In this study, a comprehensive set of 25 nonredundant heat shock factors were identified and characterized from the current version of the maize (B73) genome. In a former publication, 22 maize Hsf isoforms were reported, which were composed of 16 Hsfs having intact ORFs and six 5' truncated Hsfs . The following are likely responsible for these discrepancies. In the previous studies, the maize genome had not been completely sequenced, 22 maize Hsf genes were identified by searching the publicly available maize EST and genomic sequence survey (GSS) databases for homology to rice Hsfs. In our study, the maize genome has been completely sequenced, therefore the maize genome database used here is more precise and complete than what was previously available.
Although the maize genome is approximately 6-fold larger than rice (2,300 Mb:389 Mb), the gene number is similar (3,2000:3,7000) and their genetic map organization is highly conserved. We found maize and rice have the same number of Hsfs . This partially accounts for the support of Hsfs conservation in these two species during the evolutionary process. In the investigation of conserved Hsf domains, we observed two class A Hsfs (ZmHsf-02, ZmHsf-24) lacking the AHA motif, which is essential for class A Hsfs transcription activity. Previous study suggests  these proteins bind to other class A Hsfs forming hetero-oligomers to achieve their functions.
Phylogenetic analysis of Hsfs in maize, rice and Arabidopsis indicated that ZmHsfs are more closely allied with OsHsfs than AtHsfs, consistent with the evolutionary relationships among maize, rice and Arabidopsis i.e. two monocots in the Poaceae Subclass Commelinidae and one dicot in the Brassicaceae Subclass Dilleniidae. The fact that all three classes (A, B and C) identified in maize, rice and Arabidopsis genes implies that the Hsf genes originated prior to the divergence of monocots and dicots. Hsfs of rice and maize appear more close relationship between each other in subclass A1 than to Hsfs from Arabidopsis. Such observations suggest the expansion of these Hsf genes following divergence of monocots and dicots.
The phylogenetic analysis showed that AtHsf-04 (HsfA2 type) and AtHsf-08 (HsfC1 type) were not grouped into subclass A2 and class C, respectively, and subclass A2 and class C were OsHsfs and ZmHsfs clusters. ZmHsfs and OsHsfs belong to the same clade, indicating that Hsfs of these subclasses expanded in a species-specific manner from common ancestral genes that were present prior to diversification of the monocot and dicot lineages. Phylogenetic data also proposed that subclass A2 and class C Hsfs were expanded in monocots but not in Arabidopsis. A single HsfA2 (AtHsf-04) is present in Arabidopsis. However, maize has four members and rice contains five in subclass A2. Class C consists of three maize and four rice members, while Arabidopsis has only one class C type member (AtHsf-08).
In addition, possible gene loss during the course of evolution was supported by phylogenetic reconstruction. Subclasses A7 and A8 exhibit interesting characteristics that monocots were not found in these two subclasses. Accordingly, this might indicates two dicot specific gene subclasses. Gene duplication events play a significant role in the amplification of gene family members in the genome [26, 27]. Research has estimated the fraction of retained paralogs is 72% in maize, having occurred over the course of 11 million years of evolution . The expansion mechanism of the maize Hsf gene family was analyzed to understand gene duplication events. Nine pairs of maize Hsf gene paralogs were identified. Among the paralogs, only one pair is involved in regional duplication in chromosome 1, however, two members in each of the other eight pairs were arranged between chromosomes. This result suggested the maize Hsf gene family expansion originated in a high number of large segmental duplications. An increase in the number of gene regulators (i.e. transcriptional and developmental regulators and signal transducers) is an essential factor in the evolution of more complex systems in different species . It is hard to achieve the expansions of these regulator gene classes only through single-gene duplications, which points to the importance of genome duplications in expanding the regulatory gene repertoire . It was estimated that more than 90% increase in regulatory genes had been caused by genome duplications in the Arabidopsis lineage in the last approximately 150 million years . Similarly, individual gene family expansion follows this rule. In plants, genome duplications have mainly contributed to expression of the Aux/IAA family of auxin response regulators . Data from studies of the maize genome revealed that its genome has experienced two rounds of genome duplications, an ancient duplication prior to the maize-rice divergence and a recent event following triploidization . The association of Hsf gene expansion in maize with these two rounds of maize genome duplication explains this observation and in addition sheds light on the evolutionary process of the maize Hsf gene family. Furthermore, segmental duplications occur more often in more slowly evolving gene families, e.g. MYB gene family . Due to the major role of segmental duplications in the Hsf gene family evolution, the maize Hsf gene family might hold a slow evolutionary rate.
Several approaches were employed for maize Hsf gene expression analysis by EST database. ZmHsf genes exhibited distinct expression patterns in different tissues or organs. One explanation is that ZmHsf genes have different expression patterns in various tissues and at multiple developmental stages. Expression profiles of 12 class A rice heat shock transcription factor genes have been resolved and the OsHsfA genes displayed tissue-specific expression under normal conditions . AtHsfA9 was exclusively expressed during the late seed development stage and controlled by the seed-specific transcription factor abscisic acid-insensitive 3 (ABI3) . Furthermore, the expression data revealed that the majority of duplicated ZmHsf gene pairs exhibited diverse expression patterns between two members. It suggested that functional diversification of the surviving duplicated genes is a major feature of the long-term evolution .
Expression analysis of quantitative RT-PCR showed that maize Hsf genes exist different expression levels by heat stress. In this study, we have detected three HsfA2-type ZmHsfs (ZmHsf-01, ZmHsf-04 and ZmHsf-17) with significantly higher expression, when subjected to heat stress. The result indicated that the ZmHsfA2 subclass was closely related with maize heat shock response. Moreover, six genes were remarkably up-regulated under heat stress condition, i.e. ZmHsf-01, ZmHsf-03 and ZmHsf-23, and et al., which suggested specific roles for these genes in maize during heat stress. It is noteworthy that three ZmHsfs (ZmHsf-03, ZmHsf-11 and ZmHsf-25) assigned to class B appeared to be strongly induced by heat stress. The Hsfs belong to class B lack certain structural features of the class A activator Hsfs. Class B-Hsfs may serve as transcriptional repressors or coactivator cooperating with class A Hsfs. But the functional roles of these three Class B-Hsfs in maize will require further investigations. It is likely that the Hsf genes remaining unaltered or down-regulated in expression may locate at downstream in the hierarchy of the events involved in heat shock response or are repressed by other members of the family . In addition, if Hsp proteins accumulate enough, they may be involved in feedback regulation to repress Hsfs activity, such as Hsp70 proteins. In the nine duplicated gene pairs of maize, the significant divergence of expression levels between the two members of each gene pair implied that duplicated genes had various functions in the response to heat stress in the evolutionary history.