The Role of C/EBP Isoforms in the Control of Inflammatory and Native Immunity Functions*

CAAT/enhancer-binding proteins (C/EBPs) are a family of leucine zipper transcription factors involved in the regulation of various aspects of cellular differentiation and function in multiple tissues. Six different members of the family have been isolated and characterized (C/EBPa to z), all sharing a strong homology in the carboxyl-terminal domain, which carries a basic DNA-binding domain and a leucine zipper motif. The general characteristics and patterns of expression of the C/EBP family have been described in the first minireview of this series (1). Here I will focus on the functions of several C/EBP family members in regulating various aspects of inflammation and immunity in the liver and in cells of the myelomonocytic lineage, in vitro as well as in vivo.

CAAT/enhancer-binding proteins (C/EBPs) 1 are a family of leucine zipper transcription factors involved in the regulation of various aspects of cellular differentiation and function in multiple tissues. Six different members of the family have been isolated and characterized (C/EBP␣ to ), all sharing a strong homology in the carboxyl-terminal domain, which carries a basic DNA-binding domain and a leucine zipper motif. The general characteristics and patterns of expression of the C/EBP family have been described in the first minireview of this series (1). Here I will focus on the functions of several C/EBP family members in regulating various aspects of inflammation and immunity in the liver and in cells of the myelomonocytic lineage, in vitro as well as in vivo.

Regulation of Liver Acute Phase Genes
Among the many liver-specific or liver-enriched genes whose expression is variably regulated by members of the C/EBP family of transcription factors are a very prominent class of genes coding for acute phase (AP) proteins. These are plasma proteins whose levels of expression are either positively or negatively regulated during the acute phase of inflammation (reviewed in Ref. 2). As initially suggested by the observation that hepatocytes are able to respond to local tissue injury at distal sites, it has been demonstrated that a number of cytokines and hormones are involved in the regulation of the AP response (reviewed in Ref. 3). AP genes have been divided into two major classes according to the pattern of responsiveness to cytokines. For maximal induction class I genes require a combination of both interleukin (IL)-1 and IL-6, sometimes with the additional need for glucocorticoids. In contrast, class II genes are solely responsive to IL-6 and related cytokines, either alone or in combination with dexamethasone. Functional C/EBP-binding motifs (initially known as type I IL-6-responsive elements (IL-6REs)) have been characterized on the promoters of most class I genes (hemopexin, haptoglobin, ␣ 1 -acid glycoprotein, serum amyloid A1, A2, and A3, complement C3, C-reactive protein), often in association with IL-1-responsive NF-B sites (see below) (4 -14). In contrast, in class II genes such as ␣ 2 -macroglobulin mainly type II IL-6REs, which bind the cytokine-inducible transcription factors Stat3 and Stat1 (reviewed in Ref. 15), have been identified. Type II IL-6REs are also found in promoters of class I genes.
The activity and/or mRNA and protein levels of various C/EBP genes are differentially modulated in response to inflammatory stimuli and to recombinant cytokines. Indeed C/EBP␤, the second C/EBP family member to be isolated, was originally identified thanks to its inducibility by IL-6 or by IL-1 in human hepatoma cells and in a glioblastoma cell line, respectively (16,17). It has been subsequently determined by many independent studies that both C/EBP␤ and -␦ are strongly up-regulated at the transcriptional level by inflammatory stimuli such as turpentine oil and bacterial lipopolysaccharide and by recombinant cytokines such as IL-6, IL-1, and TNF-␣ (reviewed in Ref. 18). Conversely, C/EBP␣ is slightly down-regulated under the same conditions (19). Studies of the interplay of different C/EBP factors on the promoters of several AP genes in hepatocytes have shown that at steady state, the majority of DNA-protein complexes contain various forms of C/EBP␣ homodimers and C/EBP␣/C/EBP␤ heterodimers (20). Upon AP induction, however, the amount of complexes containing C/EBP␣ is dramatically reduced, replaced by C/EBP␤ and C/EBP␦, although the exact composition of these complexes varies in the different experimental systems (6 -9, 21, 22). Although C/EBP␦ mRNA and protein levels are almost undetectable under uninduced conditions, C/EBP␤ is relatively abundant in several tissues, including the liver, even before induction. C/EBP␤ is known to undergo a series of post-translational modifications that modulate its activity. The first suggestion that induced post-translational modifications can increase the transcriptional activating potential of C/EBP␤ came from studies on Hep3B cells, in which transfected C/EBP␤ was able to induce transcription of a reporter gene much more efficiently in the presence of IL-6 (16). Subsequently, several phosphorylation sites have been demonstrated on this protein. Phosphorylation of serine 276 has been shown to take place in a pituitary cell line in response to intracellular Ca 2ϩ increase via a calcium/calmodulin-dependent kinase (23); activation of the protein kinase C pathway causes phosphorylation of Ser-105 in HepG2 cells (24). Finally, activation of mitogen-activated protein kinase following induction of the Ras pathway leads to phosphorylation of Thr-235 (25). Although all these events lead to increased transactivating potency of C/EBP␤, only the last one can be linked to IL-6 signaling. In contrast, the C/EBP␦-dependent activation of target AP genes is considered to be solely secondary to transcriptional activation of its gene. Accordingly, ectopical expression of C/EBP␦ in hepatoma cells is sufficient to transactivate responsive promoters in a cytokine-independent manner (21).

Regulation of Other Genes Involved in Inflammation
Far from being liver-specific, induction of both C/EBP␤ and -␦ levels by inflammatory stimuli occurs in most tissues analyzed, thus suggesting a more general role of these two factors in inflammation (17,26). Notably, expression of various C/EBP isoforms is differentially induced during macrophage and/or granulocyte differentiation. C/EBP-binding motifs have been identified in the functional regulatory regions of various genes expressed by cells of the myelomonocytic lineages, including those encoding the inflammatory cytokines IL-6, IL-1␤, and TNF-␣ (17,(27)(28)(29), other cytokines such as IL-8 and IL-12 (30,31), genes encoding proteins important for macrophagic or granulocytic functions such as inducible nitric oxide synthase (32), lysozyme (33), myeloperoxidase (34), and neutrophil elastase (35), the gene encoding the granulocyte colony-stimulating factor (G-CSF) (36), and the macrophage, granulocyte, and granulocyte-macrophage receptor genes (37). Interestingly, C/EBP sites are also present on several viral promoters (38 -43). Of particular relevance in this context is the finding that C/EBP sites are specifically required for replication of the human immunodeficiency virus-1 in macrophages but not in CD4 ϩ cells (44). However, due to the complexity of the C/EBP family and to the co-expression of several family members in the same cell, the relative role of each C/EBP isoform in the regulation of all these genes is still unclear.

Interactions with Other Transcription Factors
Studies of different promoters have assigned a predominant role for the induction of particular genes to either C/EBP␤ or C/EBP␦ (6 -9, 21, 22). Although this may be partly because of different experimental conditions, it is also likely to reflect physiological differences due to promoter context, to the relative affinity of specific sites for C/EBP proteins, and to the interactions with other transcriptional activators or co-activators. It is also interesting to note that different promoters all containing C/EBP binding sites can undergo completely divergent regulation. For example, the negative AP reac-tant mouse albumin gene carries various high affinity C/EBP motifs in its promoter, but its expression is very high at steady state and is down-regulated by IL-6 (45); in contrast the human C-reactive protein promoter, with two weak C/EBP sites, is poorly expressed at the steady state but is stimulated by IL-6 (10). Clearly, the specific composition of these two promoters and the participation of different non-C/EBP factors in their complex regulation can partially explain this apparent paradox. On the other hand, the specific transactivating capacities of the different C/EBP polypeptides are also likely to play an important role. C/EBP␣ is for example a stronger transcriptional activator than C/EBP␤ (16); moreover, both C/EBP␣ and C/EBP␤ mRNAs can give origin to truncated forms that can act either as inhibitors or as weak activators (20,46). In vivo, many different combinations in either homodimeric or heterodimeric forms are possible, giving rise to multiple polypeptides with distinct physiological properties whose differential occupancy of promoters may be dictated by their relative abundance and affinity and by proteinprotein interactions with other factors binding to adjacent sites.
A number of different transcription factors have been reported to be able to physically and functionally interact with C/EBP members and in particular with C/EBP␤. Among those, the interactions with members of the NF-B family of transcription factors (47,48) are particularly intriguing because they link the pathways of two major mediators of inflammation, IL-1 and IL-6. Interestingly, adjacent C/EBP and NF-B motifs are found in the promoters of many AP class I genes that require both IL-1 and IL-6 for their induction, as well as those of several cytokine genes, suggesting that cooperative interaction between the two families of transcription factors may represent a general mechanism of coordinating transcriptional responses to selected stimuli. Indeed, synergistic activation by C/EBP and NF-B members has been demonstrated for the genes encoding the acute phase proteins serum amyloid A1, A2, A3, and ␣ 1 -acid glycoprotein, as well as the cytokines IL-6, IL-8, and IL-12 and the G-CSF (11-13, 30, 31, 36, 49, 50). Cooperativity between the two families of transcription factors has also been demonstrated in the case of the human immunodeficiency virus long terminal repeat (43). C/EBP and NF-B interactions have also been shown to lead to antagonistic effects (47,51), again suggesting that promoter architecture and specific cell type are likely to play a major role. Although the mechanisms responsible for cooperative effects have not yet been entirely clarified, productive interaction requires the integrity of both the NF-B rel homology domain and the C/EBP leucine zipper motif (48). Increased affinity of C/EBP and NF-B for their respective sites has been demonstrated (43,47), and DNA-protein complexes containing both proteins have been detected using both NF-B and C/EBP sites (11,43,50).

A Lesson from Life
The availability of mouse strains in which the genes coding for different C/EBP members have been inactivated provided a model to test the relative importance of their different functions in the inflammatory pathway as well as in numerous other systems. Sometimes in apparent contrast with expectations it has certainly taught us a lesson about the dramatic oversimplification of extrapolating from a tissue culture plate to a living organism.
Acute Phase Genes-Analysis of AP mRNAs in C/EBP␤-deficient mice has shown that not all AP genes thought to require C/EBP␤ for their induction are in fact equally regulated by this factor (52); indeed, induction of both hemopexin and haptoglobin mRNAs was normal in the C/EBP␤Ϫ/Ϫ mice, whereas induction of serum amyloid A and P mRNAs was reduced and induction of C3 was totally impaired, at least at the protein level. The most prominent role of C/EBP␤ in the transcriptional regulation of the serum amyloid A, serum amyloid P, and ␣ 1 -acid glycoprotein genes appeared to be the maintenance of the induced state rather than the initial induction; after an initial accumulation comparable with that of the wild type mice, the mRNA levels for these three genes started to decrease in the C/EBP␤Ϫ/Ϫ mice, dropping at almost background levels by 24 h, the time point at which the wild type mice reached (in contrast) their peak. Because induction of both C/EBP␤ and C/EBP␦ mRNAs occurs relatively late after the inflammatory stimulus, 2 it is likely that transcription factors like NF-B and Stat3, whose activation is much faster but transient, are responsible for the first burst of induction, being partly replaced after a few hours by C/EBP members. In agreement with this idea, it has been recently demonstrated that Stat3 is involved in the IL-6-induced up-regulation of both C/EBP␦ and C/EBP␤ gene promoters (53). This suggests a sequential model in which the induction of inflammatory cytokines such as IL-1 and IL-6 in response to inflammatory stimuli would first trigger the activation of pre-existing, inactive forms of Stat3 and NF-B (illustrated in the upper part of Fig. 1,  early inflammation). These factors would in turn initiate the activation of both class I and class II genes although at the same time Stat3 would induce new synthesis of both C/EBP␦ and C/EBP␤. The preexisting population of C/EBP␤ molecules, activated by phosphorylation, would also participate in this initial induction of acute phase genes by interacting with NF-B. However, the main role played by C/EBP factors in the induction of AP genes is linked to the synthesis of new C/EBP␤ and ␦ polypeptides, which will substitute for the early factors NF-B and Stat3, thus allowing the activated status of AP genes to be maintained (lower part of Fig. 1, late inflammation). According to this model, C/EBP␦ is likely to act even at a later stage in the induction of AP genes, and its induction, which appears to be normal in C/EBP␤-deficient mice, 2 is likely to partially compensate for the absence of C/EBP␤. Late induction of AP genes should therefore be severely impaired in mice lacking both C/EBP␤ and -␦. This double mutant mouse strain has been recently generated (54), but no information is available about regulation of the AP response. C/EBP␣-deficient mice die shortly after birth (55), preventing an analysis of the liver AP response. Although liver-specific inactivation of the C/EBP␣ gene was recently described (56), no information is available on the regulation of the AP response in these mice.
Cytokine Genes-More puzzling are the results of the analysis of cytokine gene expression in the different C/EBP mutant mice. Induction of serum IL-6 was unchanged in C/EBP␤-deficient mice, whereas serum TNF-␣ induction was impaired, indirectly suggesting a more important role for the factor in the control of the TNF-␣ than of the IL-6 gene (52). Among all the cytokines thought to be regulated by C/EBP␤, only G-CSF mRNA induction is impaired in peritoneal macrophages from C/EBP␤Ϫ/Ϫ mice (57). Notably, IL-6 serum levels were elevated in aging C/EBP␤Ϫ/Ϫ mice (58,59), further suggesting that this factor is totally dispensable for IL-6 gene activity. No defects in cytokine production have been detected in macrophages from C/EBP␦-deficient mice. 3 Interestingly, although cytokine gene expression has not been analyzed in C/EBP␣-deficient mice, G-CSF receptor mRNA is almost undetectable in their liver (60). In contrast, inactivation of the gene encoding C/EBP⑀, a recently isolated family member specifically expressed in cells of the myelomonocytic lineages (61,62), results in decreased levels of several cytokine mRNAs (interferon-␥, TNF-␣, IL-2, IL-4, and IL-12 p40) in the spleen (63). Unfortunately, neither the inducibility of these cytokines nor the levels of IL-6, IL-1␤, G-CSF, granulocyte-macrophage CSF, or macrophage CSF have been analyzed yet in the C/EBP⑀ mice.
C/EBP Isoforms Coordinate the Differentiation and Function of Myelomonocytic Cells-Perhaps the most striking picture emerging from the analysis of different knock-out mice in the field of inflammation and immunity is that of a coordinated role of different C/EBP family members in regulating the differentiation and function of cells of the myelomonocytic lineage. Several lines of evidence suggested an important role of the various C/EBP isoforms in myelomonocytic cells. The expression of C/EBP␣, -␤, and -␦ is differentially regulated in myelomonocytic cell lines (64,65). In vivo, C/EBP␣ expression is relatively high in immature granulocytic cells but is down-regulated in most mature granulocytes (65), and C/EBP⑀ is preferentially expressed during granulocytic differentiation both in cell lines and in human primary CD34 ϩ cells (61,62). Moreover, as already mentioned, C/EBP family members can specifically transactivate the promoters of several myeloid-specific genes, and the chicken homologue of C/EBP␤, NF-M, is a myeloidspecific factor mediating eosinophils differentiation (66).
Remarkably in agreement with these findings, mice in which the genes encoding C/EBP␣, -␤, and -⑀ have been inactivated show specific defects in either macrophagic (C/EBP␤) or granulocytic (C/EBP␣ and C/EBP⑀) differentiation and/or functions (see below). In contrast, no abnormality was detected in C/EBP␦-deficient mice.
C/EBP␤ appears to play an important role in determining activation and/or terminal differentiation of macrophages. Indeed, C/EBP␤-deficient mice developed a lymphoproliferative disorder that can be linked to defects in macrophagic activation, as suggested by defective nitric oxide (NO) production by splenic macrophages and by the observation that no active IL-12, normally produced by activated macrophages, could be detected in the serum of the mutant mice infected with Candida albicans (58,59). Macrophages from C/EBP␤-deficient mice were also defective in intracellular killing of Listeria monocytogenes and displayed impaired tumoricidal and tumoristatic activity (57). The peritoneal macrophages used for these studies were able to produce normal amounts of NO, which is thought to play an important role in the elimination of intracellular bacteria and parasites (67), thus suggesting that a NO-independent, C/EBP␤-dependent pathway may be involved in Listeria killing and tumoricidal activity. Mainly as a result of these specific defects in macrophage functions, the C/EBP␤Ϫ/Ϫ mice are extremely susceptible to infections with microorganisms such as C. albicans and Listeria. No specific defects related to granulocytic differentiation or activation have been detected in these mice, although no direct tests have been performed.
In contrast, C/EBP␣ and C/EBP⑀ seem to play a crucial role in specific stages of granulocytic differentiation. C/EBP␣-deficient mice (55) were strikingly found to totally and selectively lack mature granulocytes (60), probably as a result of a specific maturational block of myeloid precursors toward mature neutrophils and eosinophils. This defect appears to be intrinsic to the hematopoietic precursor cells, and the loss of granulocytic maturation correlated with loss of expression of G-CSF receptor in the liver. However, the observation that mice in which the gene encoding G-CSF receptor has been inactivated still can form mature neutrophils (68) suggests that other still unidentified critical targets of C/EBP␣ must exist that make this factor essential for granulocytic differentiation. In agreement with these findings, it has been recently shown that conditional expression of C/EBP␣ in bipotential myeloid progenitors is sufficient to trigger neutrophilic differentiation (69). Unfortunately, analysis of the immune functions of the C/EBP␣Ϫ/Ϫ mice was not possible because the mutation caused perinatal death (55). In contrast to the complete maturational block toward granulocytic cells observed in the C/EBP␣Ϫ/Ϫ mice, mice deficient in C/EBP⑀ displayed only slightly reduced numbers of eosinophils and high numbers of granulocytes, which were, however, morphologically atypical and not fully functional, as suggested by a defective oxidative burst (63). In agreement with defective granulocytic function, these mice succumbed to opportunistic infections between 2 and 5 months of age, presenting various infection-related tissue lesions. In conclusion, although C/EBP␣ appears to act as a main differentiative switch toward the granulocytic lineage, the main function played by C/EBP⑀ in regulating granulocytic functions appears to be specular to the function of C/EBP␤ in macrophages, that is the specification of specialized functions and of terminal differentiation.
FIG. 1. Sequential model of induction for acute phase genes. Inflammatory stimuli induce the production of circulating cytokines (mainly IL-6, IL-1␤, and TNF-␣). Both TNF-␣ and IL-1␤ are known to trigger nuclear translocation of NF-B and activation of the Ras pathway, as well as transcriptional induction of C/EBP␤ and C/EBP␦ (whose mechanism is still unknown), whereas IL-6 mediates activation of Stat3 and of the Ras pathway, which ultimately leads to phosphorylation and activation of C/EBP␤. This wave of post-translational activation of pre-existing transcription factors triggers the early induction of responsive AP genes. At the same time activated Stat3 also induces transcription of the C/EBP␤ and -␦ genes. The newly synthesized C/EBPs bind in different combinations to the AP gene promoters, either functionally replacing the other factors or synergizing with them depending on promoter composition and duration of the stimulus, thus maintaining the induced state. This scheme only illustrates a model proposed for the coordinated role of the transcription factors NF-B, Stat3, and C/EBP␤ and -␦ in the induction of AP genes and is not meant as an exhaustive description of all possible activators of these factors. MAPK, mitogen-activated protein kinase; MEK, mitogen-activated protein kinase kinase; JAK, Janus family tyrosine kinase.

Conclusions
As has often been the case with gene-targeting experiments, the generation and analysis of single C/EBP knock-out mice, although confirming the importance of different family members in regulating various aspects of inflammation and immunity, have on the other hand revealed new layers of complexity that will require more systematic studies of gene expression in the different C/EBPdeficient strains and in multiple mutant strains lacking more than one family member. However, it has already clearly emerged that the relative role of different C/EBP isoforms in regulating the expression of distinct classes of genes may vary according not only to the cell type but also to the quality and quantity of the signal, being determined by specific promoter architecture and by interactions with other transcription factors. Moreover, because members of the C/EBP family do play important roles in inflammation by regulating the functions of both cytokine-producing effector cells (macrophages and granulocytes) and target cells (hepatocytes), the cell specificity of the observed phenotypes is not always clear. The tissue-specific inactivation of different C/EBP genes either in the liver on in different myelomonocytic lineages would allow a more precise dissection of their specific functions in each cell type. Finally, it appears that the main function of C/EBP factors in myelomonocytic cells is that of determining differentiation and expression of specialized functions, thus suggesting that C/EBP␣, -␤, and -⑀-deficient mice will be precious tools to study the regulation of myeloid cells differentiation at the transcriptional level.