Increased proliferation of hepatic periportal ductal progenitor cells contributes to persistent hypermetabolism after trauma

Abstract Prolonged and persistent hypermetabolism and excessive inflammatory response after severe trauma is detrimental and associated with poor outcome. The predisposing pathology or signals mediating this complex response are essentially unknown. As the liver is the central organ mediating the systemic metabolic responses and considering that adult hepatic stem cells are on top of the hierarchy of cell differentiation and may pass epigenetic information to their progeny, we asked whether liver progenitor cells are activated, signal hypermetabolism upon post‐traumatic cellular stress responses, and pass this to differentiated progeny. We generated Sox9CreERT2: ROSA26 EYFP mice to lineage‐trace the periportal ductal progenitor cells (PDPCs) and verify the fate of these cells post‐burn. We observed increased proliferation of PDPCs and their progeny peaking around two weeks post‐burn, concomitant with the hepatomegaly and the cellular stress responses. We then sorted out PDPCs, PDPC‐derived hepatocytes and mature hepatocytes, compared their transcriptome and showed that PDPCs and their progeny present a significant up‐regulation in signalling pathways associated with inflammation and metabolic activation, contributing to persistent hypermetabolic and hyper‐inflammatory state. Furthermore, concomitant down‐regulation of LXR signalling in PDPCs and their progeny implicates the therapeutic potential of early and short‐term administration of LXR agonists in ameliorating such persistent hypermetabolism.

post-burn pathology. 7 Considering the ability and plasticity of continuous self-regeneration of the liver, 8 we speculated and sought to test that pathological changes in hepatocytes' proliferation and liver regeneration under stress conditions contribute to such prolonged hyper-inflammatory and hypermetabolic states.
Although it has been well demonstrated that the liver has the capacity to regenerate and up to 2/3 of the loss of the liver parenchyma can be recovered by regeneration without jeopardizing the viability of the entire organism, 9,10 there are still controversies on how such a regeneration happens including whether there is single or multiple sources of stem cells, what the triggers of the liver regeneration are, and how the liver regeneration is regulated. [11][12][13] As the portal triads are where the facultative regeneration of hepatic parenchyma occurs under liver damage and stress conditions, 14 in line with the existing streaming liver theory 15,16 that the regeneration and maturation of hepatocytes start from the portal venule, proceed across the liver plates and end with clearance in the central venule, we suggested that liver regeneration under profound stress condition would be dominated by proliferation and differentiation of periportal ductal progenitor cells (PDPC) which are bi-potential progenitor cells that can give rise to either hepatocytes or cholangiocytes, 17 whereas liver regeneration under physiological or mild stressful conditions was dominated by self-duplication of mature hepatocytes. 12 We further speculated that those hepatocytes regenerated under significant stress conditions after major burn injury might possess aberrant and persistent inflammatory and/or hypermetabolic profiles and thus contribute to prolonged pro-inflammatory states and hypermetabolism that are commonly seen in major burned patients. 5,6 2 | MATERIAL S AND ME THODS  Table S1.

| Animal studies
Eight-to eleven-week-old male mice with the genetic background of Sox9-cre/ERT2 +/− : ROSA26 EYFP +/+ were included for the animal experiments. Tamoxifen (Sigma) was dissolved at 20 mg/mL in corn oil (Sigma) and administered subcutaneously at a dosage of 100 mg/kg body weight. Tamoxifen was administered once daily for 3 consecutive days. Wild-type mice of the same age and non-tamoxifen control were also kept for baseline determination. The mice were randomly divided into sham and burned groups and received 30% total body surface area (TBSA) scald burn 18 or sham treatment immediately after the first injection of Tamoxifen. The mice were killed on post-burn day 2, 7, 14, 21, 28 and 42 (referred to as different observational groups). N = 6 for each group including sham control. Mice killed on post-burn day 2 received 2 doses of tamoxifen injection.

| Liver tissue collection and digestion
Upon killing, the inferior vena cava was cut and the whole liver was collected after brief portal vein perfusion with PBS (2 mL). The liver was weighed, and 2 small pieces of liver were taken and frozen immediately on dry ice and then stored at −80°C for gene expression and Western blot analyses. Another piece of liver tissue was fixed in 10% buffered formalin at 4°C overnight, transferred to 70% ethanol and then paraffin embedded for histology. The rest of the liver tissue was chopped into fine particles less than 1 mm 3 and transferred to 5 mL digestion cocktail (200 U dispase (Sigma, Cat#4693), 270 mg Type I collagenase (Life Technologies, Cat#07912) in 100 mL DMEM with 1% Ab/Am) for cell staining and flow cytometry analysis and cell sorting.

| Western blotting
Liver homogenate lysates (50 μg of protein per well) were separated by 10% SDS-PAGE gel, proteins were transferred to nitrocellulose membrane as previously described, 19 and then, blots were probed using the antibodies listed above. Band intensities were detected, normalized and quantified with the Chemidoc and Image Lab 5.0 software (Bio-Rad Laboratories). GAPDH was used as a loading control.

| Cell staining and flow cytometry
Cells were incubated in digestion cocktail in 37°C for 40 minutes, then added equal volume of DMEM with 10% FBS and filtered through 40 μm strainer to a new tube. The cells were then washed with FCM buffer (0.5% BSA in HBSS). Cell count was performed with trypan blue using TC20TM automatic cell counter (Bio-Rad Laboratories). 5 million cells (viability is between 30%-50%) were transferred to flow tube and washed with FCM buffer. The cells were then blocked with CD16/32 on ice for 10 minutes followed by incubation with antibodies against EpCAM in FCM buffer on ice for 30 minutes. The sample was then washed once with

| Immunofluorescent multi-channel staining of liver
Antibody staining was performed as described previously. 19,20 Primary antibodies were the same as in Western blotting.

| Microarray transcriptomic analysis
Sorted cells were centrifuged, the pellet was dissolved in TRIzol

| Statistical analysis
The statistical analysis was performed using Prism version 5.01 (GraphPad Software). One-way ANOVA with Bonferroni's multiple comparison test was used unless otherwise specified, and P < .05 was considered statistically significant.

| Increased proliferation of PDPCs contributes to hepatomegaly after major burn injury
To specifically lineage-trace the proliferation of the PDPCs in the liver after thermal injury, we generated Sox9-cre/ERT2:ROSA26 EYFP mice in which the expression of EYFP in PDPCs is inducible To test whether such proliferation of PDPCs contributes to the increase in the hepatic parenchyma, we measured the weight of the whole liver of the mice and compared with the body weight of the mice when killed. As the body weight of the mice was generally stable during the whole observation period ( Figure S2), the concomitant significant increase in liver/body weight ratio around PBD14 and 21 indicated hepatomegaly in this period of time ( Figure 1E).
We also performed immunofluorescent staining of the liver tissue sections with anti-GFP antibody to examine the distribution of the EYFP + cells in the liver tissue (Figure 2A Interestingly, while the proliferation of PDPCs peaked around PBD14 and significantly attenuated afterwards ( Figure 1C), the increase in the liver mass peaked around PBD21 ( Figure 1E). Also, we were unable to see the pattern of streaming of liver regeneration from portal triads to central venule systems (Figure 2A). These results suggest that the liver regeneration after major burn injury is not from a single type of progenitor cells but via an orchestrated proliferation of both PDPCs and mature hepatocytes.

| The hepatic cellular stress response is implicated in signalling the proliferation of PDPCs after major burn injury
In our previous studies, we have demonstrated the augmented hepatic ER stress and subsequent increase of apoptosis in the liver in the early post-burn period in several rodent thermal injury models. 23,24 In the current study, we examined the expression of multiple cellular stress markers including phospho-IRE1α, CHOP, phospho-eIF2α versus eIF2α, ATF4, BiP and HSP90 in the liver tissue by Western blotting ( Figure 3A) of whole liver lysate samples and densitometry analysis showed that significant hepatic stress response occurred and culminated between PBD2 to PBD7, persisted to PBD21 and resolved after PBD28 to almost normal at

| Transcriptomic analysis revealed enhanced pro-inflammatory signalling and hypermetabolism in PDPC and their progeny after burn injury
We next asked whether and how the increased number of the PDPCs contributes to the persistent pro-inflammatory response down-regulated signalling pathways in the EYFP + cells on PBD7 as compared with that of sham (Table S2). When compared the EYFP + cells with the EYFPcells on PBD7, we found 52 up-regulated signalling pathways and 12 down-regulated signalling pathways (Table S3).

| Persistent hepatic pro-inflammatory response and hypermetabolism, as well as implicated increased turnover of liver parenchyma after major thermal injury
Specifically, the level of hepatic LXRα significantly decreased between PBD2 to PBD14 with the concomitant presence of the cellular stress response ( Figure 3) and then significantly increased around PBD21 and PBD28 along with the resolving of the inflammation and the restoration of homeostasis after the major burn injury ( Figure 6B). We observed rapid and significant increase in the expression of hepatic pro-IL-1β from PBD2 to PBD21, resolving to the level of sham animals on PBD28 ( Figure 6C). There seemed to be delayed and more temporal increase in the level of hepatic matured IL-1β which peaked around PBD14 ( Figure 6D). Similarly, the level of the hepatic phospho-p38 MAPK versus total p38 MAPK increased significantly around PBD7 to PBD14 ( Figure 6E) which, together with the changes in the expression of hepatic LXRα and IL-1β, implicated the activation of the immunological responses and inflammatory signalling between PBD2 to PBD14 or 21. Moreover, we measured the level of expression of CPT1A, which is the rate-limiting regulator of hepatic β-oxidation, 26 and it was demonstrated that there was increased β-oxidation from PBD2 to PBD21, reflecting the increased energy demand and expenditure after burn injury ( Figure 6F). We also examined the expression level

| D ISCUSS I ON
In the current study, using the reporter mice strain of Sox9-cre/ ERT2:ROSA26 EYFP, we were able to lineage-trace the prolifera- It has been well accepted that severe liver damage and the impairment of the renewal of hepatic parenchyma by self-duplication of mature hepatocytes trigger the proliferation of the PDPCs for the liver regeneration. 21 Here, we have demonstrated that 30%TBSA scald burn is an insult strong enough to activate PDPCs proliferation.
It has also been suggested that ER stress and subsequent activation of unfolded protein response (UPR) enable the cell to either resolve stress or initiate apoptosis and while primary stem cells are more prone to cell death under stress, closely related progenitor cells exhibit more adaptive response leading to their survival. 27  with the EYFP + cell population over 20% of the total hepatocytes count. The drastic increase in the EYFP + cells seen on PBD7 and PBD14 is concomitant with the increase in the liver/body weight ratio. However, the EYFP + cell population decreased from over 20% to around 15% of the total hepatic parenchyma from PBD14 to PBD21 and the liver/body weight ratio peaks around PBD21 within the context of stable or slightly increased body weight. This indicates that cells other than PDPCs also contribute to the increased liver mass from PBD14 to PBD21.
2. When examining the histological pattern of the liver regeneration after burn injury, we did not see the pattern of streaming of the hepatocytes from portal triads to central venule systems. We found most of the EYFP + cells are along the portal venule from PBD2 to PBD7, disseminating to the liver plates around PBD14 and PBD21, but seldom stretching out to the central venule system afterwards.
To better explain the above phenomena, we speculate that the liver regeneration in the early post-burn period is mainly via the proliferation of PDPCs which is activated by significant cellular stress response and liver damage, whereas two to three weeks after injury, with the approaching of the wound closure and the restoration of total body homeostasis, the cellular stress response is attenuating and the liver regeneration is gradually taken over by the physiological self-renewal of the mature hepatocytes. 28 Nevertheless, we noticed that not only cell proliferation but also fat infiltration might contribute to the increase of liver mass in early post-burn period while such a fat infiltration subsided within two weeks post-injury ( Figure S3). Moreover, when we determined the hepatic PARP level after burn injury, we noticed a bi-phasic increase of the expression of cleaved form of PARP around both PBD2 and PBD21, indicating increased apoptosis at these two time-points post-burn. It is clear that the first phase of the increase correlates with acute hepatic stress response and increased apoptosis after burn injury which is consistent with the increased expression of the multiple cellular stress markers. 24,29 The second phase of the increase in the cleaved PARP is concomitant with the decrease in the EYFP + cells from around 25% on PBD14 to 15% on PBD21 and 10% on PBD28, implicating the clearance of the PDPCs and PDPCderived hepatocytes when the homeostasis is finally restored after the injury.
Transcriptomic analysis in the current study reveals the signifi- and is critical for immune and immunological responses. 31 Also, the synergistic interaction among these signalling pathways is evident. 32 Taking together, from the point of view of translational medicine, it is appropriate to consider the expanded population of PDPCs and the activation of the above signalling pathways in these cells as the potential contributing factors to the prolonged inflammatory response and hypermetabolism seen in major burned patients.
Based on our observation, the duration of the pro-inflammatory response and metabolic derangement in the burned mice is around 3 to 4 weeks, peaking at around 2 weeks post-burn. As the maturation rate of the mice aged 1 to 6 month is about 45 times that of human, 33 three weeks in mice could be roughly equivalent to 2 years in humans. The duration of the pro-inflammatory response and metabolic derangement we observed in this burned mice study is thus consistent with the clinical observations of the persistent pro-inflammatory states and hypermetabolism in major burned patients.
We found the down-regulation of the hepatic LXR/RXR signalling pathway concomitant with the activation of the above pro-inflammatory pathways after burn injury by both the transcriptomic analysis and determination of the changes in the level of expression of LXRα in the liver tissue. It is interesting to notice that, on the one hand, LXR signalling is inhibitory to inflammatory responses 34 and thus the down-regulation of the LXR signalling pathway at least correlates with, if not contributes to the activation of the pro-inflammatory responses; 35 on the other hand, LXR signalling is pivotal to lipid homeostasis in mammals and the repression of the LXR signalling implicates impaired lipid metabolism post-burn. 36 More importantly, this may suggest a novel therapeutic target for the care of the severe trauma patients early after the injury. It will be interesting to see whether early and short-term application of LXR agonists to those major trauma patients can be beneficial to the control of overwhelming stress response and pro-inflammatory response, as well as the amelioration of the subsequent and persistent metabolic derangement. There have been several LXR agonists in different phases of clinical trials for the treatment of atherosclerosis.
However, a major issue of concern is their undesirable effects on hepatic lipogenesis and thus the increased risk of hepatic steatosis if they are used for long time. 25 We would propose clinical trial to see whether short-term administration of these LXR agonists to major burn patients in their early post-injury phase would be beneficial, safe and feasible. If RNA sequencing can be done, more information could be available including long non-coding RNAs, micro-RNAs and RNA modifications such as splicing and cleavage. These are all very important for the mechanistic study of the gene transcriptional regulation and control. Also, more accurate information could be available as the copy numbers can be collected directly without the possible skew of the information via PCR amplification. Also, we can look forward to more comprehensive understanding of the dynamic changes of the signalling pathways by transcriptomic profiling if PBD14, 21, 28 and 42 samples can be included for the analysis.
In conclusion, hepatic cellular stress responses and cell damage stimulates proliferation and differentiation of PDPCs with activated pro-inflammatory and stress signalling, contributing to the persistent pro-inflammatory response and metabolic activation after major burn injury. LXRα agonists may have potential therapeutic effects to ameliorate such pro-inflammatory response and hypermetabolism if administered early after the injury.

ACK N OWLED G EM ENTS
This research was supported by the National Institutes of Health (2R01 GM087285-05A1), Canadian Institutes of Health Research (123336), the CFI Leaders Opportunity Fund (25407) and a generous donation from Toronto Hydro.

CO N FLI C T O F I NTE R E S T
The authors confirm that there are no conflicts of interest.

AUTH O R CO NTR I B UTI O N S
LD, SA and MGJ contributed to conceptualization; LD and SA contributed to methodology; LD and YY contributed to investigation; LD contributed to writing-original draft; SA and MGJ contributed to resources and writing-review and editing; MGJ contributed to funding acquisition and supervision.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.