empty pericarp2 Encodes a Negative Regulator of the Heat Shock Response and Is Required for Maize Embryogenesis

emp2 Is an Embryo-Lethal, Defective-Kernel Mutation

The emp2 defective-kernel mutation of maize is a recessive, lethal mutation required for both endosperm and embryo development (Scanlon et al., 1994, 1997). At maturity, emp2 mutant kernels contained few if any endosperm or embryo structures (Figure 1A) . Despite the empty-pericarp phenotype of mature mutant kernels, endosperm development did progress (albeit in a retarded manner) beyond the early stages before development was aborted (Figure 1B). The scarce endosperm material that did form in mutant kernels was either reabsorbed or became necrotic before kernel maturity. Likewise, embryogenesis in emp2 mutant kernels proceeded until the coleoptile stage, or more rarely the first leaf stage, whereupon development was arrested and the mutant embryos eventually decomposed (Figures 1C to 1J) (Scanlon et al., 1997). Close examination revealed that emp2 mutant embryo development was retarded severely at all stages compared with the wild type, although mutants did form a shoot apical meristem, a coleoptile (a portion of the maize cotyledon) (Weatherwax, 1920), and in some cases a leaf primordium before their eventual abortion (Figures 1H to 1J). Therefore, EMP2 functions very early in kernel development and is required for progression beyond the coleoptile stage/stage 1 of maize embryogenesis. Genetic analyses mapped the emp2 locus to the long arm of chromosome 2, whereas dosage analyses suggested that the emp2-R mutation is a fully recessive, null allele (Scanlon et al., 1994, 1997). Culturing of mutant embryos explanted from emp2 kernels at 12, 14, 16, 18, and 20 days after pollination (DAP) failed to rescue the embryo-lethal phenotype (Scanlon et al., 1997), indicating that the emp2 lesion is not a nutritional deficiency.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Mutant emp2 Kernels Abort Early in Development.

(A) Self-pollinated ears of emp2-R/Emp2 heterozygous plants segregate 1:4 collapsed, defective-kernel phenotypes.

(B) Endosperm development in emp2 mutant kernels. Although kernel filling does occur at earlier stages (20 and 24 DAP) in emp2-R mutant seeds, endosperm development eventually is aborted. At 36 DAP, much of the endosperm material that has accumulated in mutant kernels has been reabsorbed, and the mutant kernels collapse.

(C) to (J) Developmental comparisons of nonmutant sibling embryos at 7 DAP (C), 12 DAP (D), 14 DAP (E), and 24 DAP (F) and emp2-R mutant embryos at 12 DAP (G), 14 DAP (H), 16 DAP (I), and 24 DAP (J) reveal that mutant kernels are blocked variably after the development of the coleoptile or, more rarely, the first foliar leaf. Mutant embryo development also is retarded compared with that in nonmutant siblings, and in no cases did embryogenesis proceed further than the formation of a single leaf primordium. At 12 DAP, the nonmutant embryo (D) has no suspensor and has formed a coleoptile, a root pole, and a shoot meristem. By contrast, the 12-DAP emp2 mutant embryo is just initiating a shoot meristem and has retained the suspensor (G). At 14 DAP, nonmutant siblings have developed two leaf primordia (E), whereas the emp2 mutant is just initiating the coleoptile (H). Necrosis and tissue collapse commence at 16 DAP in mutant kernels (I). At 24 DAP, the nonmutant embryo has developed four leaf primordia (F). The 24-DAP emp2 mutant sibling embryo has initiated a coleoptile and a single leaf primordium; however, the embryo is collapsing and becoming necrotic (J). Note that the majority of emp2 mutant kernels examined had not developed leaf primordia but arrested development at the coleoptile stage. Numbers 1 to 4 designate leaf primordia at various stages of development, where 1 indicates the youngest primordium. c, coleoptile; en, endosperm; m, shoot apical meristem; p, pericarp; r, root meristem; s, suspensor; sc, scutellum. Bars = 50 μm.

Cloning of the emp2-R Mutation via Transposon Tagging

The emp2 reference mutation, emp2-R, was discovered in maize lines that were mutagenized with Mutator (Mu) transposons (Scanlon et al., 1994). A 3-kb Mu1-tagged EcoRI restriction fragment tightly linked to the emp2-R mutation (Figure 2A) was cloned into bacteriophage λ. A 550-bp EcoRI-BglI fragment, probe 1, is contained within the original cloned fragment flanking the Mu1 element (Figure 3) . This fragment was used as a single-copy probe by DNA gel blot hybridization analysis to confirm that the cloned fragment was linked to emp2-R (data not shown). The emp2-R mutation was identified in the self-pollinated ear of a single individual in an F1 family (hybrid Q66/Q67, which contained Mu transposon activity) comprising 20 sibling plants (Scanlon et al., 1994), indicating that the mutation arose during gametogenesis in the Mu parent. Sibling plants in this F1 family were examined to determine whether the 3.0-kb Mu1-containing EcoRI fragment was present in this line before the formation of the emp2-R mutation. If so, the 3.0-kb fragment would be present in half of the family members. Twenty sibling kernels of the single kernel that contained emp2-R were analyzed in DNA gel blot hybridizations using probe 1 (Figure 2B). All 20 plants contained EcoRI fragments of the same size as those seen in standard Q66/Q67; none contained a 3.0-kb fragment. Therefore, a wild-type allele of inbred Q67, a 1.6-kb EcoRI fragment, was modified by insertion of a 1.4-kb Mu1 element. Moreover, insertion of this specific Mu transposon and the appearance of the emp2-R mutation were concomitant events, suggesting that the cloned fragment linked to emp2-R represents a portion of the emp2 locus.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

DNA Gel Blot Analyses of Clones Linked to emp2.

(A) A Mu1 transposon-tagged EcoRI fragment (arrow) is linked to the emp2-R mutation.

(B) Nonmutant siblings (lanes 2 to 8) of the original isolate harboring the emp2-R mutation segregate for nonmutant inbred-sized (either inbred q66 or q67) restriction fragments of clones linked to the emp2-R mutation (probe 1, as described in Methods). The Mu1 transposon-tagged, 3-kb emp2-R–linked EcoRI fragment is absent from nonmutant sibling DNA, indicating that the emp2-R–linked allele did not preexist in the genetic background that gave rise to the emp2-R mutation.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Structural Map of the emp2 Gene.

The emp2 gene is composed of four introns and five exons (black boxes), including a long 5′ UTR (hatched box). The locations of Mu transposon insertions in the 5′ UTR and the intron corresponding to the known emp2 mutant alleles are indicated. The putative TATA box (GATAAA) and transcriptional start site (large arrow) are shown, as well as the HSE promoter consensus sequence, located upstream of the TATA box. AUG, translational start site; UAA, translational stop.

A 7.6-kb contiguous genomic clone containing sequences homologous with the 3-kb clone linked to emp2-R was identified from two distinct clones (constituting ∼27 kb of genomic DNA) obtained from an inbred Mo17 maize genomic DNA library (see Methods). Reverse genetic analyses were used to prove that this 7.6-kb maize genomic interval contains the emp2 locus. Oligonucleotide primers corresponding to genomic sequences throughout the emp2-linked contiguous sequence were used together with Mu transposon-specific primers in TUSC (Trait Utility System for Maize; Bensen et al., 1995) reverse genetic analysis to identify nine additional Mu transposon insertions within this cloned locus. Intriguingly, all 10 Mu insertions identified in emp2 mutant alleles were located within either the 5′ UTR or the first intron of the emp2 gene (Figure 3), despite the fact that oligonucleotide primers were prepared from throughout the emp2 gene. Complementation analyses of emp2-R heterozygous plants and plants harboring the newly identified Mu1 insertions (termed emp2-PI) revealed that all nine TUSC-derived insertions define kernel mutations that are allelic to emp2-R. These data indicate that the 7.6-kb contiguous clone contains a portion of the emp2 locus.

EMP2 Is Predicted to Encode HSBP1 of Maize, a Negative Regulator of the HSR

Sequencing of the 7.6-kb contiguous emp2 clone identified regions of exact homology with a 480-bp cDNA clone isolated from maize seedling RNA. RNA gel blot analyses (Figure 4) revealed that the emp2-homologous EST hybridized to an ∼700-bp transcript in maize tissues, indicating that the cDNA clone is not complete. Therefore, rapid amplification of cDNA ends (Clontech) was used to isolate nearly full-length emp2 cDNA; the exact transcriptional start site was identified by S1 nuclease protection assays (see Figure 6). These combined analyses revealed that the nonmutant emp2 transcript is 760 bp long, excluding the poly(A⁺) tail. Database searches and cDNA cloning indicated that multiple polyadenylation sites are used in emp2 transcripts, resulting in a broad hybridizing band in RNA gel blot analyses (Figure 4). Alignment of genomic and cDNA clones revealed that the emp2 gene is composed of five exons and four introns (Figure 3).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

RNA Gel Blot Analyses of emp2 Mutant Kernels.

(A) At 12 DAP, no differences in hsp70 or emp2 transcript abundance are detected in nonmutant (wt) and mutant (emp2) kernels. All gels represent separate hybridizations to the same blot.

(B) At 14 and 16 DAP, transcripts homologous with the heat shock genes hsp70, dnaj, and emp2 are overly abundant in emp2 mutant kernels, whereas hsp101 transcripts accumulate at 16 DAP. All gels represent separate hybridizations to the same blot.

(C) Polyacrylamide RNA gel blot hybridization reveals molecular mass heterogeneity for emp2-homologous transcripts in nonmutant and emp2 mutant 12-DAP kernels.

(D) Transcripts of emp2 and hsp101 are not upregulated in the defective-kernel mutant emp*6. Both gels represent separate hybridizations to the same blot.

(E) Abundant emp2 transcripts are detected in poly(A⁺)-selected RNA using internal emp2-specific probes (emp2 cDNA); however, the hybridization signal is much reduced in RNA from emp2 mutant kernels when the blot is stripped and rehybridized using only the 5′-most 174 nucleotides of the emp2 cDNA (5′ emp2 cDNA) as a probe. Note that more poly(A⁺) RNA is loaded in the nonmutant (wt) lane.

The emp2 cDNA clone is predicted to encode a 78–amino acid peptide that shows homology (∼36% identical and 60% similar amino acids over the full-length protein; Figure 5) with HSBP1. First identified in human by its ability to bind to the trimerization domain of HSF1, analyses in Caenorhabditis elegans demonstrated that HSBP1 functions as a negative regulator of the heat shock transcriptional response (Satyal et al., 1998). Multiple lines of evidence suggest that EMP2 functions as the maize ortholog of HSBP1 (described below and in the subsequent section). EMP2 is predicted to contain the conserved hydrophobic heptad repeat domains and to form the characteristic coiled-coil structure (Figures 5A to 5D) found in animal HSBP1 (Satyal et al., 1998; Tai et al., 2002). Furthermore, the positions of both introns found in human hsbp1 are conserved absolutely in emp2 (Figure 5E), although emp2 harbors two additional introns (Figures 3 and 5E) not found in animal hsbp genes.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

EMP2 Is Homologous with Animal HSBP1.

(A) Predicted amino acid sequence of EMP2. EMP2 contains conserved arrays of hydrophobic (residues labeled a and d) heptad repeats.

(B) Probability plot of the predicted EMP2 protein to form a coiled-coil structure using the software prediction application COILS and a window of 21 residues.

(C) and (D) Probability plots of the predicted EMP2 protein (C) and the predicted ZmHSBP2 protein (D) to form dimer and trimer multimers, as calculated by PAIRCOIL.

(E) Alignment of the predicted amino acids of the EMP2 gene product (ZmHSBP1) and other HSBP homologs from plants and animals. Identical amino acid residues are shown in black, and similar residues are shown in gray. The positions of introns conserved in human, maize, rice, and Arabidopsis hsbp genes are indicated by black asterisks; two additional introns conserved in plant hsbp genes are indicated by red asterisks. Two distinct HSBP-like proteins (HSBP1 and HSBP2) are identified in plants. Sequences were aligned with CLUSTAL W and BOXSHADE. Sequences examined are as follows (species names are given in parentheses): AtHSBP1 (Arabidopsis thaliana), CeHSBP1 (Caenorhabditis elegans), DmHSBP1 (Drosophila melanogaster), GaHSBP1 (Gossypium arboreum), GmHSBP1 (Glycine max), HsHSBP1 (Homo sapiens), HvHSBP1 (Hordeum vulgare), HvHSBP2 (Hordeum vulgare), LeHSBP1 (Lycopersicon esculentum), LjHSBP1 (Lotus japonicus), McHSBP1 (Mesembryanthemum crystallinum), MtHSBP1 (Medicago truncatula), OsHSBP1 (Oryza sativa), OsHSBP2 (Oryza sativa), PotHSBP1 (Populus balsamifera subsp trichocarpa), PtHSBP1 (Pinus taeda), SbHSBP2 (Sorghum bicolor), SbHSBP1 (Sorghum bicolor), SpHSBP1 (Schizosaccharomyces pombe), TaHSBP1 (Triticum aestivum), TaHSBP2 (Triticum aestivum), ZmHSBP1 (Zea mays), and ZmHSBP2 (Zea mays).

EMP2 is a novel HSBP-like protein in plants: homologous ESTs have been identified in five monocot species, eight eudicots, and a single gymnosperm (Figure 5E). Interestingly, although a single hsbp isoform has been identified in eudicots, two isoforms exist in monocots (Figure 5E). In addition, interspecific homologs of HSBP1 (or HSBP2) in grasses are related more closely than HSBP1 and HSBP2 from the same species. Therefore, the duplication of the hsbp gene most likely occurred before the evolutionary divergence of monocot grass species. Moreover, of the two rice genes closest to animal HSBP (Oshsbp1 and Oshsbp2) found in the entire sequenced genome, neither is predicted to encode a protein that has higher homology with animal HSBP1 than emp2 and Zmhsbp2, respectively (Figure 5E). In addition, the intron positions of maize emp2 are conserved completely among rice and Arabidopsis hsbp genes (Figure 5E). Therefore, it is unlikely that there are additional undiscovered maize hsbp genes that are more highly homologous with animal hsbp1 than emp2.

emp2 Mutant Embryos Exhibit an Unattenuated HSR

To understand the cause of emp2 mutant embryo abortion, we studied the steady state transcript abundance of the major HSPs in mutant kernels. If EMP2 functions as a negative regulator of HSR in maize, as it does in animals, it is predicted that the loss of EMP2 function may lead to an unattenuated HSR. Indeed, this is the case. Transcripts of hsp70 and dnaj genes were highly expressed in mutant kernels (but not in nonmutant sibling kernels) beginning at 14 DAP; normal levels of transcript were seen at 12 DAP (Figures 4A and 4B). In addition, transcripts of maize hsp101 also were upregulated in emp2 kernels at 16 DAP. Unlike hsp70 and dnaj, which are upregulated at the onset of heat shock, hsp101 was upregulated only after extended thermal stress (Queitsch et al., 2000). Thus, the overaccumulation of hsp101 transcripts indicates that emp2 mutant kernels are unable to attenuate the HSR. Furthermore, the timing of hsp transcript overaccumulation was well correlated with the developmental time point at which maize embryos are first competent to elicit HSR (Schoffl et al., 1998) and preceded emp2 mutant embryo necrosis at 16 DAP (Figure 1). Importantly, transcripts of emp2 and hsp101 were not upregulated in 16-DAP kernels homozygous for the unlinked defective-kernel mutation emp*6 (Figure 4D), which also causes early-stage embryo abortion. Therefore, the pattern of hsp upregulation in emp2 mutant kernels is not simply a phenotype exhibited by defective kernel mutants as a class, nor is it a stress response to embryo abortion.

EMP2 Transcripts Accumulate in Homozygous emp2 Kernels

The expression pattern of emp2 transcripts was analyzed in wild-type and emp2 mutant tissues. emp2 transcripts were detected in every nonmutant maize tissue examined, including developing kernels, roots, whole seedlings, leaves, and vegetative shoot apices (maize vegetative shoot meristem and the youngest five or six leaf primordia). At 12 DAP, no differences were detected in emp2 transcript size or abundance in nonmutant versus emp2-R mutant seeds (Figure 4A). At 14 and 16 DAP, however, there was a dramatic reduction in steady state transcript abundance in nonmutant kernels compared with that at 12 DAP (Figures 4A and 4B). This downregulation of steady state emp2 transcript coincided with the initial competence of maize embryos to elicit the heat shock transcriptional response (Pitto et al., 1983; Schrauwen et al., 1986; Vierling, 1991; Schoffl et al., 1998) and with embryonic developmental arrest in emp2-R mutant kernels. By contrast, emp2-R mutant kernels at 14 and 16 DAP exhibited significantly higher levels of emp2 transcript than nonmutant sibling kernels (Figure 4B). Upregulated emp2 transcript abundance also was observed in emp2-P1 and emp2-P9 mutant kernels at 14 DAP (data not shown). Moreover, transcript abundance in mutant kernels at 14 and 16 DAP was increased compared with that at 12 DAP (Figures 4A and 4B). In addition, sequence analyses of 20 individual cDNA clones obtained via hybridization to emp2 probes indicated no cross-hybridization to Zmhsbp2 and demonstrated that the upregulated transcripts are emp2 specific.

Sequence analyses of cDNA prepared from emp2 mutant transcripts revealed that the coding region of emp2-R mutant cDNA is normal. The apparent molecular mass of emp2-homologous transcripts was difficult to quantify from RNA gel blots, because the transcripts migrate in a broad band on agarose gels (Figures 4A and 4B). Moreover, high-resolution polyacrylamide gel hybridization revealed that emp2 transcripts are of heterogeneous molecular mass (Figure 4C). Sequence analyses of 13 emp2 mutant and 7 nonmutant transcripts revealed that a variety of different polyadenylation sites were used in emp2 transcripts (bp 623 to 760), which contributes to transcript size heterogeneity. In addition, emp2 alleles exhibited no preference for any particular polyadenylation site(s).

Mu Transposon Insertion in the 5′ UTR Region Correlates with Truncated emp2 Transcripts and Embryo-Lethal Mutations

A perplexing contradiction of the emp2-R phenotype (as well as the emp2-P1 phenotype) is the abundant accumulation of emp2 transcripts (Figure 4B) that contain unaltered coding regions in kernels homozygous for null, recessive emp2 mutations (Scanlon et al., 1994, 1997; this report). RNA gel blot analyses of poly(A⁺)-selected RNA from 16-DAP emp2-R mutant and sibling kernels revealed that sequences contained within a 174-bp fragment constituting the 5′ end of the emp2-R cDNA are underrepresented in mutant transcripts (Figure 4E). S1 nuclease protection analyses were used to locate the predominant transcriptional start site in nonmutant kernels to 17 bp downstream from the putative TATA box (Figures 6A and 6B) . By contrast, emp2-R mutant transcripts are 5′ truncated and prevalently initiate at 146 bp downstream of the preferred transcriptional start site in nonmutant siblings (Figures 6C and 6D). Interestingly, the mutant transcriptional initiation site is located 23 bp downstream from a second TATA box–like sequence (TATACA). These data suggest that Mu transposon insertion within the 5′ UTR of emp2 may cause alternative promoter utilization, perhaps resulting in an untranslatable transcript.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

S1 Nuclease Protection Analyses of emp2 Transcripts.

(A) The 5′ UTR sequence of emp2 is shown in uppercase letters, and the 30-bp promoter sequence is shown in lowercase letters. The putative upstream and downstream TATA boxes are underlined. The emp2 transcriptional start sites, as mapped by S1 nuclease protection assays, in nonmutant (position +1; presented in [B]) and emp2 mutant (position +146; presented in [E]) kernels are boxed.

(B) to (E) S1 nuclease protection assays of nonmutant (lanes 2, 4, and 7) and emp2 mutant (lanes 5, 8, and 10) transcripts reveal that the predominant transcripts in emp2 mutant kernels are 5′ truncated.

(B) Mapping of the transcription start site in the nonmutant transcripts by probe 2. A 24-bp fragment of probe 2 is protected by emp2 transcripts (lane 2).

(C) Mutant transcripts are 5′ truncated. Probe 3 is fully protected by nonmutant emp2 transcripts (lane 4) but is not protected by emp2 mutant transcripts (lane 5).

(D) 3′ transcripts are intact in emp2 mutant kernels. Probe 4 is fully protected by emp2 transcripts from both mutant (lane 8) and nonmutant (lane 7) kernels.

(E) The transcription start site of emp2 mutant transcripts. A 22-bp fragment of probe 5 is protected by emp2 mutant transcripts.

Undigested emp2 oligonucleotide probes are presented in lanes 1, 3, 6, and 9. The sequences of the emp2 oligonucleotide probes (probe 2, positions −11 to +25; probe 3, positions +16 to +48; probe 4, positions +392 to +421; and probe 5, positions +135 to +171) are described in Methods and Table 1.

emp2 Mutant Kernels Do Not Accumulate EMP2 Protein

To test our prediction that the 5′ truncated emp2 transcripts are not translated in emp2 mutant kernels, polyclonal antibodies were raised against predicted EMP2-specific antigens (see Methods) and used in immunoblot assays of proteins extracted from nonmutant and emp2-R and emp2-P9 mutant kernels (Figure 7) . The anti-EMP2 antibodies identified a single protein band of the approximate molecular mass predicted for emp2-encoded protein in 12- and 16-DAP nonmutant kernels. By contrast, no immunoreactive proteins were detected in emp2-R or emp2-P9 mutant kernels (Figure 7). These data reveal that null emp2 mutant kernels do not accumulate EMP2 protein.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

EMP2 Protein Does Not Accumulate in Null emp2 Mutant Kernels.

(A) and (B) Anti-EMP2 antibodies specifically recognize the EMP2 fusion protein. Lane 1 contains E. coli protein extract, including the bacterially expressed EMP2 fusion protein. The predicted molecular mass of the EMP2 fusion protein is ∼17 kD. Lane 2 contains extracts from bacterial constructs in which the emp2 cDNA is inserted out of frame.

(C) and (D) Analysis of 30 μg of protein extracted from nonmutant sibling (wt sib) emp2-R mutant kernels at 12 and 16 DAP and emp2-P9 mutant kernels at 14 DAP. The predicted molecular mass of the EMP2 protein is ∼8.5 kD.

anti-EMP2, immunoblots using an EMP2-specific antibody. CBB, polyacrylamide gels stained with Coomassie Brilliant Blue.

Abortion of emp2 Mutant Embryos Is Caused by an Unattenuated HSR

emp2 is an embryo-lethal, defective kernel mutation of maize (Scanlon et al., 1994, 1997). Embryogenesis in emp2 mutant kernels was aborted at the coleoptile stage/stage 1, followed by necrosis and reabsorption of kernel contents (Figure 1). Abortion of embryogenesis was preceded by an unattenuated HSR in homozygous emp2 embryos (Figure 4). Furthermore, both mutant embryo abortion and hsp transcript overaccumulation occurred during or soon after the coleoptile stage of maize development, the developmental time point at which plant embryos become competent to invoke the HSR (Pitto et al., 1983; Vierling, 1991; Schoffl et al., 1998). These data indicate that embryo abortion in emp2 mutant embryos is caused by unattenuated HSR. The HSR is known to result in transcriptional and translational arrest of numerous plant genes (reviewed by Nagao et al., 1986; Morimoto, 1991; Vierling, 1991), including those that affect cellular metabolism (Simoes-Araujo et al., 2002) and cell wall/membrane biosynthesis (Li and Showalter, 1996; Iwahashi and Hosoda, 2000; Simoes-Araujo et al., 2002). In addition, the HSR blocks essential cellular processes such as DNA synthesis and cell cycle progression (Song et al., 2001, and references therein). Thus, after competence to invoke the HSR is achieved after the coleoptile stage of development, an extended, unattenuated heat shock transcriptional response may trigger embryo abortion. Indeed, overexpression of a single HSP, HSP70, in Drosophila has been shown to reduce dramatically the rate of cell division and growth (Feder et al., 1992).

EMP2 Is an Essential Negative Regulator of the Heat Shock Transcriptional Response

HSBP1 was first described in animals as a negative regulator of the transcriptional activity of HSF1 during heat shock attenuation (Satyal et al., 1998). Here, we report that the loss of EMP2 function results in an unattenuated HSR and embryo lethality in maize kernels. In addition, the hydrophobic heptad repeat structure and the intron positions are conserved between EMP2 and animal HSBP1 (Figure 5). Together, these results suggest that EMP2 is the functional ortholog of HSBP1 in maize. However, hsbp1 knockouts in C. elegans do not show reduced viability after exposure to heat stress. This discrepancy indicates that EMP2 is an essential negative regulator of the HSR in maize, whereas redundant mechanism(s) of HSR attenuation may exist in C. elegans.

It has been demonstrated that the hydrophobic heptad repeats of HSBP1 interact with the oligomerization domain of HSF1 in animals. Although animals have 1 to 4 HSFs, plants may contain >20 (Nover et al., 2001). In addition, the oligomerization domains of animal and plant HSFs are divergent. Both class A and class C HSFs in plants have an expanded insertion of amino acid residues between the HRA and HRB domains. The extended distance between HRA and HRB in plant HSFs may make HSBP1 a necessary regulator of HSF activity. Furthermore, there are two predicted isoforms of HSBP in monocot grasses (Figure 5). Although the monocot HSBPs are extremely homologous in the heptad repeat region, HSBP2 has a distinct extension in the N terminus as well as a longer C-terminal tail than HSBP1 (Figure 5E). These differences may modify functional interactions between the heptad repeats of HSBP2 and the oligomerization domain of a specific HSF(s). Indeed, HSBP2 is likely to form dimers exclusively, as predicted by PAIRCOIL (Figure 5D), whereas EMP2/HSBP1 is more likely to form a mix of dimers and trimers. Moreover, the N-terminal extension found in ZmHSBP2 displays homology with chloroplast import motifs, suggesting that ZmHSBP2 may be targeted to the chloroplast. The coexistence of emp2 and Zmhsbp2 transcripts (data not shown) in developing embryos and the unattenuated HSR in emp2 mutant embryos further support the contention that ZmHSBP1 and ZmHSBP2 are not functionally replaceable in planta. Thus, the divergent but partially overlapping expression patterns of emp2/Zmhsbp1 and Zmhsbp2 in maize indicate that different mechanisms govern the expression of hsbp paralogs. Thus, we speculate that the two homologs of hsbp found in monocots may have evolved diversified functions in response to different upstream signals.

Mu Transposon Insertion Correlates with the Use of Alternative emp2 Transcriptional Start Sites

The emp2-R mutation is inherited recessively and renders a null mutant phenotype (Scanlon et al., 1994, 1997; this report). Despite the null mutant phenotype, emp-R kernels accumulated overly abundant levels of emp2 transcript (Figure 4). Furthermore, no Mu transposon sequences were detected within the mutant transcripts, and the coding region of emp2-R mutant mRNA was intact (data not shown). Determination of the exact size of emp2 transcripts via RNA gel blot analyses was complicated by the use of diverse polyadenylation sites, resulting in wide variation in the length of the 3′ UTR (Figures 4 and 5). However, S1 nuclease protection assays and RNA gel blot analyses revealed that the emp2 transcriptional start site in mutant kernels was 5′ truncated; emp2-R transcripts initiated at ∼146 bp downstream of nonmutant sibling mRNAs (Figure 6). Protein gel blot analyses accounted for null emp2 mutant phenotypes, revealing that the 5′ truncated emp2-R mutant transcripts yielded no detectable EMP2 protein (Figure 7). We propose that the 5′ UTR of the emp2 transcript harbors an element(s) important for its efficient translation. It has been demonstrated in both Drosophila and human that a lengthy and intact 5′ UTR is required for the preferential translation of hsp70 transcripts under stress conditions (Lindquist and Petersen, 1990; Hess and Duncan, 1996; Vivinus et al., 2001). If the 5′ UTR of emp2 has a function similar to that of hsp70, this may provide a mechanism to regulate the accumulation of EMP2 in response to heat stress, assuming that emp2 is transcribed constitutively (data not shown).

Furthermore, the data suggest that the EMP2 gene product may be negatively feedback regulated such that the presence of functional EMP2 protein leads to a reduction in steady state levels of emp2 transcript. Thus, the absence of functional EMP2 protein in mutant embryos may lead to overaccumulation of the nonfunctional mutant emp2 transcript. Additional, albeit speculative, support for such a feedback mechanism is provided by the presence of an inverted pentanucleotide repeat (nGAAnnCTTn) with identity to the consensus HSE promoter element located 400 bp upstream from the transcriptional start site of the emp2 gene (Figure 3). Therefore, it is possible that emp2 transcription can be enhanced by HSF binding to this HSE and thus may be influenced by feedback mechanisms during attenuation of the HSR.

Mu Transposons Insert Preferentially into the 5′ Region of emp2 Mutant Alleles

The emp2 locus is defined by 10 independent Mu transposon insertions, all of which are located in the 5′ UTR and/or the first intron (Figure 3). In addition, reverse genetic analyses identified five additional transposon insertions in this region of emp2, although complementation tests of these insertion stocks have not been completed. Therefore, despite that fact that emp2 gene-specific primers were prepared from several distinct regions of the emp2 locus, Mu insertions were identified only in 5′ genic regions. Previous surveys also have suggested a “preference” for Mu transposon insertion in 5′ promoter regions of maize genes (Dietrich et al., 2002), although no mechanistic explanation for this insertion site bias has been demonstrated. Nonetheless, these data suggest that genomic sequences present in the 5′ UTR, and within the first intron especially, may be important for directing the correct initiation site of emp2 transcripts.

EMP2 May Provide Additional Functions beyond Regulation of the HSR

Our results indicate that EMP2 is a negative regulator of HSR during attenuation and that, unlike HSBP1 in C. elegans, ZmHSBP1 function is required for embryo viability. Moreover, our results indicate that EMP2 performs an important developmental function(s) outside the realm of the HSR. Embryogenesis in emp2 mutant kernels was retarded severely at 12 DAP (Figure 1), well before maize kernels are competent to invoke the HSR (i.e., at the coleoptile stage; reviewed by Vierling, 1991). Moreover, no abnormal accumulation of hsp gene transcripts was observed in mutant kernels before 14 DAP (Figure 4). Thus, the developmental retardation of the emp2 mutant indicates an additional role (i.e., beyond the regulation of hsp gene regulation) for EMP2 in very early stages of maize development. The emp2 mutation provides a powerful genetic tool with which to investigate the expanded role of HSBP protein during maize embryogenesis.

Isolation and Propagation of Defective-Kernel Mutants

The emp2 mutation was identified in a self-pollinated F1 plant that was generated by outcrossing the maize (Zea mays) hybrid stock Q66/Q67 as female to plants containing active Mu transposons (tagging scheme described by Scanlon et al. [1994]). A specific F1 plant was found to be heterozygous for a dek mutation, designated emp2-R, by observing normal and defective kernels on the self-pollinated ear segregating at the ratio of 3:1. The emp2-R mutation was mapped to chromosome 2L:91 and shown to define a previously undescribed maize locus. Because emp2-R–homozygous kernels are inviable, the mutation was propagated by outcrossing emp2-R/Emp2 plants to the maize inbred lines B73 or Q66 as described previously (Scanlon et al., 1994, 1997).

DNA Gel Blot Analyses

Maize genomic DNA was isolated from immature ears or 7-day-old seedlings and analyzed by DNA gel blot hybridization analysis as described previously (James et al., 1995). Hybridization probes were as follows. Probe Mu1 is the 960-bp MluI fragment internal to transposon Mu1 (Barker et al., 1984). Probe 1 was obtained from the Mu-tagged genomic clone, a 550-bp EcoRI-BglI restriction fragment adjacent to the Mu1 insertion contained within emp2-R (Figure 4). Gel-purified double-stranded DNA restriction fragments used as hybridization probes were labeled radioactively by random primer extension. In all instances, labeled DNA probes were separated from free nucleotides by size exclusion chromatography using Sephadex G-50 (Sigma).

Cloning of Maize Genomic DNA Fragments Linked to emp2

DNA was isolated from the immature second ear of an emp2-R/Emp2 plant. Approximately 150 μg of genomic DNA was digested with EcoRI, and fragments were fractionated by electrophoresis on 0.8% agarose gels (ultra-grade agarose; Gibco BRL, Bethesda, MD). Fragments of the size range 2.5 to 4.0 kb were isolated by electroelution, purified by phenol extraction, and concentrated by ethanol precipitation. These fragments were ligated into EcoRI-digested phage λ vector NM1149 at an approximate molar ratio of 1:1, packaged into bacteriophage particles (Giga-pack gold II packaging kit; Stratagene, La Jolla, CA), and used to infect Escherichia coli strain C600 hfl. The library was screened by standard plaque hybridization using probe MJ960, and a single hybridizing plaque was identified among ∼6 × 10⁴ total plaques. After isolation of DNA from the Mu1-hybridizing recombinant bacteriophage isolate, the maize genomic DNA inserts were subcloned into the vector pBR322 to form plasmid pMS1.

Two separate genomic clones containing the full-length emp2 gene were obtained by screening a Mo17 maize genomic library (Stratagene) with probes linked to emp2 genomic and cDNA clones (Figure 3). Both clones were sequenced to identify a 7.6-kb contiguous region; the complete genomic sequence of emp2 has been deposited in GenBank. Furthermore, the positions of Mu transposon insertions in all emp2 alleles (Figure 3) were determined by sequencing of PCR products amplified with primers specific to the terminal inverted repeat of Mu and the emp2 genomic DNA (Table 1). The insertion positions have been deposited in GenBank.

Maize Transcript Analyses

For preparation of total RNA, 4 g of maize tissue was ground in liquid nitrogen with a mortar and pestle and thawed in Trizol lysis buffer (Gibco BRL). After extraction in chloroform, the RNA was precipitated with ethanol and resuspended in water. Yields of total RNA isolated by this procedure typically ranged from 500 to 875 μg/g fresh tissue weight. Polyadenylated RNA was selected by the PolyA Tract mRNA isolation system (Promega). Procedures for RNA gel blot analysis and hybridization of total maize RNA from agarose gels were described by Seeley et al. (1992). Procedures for polyacrylamide gel blot analyses of RNA were described by Thompson et al. (1992).

RNA gel blot hybridizations using oligonucleotide probes were performed as described by Thompson and Meagher (1990). For use as an RNA loading control, a 26-bp oligonucleotide probe homologous with the soybean 18S rRNA gene (Tanzer and Meagher, 1994) was used as a hybridization probe. The emp2 cDNA probe was amplified from the intact Emp2 cDNA clone using the F1 and R1 oligonucleotide primers (Table 1); the predicted PCR product corresponds to nucleotides 297 to 545 of the emp2 cDNA. The 5′ emp2 cDNA probe was composed of the first 174 bp of the emp2 cDNA and was obtained by PCR amplification using primers F2 and R2 (Table 1). The maize hsp101 cDNA was a gift from T. Young (University of California, Riverside), and the maize hsp70 cDNA clones were provided by P. Rogowsky (Ecole Normale Superieure de Leon, University of Lyon, Lyon, France). The cDNA probe corresponding to 320 bp from the 3′ end of the maize dnaj gene was amplified from maize kernel cDNA using the PCR primers F3 and R3 (Table 1). 5′ rapid amplification of cDNA ends was performed according to recommended protocols supplied with the Marathon cDNA Amplification Kit (Clontech, Palo Alto, CA) using the emp2-specific primers R4 and R5 (Table 1). SMART PCR cDNA libraries were constructed from 1 μg of total RNA from emp2 mutant and nonmutant kernels collected 16 days after pollination according to the manufacturer's recommended protocol (Clontech). Approximately 300,000 individual plaques were screened with radiolabeled emp2 cDNA probe (Table 1).

S1 nuclease protection assays were performed using end-labeled oligonucleotide probes (Table 1) as described (Goldrick et al., 1996). Probes 2 and 3 (Table 1) of the emp2 cDNA were used to quantify the abundance of 5′ and 3′ transcripts, respectively, in nonmutant and emp2 mutant kernels. Probes 4 and 5 (Table 1) were used to map the emp2 transcription start sites in emp2 mutant kernels.

Antibody Production, Recombinant Protein Expression, and Immunoblot Analyses

Maize kernels were ground in liquid nitrogen, rinsed with PBS, and resuspended in soluble protein extraction buffer (20 mM Tris-HCl, 2 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, and 200 mM NaCl, pH 8.0). Cellular debris was pelleted by centrifugation at 1000g for 3 min. Protein gel electrophoresis, transfer, and Coomassie Brilliant Blue R350 staining were performed according to standard methods (Sambrook et al., 1989). Thirty micrograms of protein was loaded in each lane.

emp2 cDNA clones in phage λ were converted directly to E. coli by one-step excision. The EMP2 recombinant proteins were expressed in the LacZ frame of vector pTriplEx (Clontech). Constructs with emp2 cDNAs (+157 to end) cloned into the SfiI site in frame with the LacZ translation initiation site were used for EMP2 recombinant protein expression, whereas the out-of-frame emp2 construct (+20 to end) served as a negative control. The predicted molecular mass of the bacterially expressed 158-residue EMP2 fusion protein is ∼17 kD. Total cellular proteins of E. coli were extracted as described (Sambrook et al., 1989) and used in immunoblot analysis to test the anti-EMP2 antisera. The EMP2 polyclonal antibodies were generated (Biosource International, Camarillo, CA) against the last 13 residues of EMP2, with a Cys residue added to the N terminus (N-CVKKPD-EPKPADSA-C). This sequence of residues in the predicted EMP2 protein is not present in the predicted protein encoded by Zmhsbp2. Immunoblot analyses were performed according to the manufacturer's recommendation (ECL; Amersham Pharmacia) using a 1:300 dilution of crude serum as the primary antibody.

Computational and Database Analysis

The EST and cDNA sequences were translated by ORF Finder (http://www.ncbi.nlm.nih.gov/gorf/gorf.html). Secondary structures of EMP2 and ZmHSBP2 were predicted using COILS (http://www.ch.embnet.org/software/COILS_form.html), and the probabilities of multimerization were calculated by PAIRCOIL. The multiple alignment was performed using CLUSTAL W and BOXSHADE (http://searchlauncher.bcm.tmc.edu/multi-align/multi-align.html). Sequences examined are listed in the legend to Figure 5. To determine the intron-exon border of hsbp genes, cDNA sequences were aligned against genomic sequences using BLAST 2 (Basic Local Alignment Search Tool) sequences. Genes whose intron positions were determined are Hshsbp1, Zmhsbp1, Athsbp1, Oshsbp1, and Oshsbp2.

Histology of Maize Kernels

Whole maize kernels were dissected from self-pollinated, heterozygous (emp2-R/Emp2) ears and fixed in paraformaldehyde as described (Scanlon and Freeling, 1998). Samples were sectioned at 10-μm intervals and mounted on slides as described. After deparaffinization, slides were stained in fast green or safranin/fast green as described (Sylvester and Ruzin, 1994).

Upon request, all novel materials described in this article will be made available in a timely manner for noncommercial research purposes.

Accession Numbers

Accession numbers for the genes and proteins described in this article are as follows: AF494285 (first intron of the emp2 gene), AF494284 (ZmHSBP1), Os.9890 and Os.1120 (Oshsbp1 and Oshsbp2), Zm.4289 and Zm.3728 (Zmhsbp1 and Zmhsbp2), and AAF053468 (dnaj). Accession numbers for the sequences compared in Figure 5 are as follows: AtHSBP1 (Arabidopsis thaliana, AV534620), CeHSBP1 (Caenorhabditis elegans, Q9u3b7), DmHSBP1 (Drosophila melanogaster, Q9vk90), GaHSBP1 (Gossypium arboreum, BF277371), GmHSBP1 (Glycine max, BF324235), HsHSBP1 (Homo sapiens, O75506), HvHSBP1 (Hordeum vulgare, AL509946), HvHSBP2 (Hordeum vulgare, BG344770), LeHSBP1 (Lycopersicon esculentum, AW624356), LjHSBP1 (Lotus japonicus, AW428820), McHSBP1 (Mesembryanthemum crystallinum, BE034188), MtHSBP1 (Medicago truncatula, AL381382), OsHSBP1 (Oryza sativa, AU075659), OsHSBP2 (Oryza sativa, BE040146), PotHSBP1 (Populus balsamifera subsp trichocarpa, AI166489), PtHSBP1 (Pinus taeda, BG039770), SbHSBP2 (Sorghum bicolor, AW746844), SbHSBP1 (Sorghum bicolor, BG411743), SpHSBP1 (Schizosaccharomyces pombe, O14330), TaHSBP1 (Triticum aestivum, BG606102), TaHSBP2 (Triticum aestivum, BE442689), ZmHSBP1 (Zea mays, AF494284), and ZmHSBP2 (Zea mays, BG840671). Other accession numbers are as follows: Hshsbp1 (AF068754 and NT_010422.9), Zmhsbp1 (AF494284 and AF494285), Athsbp1 (AV534620 and Z97339), Oshsbp1 (AU075659 and AAAA01003466.1), and Oshsbp2 (BE040146 and AAAA01000162.1).