain within a recently reported predicted Arabidopsis receptor-kinase-like protein (accession no. AL021633).

The presence within the nucleus of a multiple-TRX modular protein with a putative Zn finger offers intriguing possibilities for the regulation of transcription factors by alteration of their redox state. The number of transcription factors known to be influenced by changes in their own redox state is fairly large and increasing. TRX was shown early on to be associated with the replication machinery as an integral part of T7 polymerase, and even E. coli TRX had been noted to be associated with the nucleoid region of some E. coli cells (Holmgren, 1979; Nygren et al., 1981). The potential association of NRX with critical nuclear events is further enhanced by the presence within NRX of sequence elements that bear limited homology to the CCAAT-box-binding transcription factor or nuclear factor, mastermind, and v-erb-A (Fig. 10). Mastermind accumulates in the nucleus and is an important neurogenic locus. The limited homology with v-erb-A is also found in the thyroid hormone receptor α-2. Representative Zn fingers also depicted in Figure 10 hint at the putative maize NRX Zn finger as being similar to the human HTF 10 sequence, with the exception of having the C pairs more closely associated. Elucidation of these potential functional associations awaits further characterization of NRX and other NRX-like proteins.

Abbreviations:

DAP

days after pollination

NRX

nucleoredoxin

PDI

protein disulfide isomerase

TRX

thioredoxin