Potential conflict of interest: Nothing to report.
Given the unknown timing of the onset of an acute systemic inflammation in humans, the fine tuning of cascades and pathways involved in the associated hepatocyte response cannot be appraised in vivo. Therefore, the authors used a genome-wide and kinetic analysis in the human Hep3B hepatoma cell line challenged with a conditioned medium from bacterial lipopolysaccharide-stimulated macrophages. A complete coverage of the liver transcriptome disclosed 648 mRNAs whose change in abundance allowed for their clustering in mRNA subsets with an early, intermediate, or late regulation. The contribution of transcription, stability, or translation was appraised with genome-wide studies of the changes in nuclear primary transcripts, mRNA decay, or polysome-associated mRNAs. A predominance of mRNAs with decreased stability and the fact that translation alone controls a significant number of acute phase–associated proteins are prominent findings. Transcription and stability act independently or, more rarely, cooperate or even counteract in a gene-by-gene manner, which results in a unidirectional change in mRNA abundance. Waves of mRNAs for groups of functionally related proteins are up- or downregulated in an ordered fashion. This includes an early regulation of transcription-associated proteins, an intermediate repression of detoxication and metabolism proteins, and finally an enhanced translation and transport of a number of membranous or secreted proteins along with an enhanced protein degradation. In conclusion, this study provides a comprehensive and simultaneous overview of events in the human hepatocyte during the inflammatory acute phase. Supplementary material for this article can be found on the HEPATOLOGYwebsite (http://www.interscience.wiley.com/jpages/0270-9139/suppmat/index.html). (HEPATOLOGY 2005;42:946–955.)
The acute phase (AP) of the inflammatory response is coordinated by a large number of mediators, such as the pro-inflammatory cytokines tumor necrosis factor (TNF)-α and interleukin (IL)-1β, mainly produced by macrophages.1 TNF-α and IL-1β then promote a second wave of cytokines, such as IL-6, mostly released by macrophages and fibroblasts.1, 2 IL-6 is a dual, pro- and anti-inflammatory cytokine. It amplifies the response of various organs to AP while it downregulates the production of TNF-α, thereby facilitating the so-called resolution phase and a return to homeostasis.1–4
Altered gene regulation in the liver is a hallmark of AP.1, 2 Specifically, the AP regulates many liver-expressed genes involved in innate immunity and coding for AP-regulated intracellular proteins (APRIPs) and plasma acute phase proteins (APPs), which are transiently up- or downregulated and consequently classified as positive or negative APRIPs/APPs.5–8 These regulations entail transcriptional or post-transcriptional step(s), which results in altered mRNA abundance, stability, or translation,6, 9 but the relative importance of these controls remains to be assessed. Transcriptome studies in rodents have partly dissected the elaborate and dynamic process that takes place in the liver in the course of AP.10, 11 In contrast, a genome-wide and kinetic view of the AP-induced changes in the human liver is not available. Owing to a coverage of the whole human liver transcriptome with a dedicated microarray, we recently identified the APRIP and APP mRNAs whose altered abundance best correlates with the extent of an acute, systemic inflammation in vivo.12 However, deciphering the complete regulatory steps of signaling cascades and pathways involved in the response of the human liver to inflammation cannot be gained from studies in vivo that lack essential information such as the time of AP onset. Therefore, an analysis of the kinetics of transcriptome alteration in the human hepatocyte challenged the pro-inflammatory cytokines in vitro should provide a privileged window on the liver response to AP. With this approach, we have now observed that several waves of mRNAs for groups of functionally related proteins are up- or downregulated in an ordered fashion. Analysis of mRNA transcription, stability, and translation further indicated that in most instances these control steps act independently or, more rarely, cooperate or even counteract in a gene-by-gene manner, which still results in a unidirectional change in mRNA abundance.
Stimulation of a Hepatoma Cell Line With Pro-inflammatory Cytokines.
The human Hep3B hepatoma cells (ATCC HB-8064) plated at 33% confluence were cultured for 48 hours, and the culture medium was next changed for a mixture made of a serum-free medium added with 20% (vol/vol) stimulated-macrophage conditioned medium (CM) enriched in TNF-α, IL-1β, IL-6, and IL-8 or nonconditioned medium (NCM) used as a control.12 Paired cultures were challenged with CM versus NCM for a given length of time in 3 (time-course) or 2 (stability, transcription, or translation) independent experiments.
Determination of mRNA Abundance by Microarray or Polymerase Chain Reaction.
Total RNAs were labeled and hybridized to our “Liverpool” microarray, which provides complete coverage of the human liver transcriptome (approximately 10,000 genes).12 Quantitative reverse transcription polymerase chain reaction (q-RT-PCR) of mRNAs with the primers listed in our Supplementary Table 1 was done as described.12
Analysis of mRNA Stability by Actinomycin D and Microarray.
The cells were stimulated with CM or NCM for a fixed time. The medium was replaced by serum-free medium containing 10 μg/mL actinomycin D (Sigma, St. Louis, MO), the dishes were kept at 37°C for 0, 15, 60, or 240 minutes (1 dish per time) and the resulting RNAs were labeled and hybridized to our microarray. Every mRNA 3′ untranslated region (3′UTR) was retrieved from the ENSEMBL data library, and a search for AU-rich elements (AREs) was made with a series of 30 published AREs,13, 14 owing to a locally developed PERL script.
Analysis of RNA Transcription by Run-on and Microarray.
Nuclear primary transcripts labeled with [α-32P]UTP by run-on assay were prepared as described by in Daveau et al.15 and hybridized to our microarray for 64 hours followed by autoradiography for 1 week.
Analysis of mRNA Translation by Polysome Isolation and Microarray.
Polysome fractionation was done essentially as described elsewhere.16–18 The polysome-free or polysome-enriched fractions were pooled separately and then labeled and hybridized to our microarray.
Microarray Data Handling and Mining.
Our general procedures for data handling were detailed previously.12 For every detected mRNA, the normalized paired values obtained under NCM versus CM at a given time were considered to be significantly induced or repressed (folds) when their difference was outside a funnel-shaped confidence interval (P < .05) calculated from every mRNA detected within the experiment.12 In time-course experiments, an mRNA abundance was considered to be CM-regulated at a given time point whenever a significant induction or repression occurred in at least 2 of 3 independent experiments. For mRNA stability, run-on, or polysome-related experiments, the mean values measured at a given time were used for determination of confidence intervals, from which outlier transcripts were considered to be significantly regulated (further specific calculations are provided in the legends of the tables and figures). K-means clustering was done with the Genesis software.19 Protein functions and groups of functionally related mRNAs were based on the Gene Ontology Consortium.20
Protein Electrophoresis and Immunodetection.
SDS-PAGE and immunodetection were performed as described.21 Goat antibodies against monoamine oxidase B (MAO-B) (catalogue ref. sc-18401) or calmodulin 1 (CAM-1) (sc-1989) and mouse antibodies against RNA polymerase II (DNA-directed) polypeptide F (POLR-2F) (sc-21752) were from Santa Cruz Biotechnology (Santa Cruz, CA). Alexa Fluor 680–labeled rabbit anti-goat or anti-mouse IgGs used as a secondary antibody were from Molecular Probes (Eugene, OR). Fluorescent protein bands were quantified with the Odyssey Imaging System from Li-Cor (Lincoln, NE).
Kinetics of Cytokine-Induced Changes in mRNA Abundance.
A time-course (0, 15, 30 minutes; 1, 3, 6, or 16 hours) of mRNA abundance changes was studied in the human Hep3B hepatoma cell line challenged with a pro-inflammatory cytokine-enriched CM versus control NCM. Our “Liverpool” microarray was used to identify every mRNA whose abundance exhibited a statistically significant difference under CM challenge at one or more given time points, which resulted in a selection of 648 such mRNAs, referred to as the Hep3B/CM mRNAs. To identify subsets of mRNAs with a similar timewise regulation of abundance, these Hep3B/CM mRNAs were next separated into clusters by k-means clustering. The latter is an unsupervised procedure that requires the number of clusters to be chosen beforehand.19 We found that 11 clusters C1 to C11 conveyed appropriate information; all but 1 (C11) presented a typical up or down and time-dependent pattern (Fig. 1A). The complete list of mRNAs within each cluster is provided as Supplementary Table 2. C1 to C10 correspond to a down- (C1–C3) or upregulated abundance (C4–C10). Moreover, an early (<1 hour) change in C1, C4, and C5, an intermediate (1–3 hours) change in C2, C6, and C7, and a late (>6 hours) change in C3 and C8–C10 point to early or late CM-responsive genes. The mRNAs with an increased abundance predominated within the entire subset of early genes C1, 4, 5 (149 of 202 mRNAs, 73.8%). This feature was also found, albeit to a lower extent, in the two subsets of intermediate or late genes C2, 6, 7 (64.6%) and C3, 8, 9, 10 (57.2%) (early vs. late genes: P < 10−3 by chi-square test). C11 contained mRNAs whose abundance poorly correlated with time. As a negative control, a cluster C12 made with 50 mRNAs randomly taken from all those that did not exhibit any change in this study provided a flat curve.
As an external control for the above selection, one mRNA taken from every cluster C1 to C10 was randomly tested by q-RT-PCR. In all instances, the kinetics of abundance as found by microarray or q-RT-PCR were quite similar (Fig. 2). Moreover, for most of these mRNAs, the direction and kinetics of change in abundance were quite similar in CM-challenged Hep3B or HepG2 hepatoma cells (Supplementary Fig. 1), which makes a cell line–specific effect unlikely.
Abundance of Functionally Identified mRNA Subpopulations and Their Kinetics.
We first verified that our series of 648 Hep3B/CM mRNAs fit a pro-inflammatory cytokine-induced regulation. In Fig. 3, many Hep3B/CM mRNAs that code for (1) critical proteins of the major cytokine-driven cascades or (2) other proteins in the hepatocyte under acute inflammation2, 6–8, 11, 12, 21–26 were regulated as expected.
The mRNAs that code for proteins involved in (1) the immune response at large or (2) the inflammatory response were further identified by an ontology approach.20 When comparing the numbers of such mRNAs found in either the Hep3B/CM mRNA population (see details in Supplementary Table 2) or in the entire population of mRNAs in the quiescent liver,12 both functional groups were significantly enriched in the former (in both instances, P < 10−4 by chi-square test). Again, this demonstrates that our selection of the Hep3B/CM mRNAs fits with a major influence of pro-inflammatory cytokines on mRNA abundance.
We searched for a time-dependent regulation of other, functionally defined mRNA subpopulations in CM-challenged Hep3B cells. By ontology, we identified 13 major subpopulations of mRNAs corresponding to proteins with a well-identified function (Supplementary Table 2). Among the clusters C1 to C11, 7 clusters contained at least 1 significantly under- or over-represented mRNA subpopulation (Supplementary Table 3). A further comprehensive analysis is presented in Fig. 1B and Supplementary Fig. 2. Striking observations include (1) a trend to transcriptional upregulation (boxes C4–C5 and C6–C7) and translational repression (C1) at the early-to-intermediate phase of the cell response, (2) a repression of detoxication and hepatic metabolism in the intermediate phase (C2), and (3) an upregulated synthesis and transport of membranous or secreted proteins as well as increased protein degradation in the late phase (box C8–C10).
Stability-Dependent mRNA Abundance.
We examined to which extent the stability of our set of 648 Hep3B/CM mRNAs was affected after a CM-versus-NCM challenge for either 30 minutes or 16 hours. The abundance of every mRNA was determined at various times after transcription arrest by actinomycin D. Genome-wide determinations of RNA stability still await standardized analysis.14 Therefore, our specific calculations are summarized in Supplementary Table 4. After 30 minutes of CM challenge, 2 Gaussian populations of mRNAs were observed, as shown in Fig. 4A, left. One population (mean slope value = 0) had a narrow variation of stability that was considered unchanged by CM, and this was used to calculate a normal range of values (horizontal thick bar in Fig. 4A). Within the other population, the mRNAs had highly variable slope values, which significantly departed (P < .05) from the normal range. Quite similar data were obtained after a CM challenge for 16 hours (Fig. 4A, right). Altogether, 218 Hep3B/CM mRNAs did not present any variation of stability, whereas the other 430 Hep3B/CM mRNAs exhibited a CM-induced change of stability (the latter mRNAs are noted as such in Supplementary Table 2). As seen in Fig. 4A, the mRNAs with a decreased stability (negative slope) predominated (68.4% of all 430 mRNAs) as compared with those with an enhanced stability (31.6%) after 30 minutes as well as 16 hours of CM challenge. As shown in Fig. 4B, the proportion of mRNAs with an enhanced or decreased stability was not significantly different in box C1–C3 (decreased abundance) compared with box C4–C10 (increased abundance), thus ruling out that altered stability alone accounted for an up- or downregulated mRNA abundance. However, the relative distribution of mRNAs with altered stability in the clusters C1 to C10 was not random. The number of mRNAs with enhanced stability was significantly higher, and the number of unstable mRNAs was significantly lower, in box C8–C10 (clusters of increased abundance) as compared with C3 (decreased abundance) (stars in Fig. 4B). This stability/abundance relationship is logical and validates our experimental approach. Moreover, this relationship increased timewise from box C4C5 up to C8–C10, whereas it was not observed when comparing C1, C2, and C3. Taken together, our data indicate that (1) a loss of stability controls, at least partly, the abundance of many AP mRNAs and (2) stability enhancement occurs infrequently and mostly acts on the late mRNAs.
Because mRNA 3′ UTR may be involved in stability,14 we investigated whether the stability-regulated mRNAs identified above exhibited some characteristic physical features. As shown in Supplementary Table 5, the 3′ UTR of the mRNAs with modified stability was significantly shorter than that of control mRNAs, which fits earlier observations made at the level of absolute decay rate.14 Moreover, the frequency of occurrence of 3 AREs that are known to be associated with a change in mRNA stability13, 14 was significantly lower in regulated than in control mRNAs. These data support our identification of mRNA populations with a modified or unchanged stability post-CM challenge.
Transcriptional Control of mRNA Abundance.
We also examined to what extent the variations in abundance resulted from a change in transcription in our set of 648 Hep3B/CM mRNAs after 30 minutes or 16 hours of CM challenge. Comparing the relative abundance of primary transcripts in nuclei from CM- versus NCM-challenged cells identified 191 or 66 transcripts that were significantly regulated at 30 minutes or 16 hours, with very little overlap (Fig. 5A). They are so noted in Supplementary Table 2. Remarkably, the downregulated primary transcripts predominated at 30 minutes (143 of 191, 74.8%), but they were a minority at 16 hours (21 of 66, 31.8%) (P < 10−4 by chi-square test). This indicates that transcriptional repression predominates during the early AP, whereas transcriptional activation predominates at a later stage. As shown in Fig. 5B, the fraction of primary transcripts indicative of a decreased or enhanced transcription were not significantly different in the mRNAs of decreased abundance (box C1–C3) versus those of increased abundance (box C4–C10), thus ruling out that transcription alone accounted for an up- or downregulated mRNA abundance. However, the number of upregulated primary transcripts appeared to be higher and the number of downregulated transcripts lower in box C8–C10 (mRNAs with a late increase in abundance) as compared with C3 (late decrease), although this was not significant because of the small sample size. This transcriptional activity/mRNA abundance relationship is logical and supports our run-on analysis. Taken together, our data indicate that (1) altered transcription controls, at least partly, the abundance of many AP mRNAs, and (2) increased transcription mostly affects the late AP genes.
Within the mRNAs whose transcription and stability both were found to be altered in this study, those with opposite regulations of transcription and stability were as numerous as those with 2 up- or downregulations (P > .5, not detailed), and the former were evenly distributed in all clusters. Therefore, additive or subtractive effects of transcription and stability have similar occurrence. Moreover, within the subset of mRNAs undergoing 2 up- or 2 downregulated transcription and stability, the extents of both parameters did correlate (P < .05, not detailed), suggesting cooperation. On the contrary, these parameters were anti-correlated within the subset of mRNAs with opposite transcription and stability (P < .05, not detailed), and hence transcription and stability may antagonize in an imbalanced fashion, which still results in unidirectional change in mRNA abundance. None of the functionally defined subpopulations previously noted in Fig. 1B and Supplementary Fig. 2 appeared to be preferentially associated with any transcription/stability combination (not detailed).
Cytokine-Induced Changes in the Polysome Fraction of mRNAs.
Because changes in mRNA and protein levels do not necessarily correlate,27 translation of a given mRNA could be CM-modulated, regardless of whether its abundance was altered. Therefore, and regardless of the 648 Hep3B/CM mRNAs listed, we searched for a CM-associated change in the ratio of (polysome-bound molecules/[free + monosome]-bound molecules) for every mRNA that was detectable in the Hep3B cells. As above, this was carried out at 30 minutes or 16 hours of CM challenge. The identification of mRNAs whose relative abundance in these 2 populations was most significantly (P < .05) modified by a CM challenge is illustrated in Supplementary Fig. 3. These mRNAs were considered to be engaged in a CM-dependent (up- or downregulated) change of translation extent. This resulted in a final selection of 34 or 40 such mRNAs at 30 minutes or 16 hours of CM challenge, respectively. At either time, approximately half of the mRNAs had undergone an increased translation, whereas translation of the remaining mRNAs was decreased (nonsignificant difference by chi-square test), thus indicating that the translational control of protein production in AP is bidirectional. The complete list of these mRNAs (Supplementary Table 6) shows very little overlap between the mRNAs whose translation is regulated at 30 minutes or 16 hours. In fact, no correlation existed between the extent of translation observed at 30 minutes and 16 hours for the 34 (r = 0.002, P = .99) or 40 mRNAs identified previously (r = 0.13, P = .41). This observation points to a strongly time-dependent control of translation for most AP-relevant proteins. Remarkably, the mRNAs with an altered translation included (1) actors of the inflammatory response, at 30 minutes (RAB family members) or 16 hours (e.g., metallothionein 1H, leukemia inhibitory factor, MAP kinase kinase-1, TNF receptor superfamily member 11a), (2) prominent actors of protein degradation (ubiquitin protein ligase E3A and proteasome subunit β6, upregulated at 30 minutes), and (3) actors of translation (ribosomal proteins L41 and S23) whose upregulation at 16 hours suggested a positive feedback loop for a late enhancement of translation.
As noted in Supplementary Table 6 (last column), 8 mRNAs regulated in translation were previously found to be regulated in abundance, and with only one exception (cytokine-like nuclear factor n-pac) both regulation levels acted in the same direction. Strikingly, the 2 mRNAs with the most tightly up- or downregulated translation at 30 minutes of CM challenge were also regulated in abundance (RNA polymerase II polypeptide F; RAB18). Likewise, several mRNAs with a highly upregulated translation at 16 hours of CM challenge were also upregulated in abundance. In these situations, mRNAs belonging to the late clusters of upregulated abundance (C9, C10) predominated. Taken together, our data suggest that in a limited number of cases, shifts in mRNA abundance and translation can cooperate for an upregulated protein synthesis during the late AP, notably when a strong translational control takes place.
Protein electrophoresis and immunodetection were used as a control for translationally regulated mRNAs. Three mRNAs with a time- and CM-dependent shift in translation, namely, POLR-2F, CAM-1, and MAO-B, were selected. As shown in Supplementary Fig. 4, the abundance of the PolR-2F protein increased after a 30-minute CM challenge, whereas that of CAM-1 or MAO-B decreased after a 30-minute or 16-hour CM challenge, respectively. These data cannot be accounted for by a concomitant change in mRNA abundance (lower diagrams in Supplementary Fig. 4), and hence they support our identification of the mRNAs with a CM-dependent translation.
The percentage of regulated mRNAs found in this study (approximately 7%) is quite similar to the number of liver mRNAs regulated during the AP in human or mouse in vivo.11, 12 Most importantly, the proper direction and kinetics of changes in mRNA abundances for cytokines and their receptors, transcription factors (TFs), and APPs all support our choice of the Hep3B cells. Our further data (not shown) did not indicate any significant modulation of mRNAs for the suppressors of IL-6 signaling that are involved in the resolution phase of inflammation.3, 4 This is consistent with a sustained CM stimulation along the time course of this study, which likely prevented the target Hep3B cells from returning to a quiescent state. Moreover, the discrete waves of stimulation by TNFα and IL-1β, then IL-6, which occur in vivo,1, 2 could hardly be mimicked timewise in vitro, which may have prevented the resolution phase from occurring. Finally, we preferred not to study the effect of a single proinflammatory cytokine. For instance, IL-1β is known to promote an opposite effect on some APPs, depending on whether it acts in the context of other cytokines such as IL-6.1, 2, 6
The over- or under-representation of functional groups along the AP time course discloses an early control of TF-encoding mRNAs, which subsequently results in an up- or downregulated production of other mRNAs for proteins with mostly hepatocyte-specific functions. Downregulation of enzymes involved in detoxication and metabolism (mostly in C2) represents a noticeable example, which includes several alcohol dehydrogenases and glutathione transferases, as well as key enzymes for metabolism of glucose (glucose-6-phosphatase), fatty acids (stearoyl-CoA desaturase) or cholesterol (24-dehydrocholesterol reductase). This downregulation can be at least partly accounted for by the concomitant downregulation of HNF-4 and upregulation of STAT-3, because the former is an activator of hepatocyte-specific metabolism at large,28 and the latter is a repressor of the glucose-6-phosphatase–dependent gluconeogenesis.29 This transient downregulation takes place in a highly time-sensitive manner, as it disappeared during the late phase of cytokine challenge in this study, despite the possible lack of a resolution phase as discussed. An enhanced production of some APPs in the AP could benefit from amino acid saving resulting from a decreased synthesis of other proteins.2 We now demonstrate that the transient downregulation of detoxication and metabolism takes place before the increased synthesis of a bulk of APPs, which argues for a participation of some transiently dispensable detoxication and metabolism proteins in such an amino acid–saving scenario.
Contrary to other reports in which stability was merely inferred from combined transcriptional rate and mRNA abundance,30, 31 we actually measured the CM-associated change of mRNA decay. We have now found that two thirds of the Hep3B/CM mRNAs are regulated by stability. This is in keeping with the change of abundance noticed for mRNAs coding for several heterogeneous nuclear ribonucleoproteins (hnRNP), and particularly hnRNP D (cluster C5), which directly controls mRNA stability.32 Predominance of mRNAs with a loss of stability is a further novel finding of our study. It should not be seen as a standard response to stress, given that a shift toward stabilization has been found in other examples of cellular stress.30, 31 The current loss of stability likely results, at least in part, from the early repression of the MAP kinase-2 mRNA (cluster C1) that limits mRNA decay.33 It also appears to be driven by AREs, given the lower frequency of AREs found in the 3′UTR of regulated mRNAs versus controls in our study. AREs participate in decay control but they do not allow prediction of the direction and extent thereof.14, 33 Finally, the predominant loss of stability seen in our AP-regulated mRNAs is consistent with a requirement for (1) a transient downregulation of some mRNAs controlling normal functions in the quiescent hepatocyte (e.g., metabolism) and (2) a short-term limitation of some AP-induced mRNAs whose sustained presence could be detrimental.
Analysis of nuclear transcripts by run-on has seldom been made on a genome-wide scale.30 We developed this approach in the hepatocyte/AP context and, as expected,6, 9 we observed that transcription controls a number of APRIP/APP genes. The overall trend of this control step is time-dependent, as transcriptional repression or activation predominates at an early or late stage of AP, respectively. This feature has not previously been documented. It fits the up- or downregulated abundance of many TF-encoding mRNAs, and kinetics thereof, within our set of Hep3B/CM mRNAs. Not only can such an early decrease of given TFs directly account for a subsequent limitation of other proteins (e.g., the direct relationship between HNF-4 and metabolism-related proteins) but it also can allow for a subsequent upregulation of other TFs. For instance, GRIM-19, a STAT-3 inhibitor whose mRNA decreases early, allows for (1) an immediate upregulation of STAT-3 activity and (2) a late upregulation of STAT-3 targets, such as the C/EBPβ gene.6
Within every cluster C1 to C10, some mRNAs exhibited a change in either stability or transcription that was opposed to their final change in abundance. This feature fits with other reports of opposite regulations of mRNA transcription, stability, or translation in various contexts.14, 30, 31 This now underscores that in the AP-challenged hepatocyte transcription and stability can either cooperate or antagonize and still result in an unidirectional change of abundance. Potentially conflicting data appeared in the early phase of our kinetics, as we found (1) a predominance of mRNAs with an increased abundance, along with (2) predominant numbers of transcriptionally repressed mRNAs and mRNAs with a decreased stability. The fact that two counteracting regulations of transcription and stability often occur and still result in a unidirectional change of abundance clarifies, at least partly, this conflict.
This study provides a genome-wide and kinetic view of the human hepatocyte response to pro-inflammatory cytokines, from transcription to translation. Because the data obtained by our various approaches cover quite different scales, trends rather than absolute figures should be compared. Such trends shown in Fig. 6 (left 2 panels) include an overall predominance of mRNAs with a CM-induced decrease in stability, and an increased transcription and stability of the mRNAs whose abundance is upregulated late in the course, along with reversed regulations of the mRNAs with a downregulated abundance. Among the latter, those coding for elements of the translational machinery are repressed early, whereas translation is re-activated later (right 2 panels), which fits with a timewise increase in the number of mRNAs that are unaltered in abundance but translationally upregulated (center 2 panels). Therefore, translation represents a critical control for APRIP/APP production. In extreme cases of upregulation, mRNA abundance and translation cooperate (curved, dotted arrow between panels). Our overall view fits with: (1) an engagement of latent proteins and translation of a limited number of latent mRNAs as the primary events at the onset of the hepatocyte response, the latter also including a repressed translation or active degradation of other pre-existing proteins (e.g., IκB) and (2) a regulation of gene activity and protein synthesis that reaches its full extent at a later stage and allows for the full cell response to take place. The latter notably includes a repression of detoxication and metabolism, an enhanced translation and transport of a number of membranous or secreted proteins, as well as an enhanced protein degradation, which may in turn limit the production of upregulated but potentially harmful proteins34 in a time-dependent manner.