Delayed Hepatocellular Mitotic Progression and Impaired Liver Regeneration in Early Growth Response-1-deficient Mice*

The early growth response-1 transcription factor (Egr-1) is induced as part of the immediate-early gene expression response during early liver regeneration. In the studies reported here the functional significance of EGR-1 expression during liver regeneration was examined by characterizing the hepatic regenerative response to partial hepatectomy in Egr-1 null mice. The results of these studies showed that liver regeneration in Egr-1 null mice is impaired. Although activation of interleukin-6-STAT3 signaling, regulation of expression of hepatic C/ebpα, C/ebpβ, cyclin D, and cyclin E and progression through the first wave of hepatocellular DNA synthesis occurred appropriately following partial hepatectomy in Egr-1 null mice, subsequent signaling events and cell cycle progression after the first round of DNA synthesis were deranged. This derangement was characterized by increased activation of the p38 mitogen-activated protein kinase and inhibition of hepatocellular metaphase-to-anaphase mitotic progression. Together these observations suggest that EGR-1 is an important regulator of hepatocellular mitotic progression. In support of this, microarray-based gene expression analysis showed that induction of expression of the cell division cycle 20 gene (Cdc20), a key regulator of the mitotic anaphase-promoting complex, is significantly reduced in Egr-1 null mice. Taken together these data define a novel functional role for EGR-1 in regulating hepatocellular mitotic progression through the spindle assembly checkpoint during liver regeneration.

The liver regenerates in response to many forms of injury (1)(2)(3). Liver regeneration is important for recovery from the functional hepatic deficits induced by such injuries and for host health and viability. Following injury, hepatic regeneration is precisely regulated in its initiation and proceeds only until the original liver mass is restored. This process does not depend upon a stem cell population; rather, all of the normally quiescent mature cellular populations of the liver have the potential to re-enter the cell cycle and proliferate. Hepatocytes are the first to replicate, followed sequentially by biliary epithelial, Kupffer, stellate, and sinusoidal endothelial cells (3). Normal liver architecture, distorted during the regenerative response, is then restored. Regeneration continues until normal hepatic function, mass, and structure are restored, and then it is precisely terminated as hepatocytes re-enter their pre-replicative state of quiescence. Although some aspects of the molecular machinery that control this carefully orchestrated response to injury are now known, many of the mechanisms involved in regulating liver regeneration remain incompletely understood. Elucidation of these signaling pathways in greater detail offers the potential to increase our understanding of the hepatic regenerative response in acute and chronic liver diseases, and may ultimately lead to novel therapeutic strategies for enhancing liver regeneration after hepatic injury. Furthermore, such studies could provide a paradigm for analyses of tissue regeneration in other organ systems.
Partial hepatectomy has been an important model system for investigating the molecular signaling pathways that regulate the hepatic regenerative response (4). Analyses using this model show that activation of the tumor necrosis factor-␣-IL-6 1 -STAT3 cytokine signaling cascade is required for initiation of liver regeneration (5)(6)(7). This is followed by the coordinated generation of mitochondrial reactive oxygen species (8) and prostaglandins (9), the activation of stress-and mitogen-activated-protein kinase cascades (10), as well as specific transcription factors including NFb, AP-1, STAT3, cAMP-response element-binding protein (9), and others (11)(12)(13). These events initiate an immediate-early gene expression program that includes increased expression of additional transcription factors, including the CCAAT enhancer-binding protein ␤ (C/EBP␤ (14)), EGR-1 (15), CREM (16) and other targets, which direct growth factor-dependent hepatocellular re-entry into and progression through the cell cycle (1).
EGR-1, which is also known as NGFI-A, KROX-24, ZIF268, and TIS8, is a zinc finger transcription factor induced as part of the immediate-early gene expression program in the regenerating liver (15,17). EGR-1 was first identified based on its rapid induction in rat PC12 cells in response to nerve growth factor (18). Subsequent studies have shown that EGR-1 expression is induced in a variety of models of cellular proliferation during the transition from G 0 to G 1 and also in models of cellular differentiation (19 -22); however, the specific functional role of EGR-1 in these models has not been defined. In fact, analyses of Egr-1-deficient embryonic stem (ES) cells indicate that EGR-1 activity is not required for ES cell proliferation or differentiation, and Egr-1 null mice exhibit a normal growth rate and lack any obvious defects in cellular differentiation (23,24). In order to investigate whether EGR-1 expression is required for normal liver regeneration, the response of Egr-1 null mice to partial hepatectomy was evaluated. The results of these analyses showed that EGR-1 deficiency is associated with an impaired hepatic regenerative response that is characterized by delayed hepatocellular progression through the mitotic spindle assembly checkpoint. These data provide evidence for a novel functional role for EGR-1 in hepatocellular mitotic progression during liver regeneration.

EXPERIMENTAL PROCEDURES
Animal Husbandry-Egr-1 null mice were derived from heterozygous (Egr-1 ϩ/Ϫ) mating pairs on a C57Bl/6 background (provided by Dr. Jeffrey Milbrandt, Washington University School of Medicine). The derivation of these mice has been described previously (24). Wild type (ϩ/ϩ) sibling offspring were used as controls. Mice were maintained on 12-h dark/light cycles and on a standard diet of ad libitum mouse chow and water before and after surgery. All experiments were approved by the Animal Studies Committee of Washington University and conducted in accordance with institutional guidelines and the criteria outlined in the "Guide for Care and Use of Laboratory Animals" (National Institutes of Health publication 86-23).
Animal Surgery-Egr-1 null (Ϫ/Ϫ) and wild type (ϩ/ϩ) mice were subjected to partial hepatectomy or sham surgery under methoxyflurane anesthesia as described previously (4,9). All surgeries were performed between 8 a.m. and 12 p.m. Animals aroused within 30 min of surgery and were returned to their cages with ad libitum access to food and water. At serial times after the surgery animals were sacrificed, and their livers were harvested either into formalin fixative for histological evaluation of hepatocellular proliferation or snap-frozen into liquid nitrogen for mRNA preparation for gene expression analyses or cellular lysate generation for immunoblot analyses. Two BrdUrd labeling strategies were employed for analyses of hepatocellular proliferation. In the first, sustained release BrdUrd-containing osmotic pumps (Alza, Newark, DE) were implanted into animals at the time of surgery, thereby labeling all hepatocytes that replicated over the entire course of the experiment. For this approach, the BrdUrd-containing pumps, which delivered a flow rate of BrdUrd of 1 l/h (20 g/ml, 20 g/h), were placed intraperitoneally at the time of the surgery. In the second method, animals were injected with 100 mg/kg BrdUrd 1 h before sacrifice at serial times after partial hepatectomy, thereby labeling only those hepatocytes that were dividing at the time of sacrifice.
Histology-Liver tissue was fixed for 24 h in 10% neutral buffered formalin, and prepared for histological analysis. Paraffin-embedded sections of liver were stained with hematoxylin and eosin and for immunohistochemical assessment of hepatocellular BrdUrd incorporation (using the BrdUrd Immunohistochemistry kit, Oncogene, Boston). For each tissue section, the frequency of hepatocellular nuclear BrdUrd labeling as well as mitotic body frequency was determined by examination of at least three different random ϫ400 fields and at least 300 cells and nuclei. Mitotic bodies were classified as pre-anaphase or anaphase/post-anaphase based on histomorphological evidence for the absence or presence of chromatid segregation.
Immunoblot Analysis-Whole cell lysates were made from snapfrozen liver as described previously (9). Protein concentration was determined using the bicinchoninic acid (BCA) protein assay kit (Pierce). Twenty five g of total cellular protein was subjected to SDS-PAGE, followed by electrophoretic transfer to nitrocellulose. Filters were probed with primary antibody (EGR-1, Santa Cruz Biotechnology, Santa Cruz, CA; p38, phospho-p38, phospho-MKK 3/6, STAT3, phospho-STAT3, Cell Signaling Technology, Beverly, MA), followed by a horseradish peroxidase-conjugated secondary antibody, and then developed using the ECL system (Amersham Biosciences). Densitometric analysis was performed with Scion Image data analysis software (Scion Corp., Frederick, MD).
RNA Preparation-Total RNA was prepared from mouse liver tissue using Triazol reagent (Invitrogen) followed by purification using the RNeasy total RNA clean-up protocol (Qiagen, Valencia, CA). RNA integrity was assessed by formaldehyde/agarose gel electrophoresis and quantified by 260 nm absorbance.
Real Time RT-PCR-based Gene Expression Analysis-The mRNA expression of C/ebp␣, C/ebp␤, cyclin D, cyclin E, Egr-1, Prl-1, and Cdc20 was evaluated by using real time RT-PCR. For these studies, total mRNA, prepared as described above from wild type and Egr-1 null mouse liver harvested at serial time points after partial hepatectomy, was reverse-transcribed to cDNA by the SuperScript Choice System (Invitrogen). For each gene analyzed, an aliquot of cDNA was added to a reaction mixture containing gene-specific forward and reverse primers (C/ebp␣, forward primer, 5Ј-CCTGAGAGCTCCTTGGTCA-3Ј, and reverse primer, 5Ј-GAAACCATCCTCTGGGTCTC-3Ј; C/ebp␤, forward primer, 5Ј-ACGACTTCCTCTCCGACCT-3Ј, and reverse primer, 5Ј-GA-GGCTCACGTAACCGTAGTC-3Ј; cyclin D, forward primer, 5Ј-GAAGG-AGACCATTCCCTTGA-3Ј, and reverse primer, 5Ј-GTTCACCAGAAGC-AGTTCCA-3Ј; cyclin E, forward primer, 5Ј-CTCGGGTGTTGTAGGTT-GCT-3Ј, and reverse primer: 5Ј-CTGTTGGCTGACAGTGGAGA-3Ј; Egr-1, forward primer, 5Ј-GGCGATGGTGGAGACGAGT-3Ј, and reverse primer, 5Ј-CGGCCAGTATAGGTGATGGG-3Ј; Prl-1, forward primer, 5Ј-ATTGCTGTCCATTGTGTCGC-3Ј, and reverse primer, 5Ј-C-ACCTTCAATTAATGCTAGGGCA-3; Cdc20 forward primer, 5Ј-GAGC-TCAAAGGACACACAGC-3Ј, and reverse primer, 5Ј-GCCACAACCGT-AGAGTCTCA-3Ј), deoxynucleotides, TaqDNA polymerase, and SYBR (Bio-Rad). Quantification of cDNA was based on monitoring increased SYBR fluorescence during exponential phase amplification in a real time PCR Machine (Bio-Rad) and determination of the PCR cycle number at which the amplified product exceeded a defined threshold (the "crossing threshold"). These data were standardized to the expression of ␤2-microglobulin, a constitutively expressed transcript commonly used as a reference mRNA in analyses of gene expression during hepatic regeneration (15), and the standardized data were used to calculate fold differences in gene expression (User Bulletin 2, ABI Prism 7700 Sequence Detection System, Applied Biosystems, Foster City, CA). Specificity of this assay was verified for each analyzed gene by confirmation of predicted product size and uniformity using melt curves and agarose electrophoresis of the PCR products. Specificity was further confirmed by simultaneous analysis of a "reverse-transcribed" reaction mixture containing all components except reverse transcriptase.
cDNA Microarray-based Gene Expression Analysis-Total hepatic mRNA was recovered from regenerating wild type or Egr-1 null mouse livers harvested 48 h after partial hepatectomy as described above. For each genotype, equal quantities of purified mRNA from replicate animals were pooled, and 10 g of this mixture was submitted to the Digestive Diseases Research Core Center Genomics Core Facility (Washington University School of Medicine) for cDNA microarraybased gene expression analysis. From each of the pooled samples, single-stranded cDNA probes containing either unique Cy3-(for wild type RNA) or Cy5 (for Egr-1 null RNA)-binding oligonucleotide extensions were synthesized using the 3DNA Array350 Expression Array Detection kit (Genisphere Inc., Hatfield, PA) according to the manufacturer's instructions. These probes were mixed, purified, concentrated, and hybridized to the NIA mouse 20k cDNA Microarray (25), after which the detection reagent containing primer-coupled Cy3 and Cy5 was prepared, mixed, and hybridized to the chip. After chip hybridization and washing performed according to the manufacturer's instructions, the microarrays were scanned by the GenePix Array Scanner (Axon Instruments, Union City, CA). The resulting image data were captured, converted to numerical fluorescence signal output, and imported into an Excel spreadsheet (Microsoft, Redmond, WA). As for the real time RT-PCR analyses described above, gene expression data were normalized to ␤2-microglobulin. The data set was filtered by removal of genes whose expression fell below a threshold level of fluorescence intensity (absolute fluorescence intensity Ͻ300) in both samples or were "flagged" based on nonuniform fluorescence intensity. For the remaining genes, expression in the wild type liver was compared with that in Egr-1 null mouse liver, and genes that were induced or suppressed at least 3-fold were identified. These criteria were selected based on experience indicating that they have a high probability of identifying differences in gene expression that can be independently validated by other methods of analysis. 2 Statistical Analysis-Data were analyzed using Sigma Plot software (SPSS, Chicago). Analysis of variance for multiple groups was used to determine statistical significance for BrdUrd labeling, mitotic body frequency, gene expression, and densitometry data. Results are reported as means Ϯ S.E. p Ͻ 0.05 was used to determine statistical significance.

Hepatocellular Proliferation during Liver Regeneration in
Egr-1 Null Mice-In order to determine whether EGR-1 expression is required for normal liver regeneration, the hepatic regenerative response to partial hepatectomy in wild type and Egr-1 null mice was examined. First, mice were subjected to partial hepatectomy and simultaneously implanted with a BrdUrd-containing sustained release osmotic pump at the time of surgery. The animals were allowed to recover, then sacrificed, and their livers harvested at 48 and 72 h after surgery. These times for analysis were chosen because they correspond to completion of this first wave of hepatocellular DNA synthesis (48 h) and of the majority of hepatocellular DNA synthesis (72 h), respectively, during murine hepatic regeneration following partial hepatectomy (9). Liver tissue was analyzed for hepatocellular BrdUrd incorporation which, with this methodology, provides an integrated measure of hepatocellular proliferation over the time course of the experiment. The data from this analysis showed that there is comparable hepatocel-lular proliferation in wild type and Egr-1 null mice 48 h after partial hepatectomy (62 Ϯ 10% versus 42 Ϯ 9%, p ϭ 0.12), indicating that progression through the first wave of hepatocellular DNA synthesis is not significantly delayed in Egr-1 null mice (Fig. 1, A and B). Analysis of liver harvested 72 h after partial hepatectomy showed that hepatocellular proliferation continues to increase from 48 to 72 h in wild type mice (p Ͻ 0.01) but not in Egr-1 null mice (p ϭ 0.42), and that at 72 h after partial hepatectomy such proliferation is significantly greater in the wild type compared with the null mouse liver (82 Ϯ 4% versus 53 Ϯ 7%, p Ͻ 0.001; Fig. 1, A and B). In order to determine whether this abnormality in hepatic regeneration in Egr-1 null mice results from a delay before or after the first wave of hepatocellular DNA synthesis, another series of experiments was performed in which mice were subjected to partial hepatectomy, allowed to recover, and then injected with BrdUrd 1 h prior to sacrifice at 36 and 48 h after partial hepatectomy. In this case, analysis of hepatocellular BrdUrd incorporation identifies only those hepatocytes undergoing replication at the time of sacrifice. These time points of analysis were selected because they correspond to the peak of the initial wave (36 h) and the beginning of the second wave (48 h) of hepatocellular DNA synthesis following partial hepatectomy (6). The data from this experiment showed that hepatocellular proliferation is comparable between wild type and Egr-1 null mice at 36 h after partial hepatectomy (37 Ϯ 12% versus 34 Ϯ 7%, p ϭ 0.82) but significantly reduced in Egr-1 null mice at 48 h after surgery (22 Ϯ 2% versus 12 Ϯ 2%, p Ͻ 0.04, Fig. 2, A and B). Taken together, these data show that in Egr-1 null mice liver regeneration proceeds relatively normally through the first round of hepatocellular DNA synthesis but that thereafter hepatocellular cell cycle progression is significantly delayed.
Early Signaling Events during Liver Regeneration in Egr-1 Null Mice-A number of early molecular signaling events induced by partial hepatectomy and required for appropriate initiation of hepatocellular proliferation during liver regeneration have been identified. For example, the tumor necrosis factor-␣-IL-6-STAT3 cytokine signaling pathway is necessary for normal initiation of hepatic regeneration, with induction of serum IL-6 and activation of hepatic STAT3 peaking 4 -6 h after partial hepatectomy (6,26). Another early signaling event characteristic of and necessary for normal liver regeneration is the induction of hepatic expression of C/ebp␤ (14). This occurs at the same time that hepatic C/ebp␣ expression is suppressed. These and other early signaling events direct the induction of expression of the G 1 cyclins D and E, which peak 48 h after partial hepatectomy and regulate hepatocellular G 1 -S progression and DNA synthesis (1). Therefore, in order to provide further evidence that the disruption of hepatic regeneration in Egr-1-deficient animals is downstream of the first wave of hepatocellular DNA synthesis, the regulation of these early signaling events was evaluated in Egr-1 null mice. First, the induction of IL-6 and activation of STAT3 were investigated by ELISA and immunoblot, respectively. The results showed that each of these signals is comparably induced 6 h after partial hepatectomy in Egr-1 null and wild type mice (Fig. 3, A and B). Next, changes in hepatic expression of C/ebp␣ and C/ebp␤ were evaluated by real time RT-PCR, and the results showed identical changes in the expression of each of these transcription factors 6 h after partial hepatectomy in Egr-1 null versus wild type mice (Fig. 3C). Finally, the regulation of expression of cyclins D and E was determined, and again the results showed that the induction of each of these G 1 -S cell cycle regulators occurs appropriately 48 h after partial hepatectomy in Egr-1 null mice. Together, these data demonstrate that the early signaling events leading up to the initial round of hepatocellular G 1 -S progression occur normally in Egr-1 null mice.
Regulation of Hepatic p38 MAP Kinase Activity during Liver Regeneration in Egr-1 Null Mice-p38 MAP kinase activity is an important negative regulator of cell cycle progression, in- cluding progression through stages subsequent to DNA synthesis (27)(28)(29). In order to determine whether delayed hepatocellular cell cycle progression during liver regeneration in Egr-1 null mice is associated with deranged activation (by phosphorylation) of p38, the steady state levels of phosphorylated p38 MAP kinase in wild type and Egr-1 null regenerating liver were compared with each other. First the temporal pattern of regulation of p38 activity in normal regenerating liver was determined by immunoblot analysis of quiescent and regenerating liver for phosphorylated and total p38. The results showed that phosphorylated p38 is present in quiescent liver, decreases from 2 to 24 h after partial hepatectomy, increases 36 h after partial hepatectomy, and then declines again at 48 h after partial hepatectomy (Fig. 4A). These changes were specific for regenerating liver as they were not induced by sham surgery (data not shown). Comparable amounts of total p38 were detectable in quiescent and regenerating liver at all time points examined (Fig. 4A). Thus, the hepatectomy-induced changes in hepatic levels of phosphorylated p38 are mediated by regulation of p38 phosphorylation/dephosphorylation and not by changes in the level of total p38. Next, p38 activity in regenerating liver from wild type mice was compared with that seen in Egr-1 null mice 48 h after partial hepatectomy. This time point corresponds to progression through the initial round of hepatocellular mitosis and initiation of the second round of DNA synthesis during liver regeneration. The results showed significantly greater levels of phosphorylated p38 in Egr-1 null mouse liver compared with wild type mouse liver (Fig. 4, B and C, p Ͻ 0.02), with a comparable amount of total p38 in each. Finally, the activation (by phosphorylation) of MKK-3/6, the upstream p38 MAP kinase kinase (30), was determined in each of these tissue samples, and the results showed significantly greater levels of phosphorylated MKK-3/6 in Egr-1 null versus wild type regenerating liver 48 h after partial hepatectomy (Fig. 4, B and C, p Ͻ 0.001). Together these results indicate that the impairment in cell cycle progression after the first round of hepatocellular DNA synthesis in Egr-1 null mice is associated with inappropriately sustained p38 MAP kinase cascade activation.
Hepatocellular Mitotic Progression during Liver Regeneration in Egr-1 Null Mice-p38 has been shown to regulate cell cycle progression through G 2 (28) and also through the mitotic metaphase to anaphase transition (29). Therefore, whether the delay in hepatocellular cell cycle progression in Egr-1 null mice occurs before, during, or after mitosis was examined next. Hematoxylin and eosin-stained liver sections from wild type and Egr-1 null mice were analyzed for total mitotic body frequency after partial hepatectomy. The results showed that mitotic body frequency at 48 h after partial hepatectomy, following the first wave of hepatocellular DNA synthesis, is greater in wild type (10.0 Ϯ 2.0 mitoses per high powered field) versus Egr-1 null mice (5.6 Ϯ 0.8 mitoses per high powered field); however, this difference did not reach statistical significance (Fig. 5, A and B, p ϭ 0.06). Next, whether progression through the metaphase-anaphase mitotic spindle assembly checkpoint is delayed in regenerating Egr-1 null hepatocytes was examined. To answer this question, these liver sections were reexamined to determine the frequency of pre-anaphase and anaphase/post-anaphase mitotic bodies detectable in each. Mitoses were classified as anaphase/post-anaphase if they showed histomorphological evidence of chromatid segregation (e.g. as seen in the inset in Fig. 5B). The results showed comparable numbers of pre-anaphase mitotic bodies in regenerating liver from wild type (7.4 Ϯ 1.5 per high powered field) and Egr-1 null mice (5.2 Ϯ 1.8 per high powered field, Fig. 5B, p ϭ 0.22), but a significantly greater number of anaphase/postanaphase mitotic bodies in regenerating liver from wild type mice (2.7 Ϯ 0.5 in wild type versus 0.4 Ϯ 0.1 in Egr-1 null per high powered field, Fig. 5B, p Ͻ 0.01). Taken together, these data show that mitotic progression in Egr-1 null regenerating

FIG. 4. p38 MAP kinase cascade activation in wild type and
Egr-1 null mice. A, immunoblot analysis of hepatic phosphorylated and total p38 MAP kinase during liver regeneration in wild type mice. B, immunoblot analysis of hepatic phosphorylated and total p38 MAP kinase and phosphorylated p38 MAP kinase kinase (MKK-3/6) in wild type and Egr-1 null regenerating mouse liver 48 h after partial hepatectomy. C, densitometric analysis of steady state levels of phosphorylated p38 MAP kinase, total p38 MAP kinase, and MKK-3/6 from panel B (*, p Ͻ 0.02 versus wild type; **, p ϭ 0.68 versus wild type).

FIG. 5. Hepatocellular mitotic body frequency in wild type and
Egr-1 null mice after partial hepatectomy. A, total, pre-anaphase, and anaphase/post-anaphase hepatocellular mitotic body frequency (mitoses per ϫ400 high powered field) 48 h after partial hepatectomy in wild type and Egr-1 null mice (4 -5 animals for each genotype; *, p Ͻ 0.01 versus wild type). B, representative ϫ200 liver sections from wild type and Egr-1 null mice harvested 48 h after partial hepatectomy and stained with hematoxylin and eosin (blue arrowheads identify preanaphase mitotic bodies, and yellow arrowheads identify anaphase/ post-anaphase mitotic bodies). Inset in upper right corner of wild type liver section shows ϫ400 image of a representative post-anaphase mitotic body. mouse liver is impaired between metaphase and anaphase at the mitotic spindle assembly checkpoint. In order to further characterize the nature of this impairment, mitotic body frequency in regenerating liver from wild type and Egr-1 null mice 72 h after partial hepatectomy was examined. The results showed decreasing mitotic body frequency from 48 to 72 h in wild type regenerating liver (10.0 Ϯ 2.0 to 3.0 Ϯ 0.4 mitoses per high powered field, p Ͻ 0.01) but relatively constant mitotic body frequency over this same time span in Egr-1 null liver (5.6 Ϯ 0.8 versus 4.9 Ϯ 0.3, p ϭ 0.51, data not shown). The observation that pre-anaphase mitotic hepatocytes did not accumulate in the regenerating Egr-1 null mouse liver suggests that the nature of the regenerative impairment in these mice is delayed progression through rather than arrest at the mitotic checkpoint. An alternative possibility is that hepatocytes are arrested at the mitotic checkpoint but that they are subsequently eliminated, e.g. by apoptosis, such that they do not accumulate within the liver. However, there was no increase in hepatocellular apoptosis noted by routine histological analysis or terminal dUTP nick-end labeling staining (data not shown) in Egr-1 null mouse liver.
EGR-1 Expression during Liver Regeneration-Egr-1 gene expression in the regenerating liver was initially characterized as part of the immediate-early gene expression program directing hepatocellular G 0 -G 1 progression (15,17). Whether Egr-1 is also expressed at later times during the regenerative response has not been investigated previously. The data reported above indicate that during liver regeneration Egr-1 deficiency is as-sociated with disruption of cell cycle progression events subsequent to G 0 -G 1 -S progression, specifically mitotic progression. Therefore, the pattern of Egr-1 gene expression at later times during the hepatic regenerative response was investigated. Egr-1 mRNA levels were quantified in quiescent and regenerating liver harvested from replicate wild type mice at serial time points after partial hepatectomy or sham surgery using real time RT-PCR. The results showed an early increase in hepatic Egr-1 mRNA 2 h after either partial hepatectomy or sham surgery, and a subsequent increase in Egr-1 expression beginning at 6 h and peaking 12 h after partial hepatectomy and not seen after sham surgery (Fig. 6A). After declining toward base line, hepatic Egr-1 expression again increases from 48 to 72 h after partial hepatectomy (Fig. 6A). These data show that Egr-1 mRNA expression is specifically increased in regenerating liver after the immediate-early gene expression response. Next EGR-1 protein expression was examined by immunoblot analyses of lysates derived from quiescent and regenerating liver. The results showed the presence of an ϳ70-kDa band corresponding to EGR-1 in quiescent liver, which decreases from 2 to 6 h after partial hepatectomy, then increases from 12 to 36 h after surgery, followed by subsequent decline (Fig. 6B). EGR-1 DNA binding activity has been detected previously in quiescent liver (17), consistent with the immunoblot data reported here showing EGR-1 protein detectable in quiescent liver. As expected, the protein band corresponding to EGR-1 was not detectable in lysates derived from Egr-1 null mouse liver 48 and 72 h after partial hepatectomy (Fig. 6C). Taken together, these data show that EGR-1 is expressed in regenerating liver during hepatocellular mitotic progression.
Prl-1 Expression during Liver Regeneration in Egr-1 Null Mice-EGR-1 has been implicated as an important transcriptional regulator of expression of the phosphatase of regenerating liver-1 (Prl-1) (17, 31). Prl-1 was first identified during gene expression analyses comparing quiescent to regenerating liver and was subsequently characterized as a gene encoding a novel protein-tyrosine phosphatase highly induced during early liver regeneration (32). The Prl-1 promoter sequence contains consensus EGR-1-binding sites, and studies have demonstrated that EGR-1 is capable of transactivating the Prl-1 promoter in cell culture (17). These observations suggest that EGR-1 may be an important in vivo regulator of Prl-1expression during liver regeneration. To test this, and thus determine whether PRL-1 may be a candidate regulator of mitotic progression, the pattern of induction of Prl-1 gene expression in wild type regenerating mouse liver was compared with that seen in liver derived from Egr-1 null mice. First, Prl-1 mRNA expression was quantified in quiescent and regenerating wild type mouse liver from replicate wild type mice at serial time points after partial hepatectomy using real time RT-PCR. The results showed that hepatic Prl-1 mRNA expression is increased 3-6fold in regenerating liver between 2 and 6 h after surgery (Fig.  7A). Next, comparison of wild type and Egr-1 null mouse liver recovered 6 h after partial hepatectomy showed comparable Prl-1 expression in the presence and absence of EGR-1 (Fig.  7B). These data show that hepatic induction of PRL-1 expression following partial hepatectomy in vivo is not dependent on EGR-1.
Cdc20 Expression during Liver Regeneration in Egr-1 Null Mice-In order to identify candidate genetic targets of EGR-1 that could be involved in regulating hepatocellular mitotic pro-gression during liver regeneration, hepatic gene expressions in wild type and Egr-1 null mouse liver after partial hepatectomy were compared with each other by cDNA microarray-based gene expression analysis. mRNA recovered from regenerating liver harvested 48 h after partial hepatectomy was used for this analysis because EGR-1 protein is detectable in wild type regenerating liver (Fig. 6B), and anaphase/post-anaphase mitotic body frequency is decreased in Egr-1 null regenerating liver (Fig. 5) at this time point. Microarray data were screened for those genes whose expression was increased or decreased at least 3-fold in regenerating liver from wild type versus Egr-1null mice, because such criteria have a high probability of identifying differences in gene expression that can be independently validated by other methods of analysis. 2 Based on these criteria, 38 known genes and 5 expressed sequence tags whose expression was higher in wild type regenerating mouse liver (Table I) and 68 known genes and 37 expressed sequence tags whose expression was greater in Egr-1 null regenerating mouse liver (Table II) were identified. One of the known genes identified as more highly expressed in wild type regenerating liver, Cdc20, was of particular interest. CDC20 regulates the anaphase promoting complex (APC) that directs progression through the mitotic spindle assembly checkpoint (33,34), and thus its differential expression in wild type versus Egr-1 null mice could theoretically explain the observed impairment in TABLE I Genes expressed more highly in wild type regenerating liver Summary of microarray expression analysis comparing wild type and Egr-1 null regenerating liver 48 h after partial hepatectomy. Genes whose expression is at least 3-fold greater in wild type regenerating liver are listed, along with the fold difference in expression and Unigene number. MHC indicates major histocompatibility complex.  mitotic progression seen in Egr-1 null mice during liver regeneration. In order to confirm these microarray data and to further evaluate the time course of expression of Cdc20 during liver regeneration, Cdc20 mRNA expression was quantified by real time RT-PCR in wild type and Egr-1 null regenerating mouse liver. First Cdc20 expression in quiescent and regenerating wild type liver was determined. The results showed that hepatic Cdc20 gene expression is unchanged over the first 12 h following partial hepatectomy, and subsequently increases beginning 36 h and peaking 48 h after partial hepatectomy (Fig.  8A). This pattern of Cdc20 gene expression corresponds closely with that of EGR-1 protein expression during liver regeneration. Next the expression of Cdc20 in regenerating wild type and Egr-1 null mouse liver harvested 48 h after partial hepatectomy was determined. Consistent with the results of the microarray analysis, this showed significantly greater Cdc20 expression in wild type compared with null mouse liver (Fig.  8B, p Ͻ 0.04). Together these data suggest that EGR-1-dependent Cdc20 gene expression is required for normal hepatocellular mitotic progression during liver regeneration.

DISCUSSION
The functional role of EGR-1 in regulating liver regeneration was first suggested by studies demonstrating that Egr-1 mRNA expression is specifically increased during hepatocellular G 0 -G 1 progression as part of the immediate-early gene response of the regenerating liver (17). Indeed, this pattern of hepatic Egr-1 expression following partial hepatectomy is homologous to that seen in a number of other models of cellular proliferation (17)(18)(19)(20). Based on these observations, EGR-1 might be predicted to regulate G 1 -S phase progression during liver regeneration, as has been shown for Crem1, another immediate-early gene induced in the regenerating liver (16). Mice in which the Crem1 gene is deleted exhibit a delay in the induction of cyclin D1 expression and the progression through the initial round of hepatocellular DNA synthesis following partial hepatectomy (35,36). In contrast, the analyses of liver regeneration in Egr-1 null mice reported here show that the early signaling events leading up to re-entry of quiescent hepatocytes into the cell cycle and progression through the first wave of DNA synthesis occur normally, but hepatocellular cell cycle progression after that is markedly impaired in these animals. These data demonstrate for the first time that EGR-1 is important for normal liver regeneration and also identify a functional role for EGR-1 during liver regeneration, i.e. modu-  lation of mitotic cell cycle progression. Consistent with this, Egr-1 gene and EGR-1 protein expression are specifically increased in regenerating liver at times that are subsequent to hepatocellular G 0 -G 1 progression and that overlap with mitotic progression.
Whereas the specific signals that regulate Egr-1 expression during liver regeneration have not yet been defined, a number of candidates are worthy of consideration. For example, a similar phenotype of delayed hepatocellular mitotic progression during liver regeneration to the one reported here has also been reported in rats expressing a dominant-negative IB mutant, which prevents activation of NFB (37). The mechanism by which inhibition of NFB activation resulted in mitotic delay in that study was not identified, although the authors speculated that an NFB-responsive gene product might be required for mitotic cell cycle progression. Those animals also demonstrated increased hepatocellular apoptosis following partial hepatectomy, a phenotype that was not observed in Egr-1 null mice (data not shown). Nevertheless, these observations raise the possibility that NFB regulates Egr-1 expression during liver regeneration. In fact, the promoter sequence of the human EGR-1 gene contains a potential NFB-binding site (38). Additional candidate regulators of Egr-1 expression during liver regeneration are likely to include IL-6 and C/EBP␤, because induction of Egr-1 gene expression has been shown to be impaired in mice in which either of these genes is disrupted (14,39). As reported here for Egr-1 null mice, liver regeneration is also markedly impaired in both Il-6-null and C/ebp␤-null mice; however, in each of those cases the impairment is proximal to the first round of hepatocellular DNA synthesis, indicating that IL-6 and C/EBP␤ must also be involved in earlier signaling events necessary for normal hepatocellular G 1 -S progression during liver regeneration.
The microarray-based gene expression data presented here suggest that at least one of the targets of EGR-1 transcriptional regulation during liver regeneration is Cdc20. CDC20 regulates the APC, and it is known to be modulated by mitosisspecific transcription, CDC28 kinase-dependent activation, and targeted degradation by the activated APC (33,34). The observations reported here suggest that EGR-1 regulates hepatocellular mitotic progression during liver regeneration by directly or indirectly inducing Cdc20 gene expression.
Previous studies have shown that proliferation and differentiation in Egr-1-deficient ES cells is normal (23). Together with the results reported here, this suggests that the regulation of hepatocellular cell cycle progression in the regenerating liver and that of ES cells in cell culture must be distinct from each other. In fact, regulation of cell cycle control in ES cells is known to differ from that of differentiated cells at the G 1 checkpoint in that progression from G 1 into S phase in differentiated cells, but not in ES cells, is controlled by the retinoblastoma protein (40). Together these data suggest that regulation of mitotic progression also differs between hepatocytes and ES cells. Thus analyses of the molecular regulation of liver regeneration in response to partial hepatectomy should not only continue to provide insight into the signaling mechanisms that regulate the hepatic regenerative response itself but can also offer additional information about similarities and differences in regulation of cellular proliferation in differentiated and undifferentiated cells.