Heat Shock Protein-70 is Required for Optimal Liver Regeneration After Partial Hepatectomy in Mice

Abstract

Liver regeneration is a complex process that restores functional tissue following resection or injury, and is accompanied by transient ATP depletion and metabolic stress in hepatic parenchymal cells. Heat shock protein-70 (Hsp70) functions as a chaperone during periods of cellular stress, and induces expression of several inflammatory cytokines identified as key players during early liver regeneration. We therefore hypothesized that Hsp70 would be required in the initiation of regeneration. Investigations were carried out in a 70% partial-hepatectomy (PHx) mouse model with mice lacking inducible Hsp70 (Hsp70−/−). Liver regeneration was assessed post-operatively using the ratio of liver weight to body weight (LW/BW), and sera and tissues were collected for analysis. In addition, gene expression of Hsp related genes were assessed in a cohort of 23 human living-donor liver transplantation donors. In mice, the absence of Hsp70 was associated with reduced postoperative LW/BW, Ki-67 staining, and TNF-α expression compared to wild-type mice. TNF-α expression was also reduced in livers from Hsp70−/− mice following induction with LPS (1 mg/kg). Clinically, the transcription of multiple Hsp genes was upregulated following donor hepatectomy, especially Hsp70 family members. Together, these results suggest that the early phase of successful liver regeneration requires the presence of Hsp70 to induce TNF-α. Further studies are required to determine whether Hsp70 contributes to liver regeneration as a chaperone, by stabilizing specific interactions required for growth signalling, or as a paracrine inflammatory signal, as can occur in models of shock.

INTRODUCTION

Liver regeneration is a biological response to hepatocellular injury or loss involving a complex network of inflammatory, proliferative and metabolic signals.¹ Following partial hepatectomy (PHx), 75–95% of hepatocytes in a regenerating rodent liver undergo multiple rounds of mitosis, resulting in complete parenchymal restoration within a period of about 7 days.² This leads to a large energy demand in the growing liver tissue, as hepatocytes must suddenly balance available resources between synthesis/proliferation and maintenance of metabolic homeostasis. Transient signs of metabolic stress after PHx are observed in animal studies, including depletion of available ATP (decline in ATP to ADP ratio), mitochondrial swelling, increased mitochondrial membrane permeability, and decreased adenylate energy charge.^3–7 These changes can be extremely rapid, as early as 30 seconds after PHx, and can persist for several days post-surgery.

The heat shock protein-70 (Hsp70) family consists of a group of related molecular chaperones that respond to cellular stress, but its link to the metabolic changes during early liver regeneration is relatively unexplored. The final common pathway for the various sources of cellular stress that activate Hsp70 (including temperature, hypoxia, acidosis, and starvation) involves ATP depletion, leading to the denaturing of folded intracellular proteins, formation of disordered protein aggregates, and compromised cell viability. Inducible Hsp70 chaperone activity enables the cell to cope with the increased burden of misfolded proteins by a direct bind-and-release mechanism that promotes refolding. This mechanism relies on an intrinsic ATPase at the Hsp70 N-terminus, whose activity is catalyzed by other heat shock-responsive components, including members of the Hsp40 family and nuclear exchange factors (NEFs). Combinatorial interactions with various Hsp40 proteins and NEFs provide for a large range of Hsp70 substrate specificities.^{8, 9}

Mounting evidence suggests that Hsp70 is also released from stressed cells as an extracellular protein that can serve as a paracrine signal. Extracellular Hsp70 can stimulate innate immune mechanisms by promoting expression of tumor necrosis factor-alpha (TNF-α) and interleukin-6 (IL-6), and downstream NF-κB signaling - all of which are biologically important components of early liver regeneration.^1,10,11 We therefore hypothesized that the Hsp70 stress response is activated immediately following hepatectomy due to transient metabolic stress, and that this induction leads to production of inflammatory cytokines critical to liver regeneration, such as TNF-α and IL-6. Hence, the Hsp70 stress-response would serve as an essential early trigger for liver regeneration. This study tested whether inducible Hsp70 is required for liver regeneration in a partial hepatectomy mouse model, and whether there is a clinical correlation of HSP gene expression in humans following the related procedure of donor right hepatic lobectomy.

MATERIALS and METHODS

Animals

Animals housed in the Laboratory Animal Facility of the Children’s Hospital of Philadelphia were studied according to an approved IACUC protocol. Hsp70 knockout mice (Hsp70−/−) mice were originally developed on a 129 background, and lacked both inducible forms of Hsp70 (Hsp70.1 and Hsp70.3).¹² All control wild-type (WT) 129 and Hsp70−/− mice used in our studies were males, 6–8 weeks of age, weighing 20–25g. Hsp70−/− mice were maintained on site as a stable inbred colony, and WT mice were purchased (The Jackson Laboratory, Bar Harbor, ME). Both groups were housed in groups of 6 mice/cage and fed normal ad lib diets pre- and post-operatively.

Partial hepatectomy model

To test the role of Hsp70 in liver regeneration, 70% murine partial hepatectomy (PHx) was performed as described.¹³ Mice were anesthetized using a charcoal-filtered induction chamber and flow-meter with a mixture of isoflurane (3–5%) and supplementary oxygen (2 L/min). The abdomen was accessed via midline laparotomy, bowel and liver were gently retracted to expose the hila of the left and median lobes, which together account for ~70% of murine liver mass. Each hilum was ligated separately with a single 3-0 silk suture and resected. The abdomen was closed with a full thickness running silk suture and mice were placed in an incubator (37 °C) for a brief period of recovery (5–10 minutes). At the designated post-operative endpoints (4, 24, 48 or 96 hours), mice were given a lethal dose of pentobarbital and tissues harvested. The surgical incision was reopened, and blood was drawn directly from the inferior vena cava using a 28-gauge syringe. The right and caudate lobes were dissected free from surrounding structures, placed into ice-cold PBS and weighed on a scale. Once a weight was obtained, the right lobe tissue was divided into 8–10 specimens for analysis and partitioned into preservation tubes with RNAlater (Ambion, Austin, TX) or formalin. Plasma was collected for analysis of ALT and total bilirubin from plasma was done through the core pathology facilities (CHOP) and plasma cytokines/chemokines were measured with Luminex (Invitrogen, Carlsbad, CA). Frozen samples of liver tissue and plasma were kept at −80 °C.

qPCR

RNA purification was performed by mechanical disruption with a rotor-blade homogenizer and lysis/extraction solutions from RNAeasy Kit (Qiagen Inc, Valencia, CA). Purified RNA was reverse-transcribed with Taqman reagents (Applied Biosystems, Carlsbad, CA). cDNAs for each specimen were amplified with primers for each gene of interest and normalized against expression levels of 18S. Results were analyzed with a StepOnePlus^™ 96-well plate reader (Life Technologies, Grand Island, NY).

Endotoxin model

Male WT and Hsp70−/− mice, aged 6–8 weeks, were evaluated at baseline, and after injection of lipopolysaccharide (LPS, 1 mg/kg, i.p.) using 3 mice/group. Readouts included qPCR detection of cytokine mRNA and flow cytometric analysis in liver and spleen samples, and flow cytometry of macrophage surface markers and TNF-a using mAbs purchased from BD Biosciences.

Immunohistochemistry and image analysis

Formalin-fixed and paraffin embedded liver sections were stained with hematoxylin and eosin. Additional sections were treated in a pressure cooker with citrate buffer, incubated with an antibody to Ki-67 (Abcam AB1667, 1:400) overnight at 4 °C, followed by avidin biotin complex (Vector Laboratories PK-6100). Immunostained slides were scanned using the Aperio ScanScope^® CS slide scanner (Aperio Technologies, Vista, CA). Digitized images were analyzed using the Aperio ImageScope software (version 10.0.1346.1807; Aperio Technologies, Vista, CA) for determination of the percentage of cells with nuclear positivity from the total number of cells present on the slide. Cells with 2+ or 3+ intensity of staining were considered positive.

Acquisition of patient samples

23 LDLT donors underwent right-lobe hepatectomy at 3 US transplant centers between 2006 and 2009 as part of the Genomics and Regeneration in the Transplant Setting (GRITS) Ancillary Study to the Adult-to-Adult Living Donor Liver Transplantation (A2ALL) consortium (http://www.nih-a2all.org). Institutional Review Board approval was obtained at each participating institution (University of Pennsylvania, Columbia University, Northwestern University) prior to investigation. Two biopsies were obtained: (1) right-lobe, prior to hepatectomy (PRE); (2) left lobe, post-hepatectomy, just prior to abdominal closure (POST). The time between acquisition of the 2 samples was approximately 2–3 hours.

Luminex analysis

Sera were assayed with a Luminex 100 array reader (Luminex Corp., Austin, TX) at 5 different times (0, 4, 24, 48, 96 hours). The 20 analytes were interleukin-1α (IL-1α); IL-1β; IL-2; IL-4; IL-5; IL-6; IL-10; IL-12; IL-13; IL-17; TNF-α; granulocyte-macrophage colony stimulating factor (GM-CSF); interferon gamma (IFN-γ); IP-10 (CXCL10); KC/GRO-alpha (CXCL1); monokine induced by IFN-γ (MIG, CXCL9); monocyte chemoattractant protein 1 (MCP1, CCL2); macrophage inflammatory protein-1α (MIP-1α, CCL3); fibroblast growth factor (FGF); and vascular endothelial growth factor (VEGF).

RNA purification and microarray analysis of human liver biopsies

Total RNA was extracted from the PRE and POST biopsies using Trizol (Invitrogen, CA), after which the RNA was further purified using the RNeasy kit (Qiagen, CA), according to the manufacturer’s instructions. Biotinylated cRNA was prepared with the Ambion MessageAmp Biotin II kit (Ambion, TX) after which labeled cRNA was hybridized to Affymetrix Human Gene 1.0 ST Array GeneChips (Affymetrix, Santa Clara, CA) using standard Affymetrix protocols.

Statistics

Gene expression data from Affymetrix GeneChips were analyzed using BRB-ArrayTools software (developed by Dr. Richard Simon and BRB-ArrayTools Development Team). Normalized signals were generated using RMA, after which class comparisons were performed using a paired t-test with random variance and p<0.001 cut-off for significance.^14,15 Differentially expressed genes were identified using a False Discovery Rate (FDR) of <10% and p-value of <0.001. Liver regeneration in the PHx model was assessed with liver to body weight ratio (LW/BW) and means were compared using Student’s t-test. Pre-operative results for LW/BW, RT-PCR and IHC reflected means from 3 mice. Post-PHx, sample sizes for LW/BW, RT-PCR and immunohistochemistry were 6, 8 and 3 mice/time-point, respectively, and a p-value of 0.05 was deemed significant.

RESULTS

Liver regeneration is impaired in Hsp70−/− mice

We studied whether the 2 stress-inducible Hsp70 genes (HSPA1A, HSPA1B) were required for liver recovery and regeneration in a murine PHx model. In comparison to WT mice, Hsp70−/− mice did not exhibit any obvious developmental defects, anatomical variations or differences in lifespan at baseline. The 2 groups were also similar with regard to pre-operative liver weight (LW), body weight (BW) and LW/BW (Table 1). However, after PHx, Hsp70−/− mice had significantly lower LW/BW ratios compared to WT, indicating a reduced capacity for regenerative growth (POD1 1.5% vs. 1.9%, p<0.001; POD2 1.9% vs. 2.8%, p<0.001; POD4 2.6% vs. 3.2%, p=0.02) (Fig. 1). This did not appear related to differences in hepatocellular injury or synthetic function, as Hsp70−/− and WT mice had comparable plasma levels of ALT and total bilirubin at pre- and post-operative time-points (Fig. 2).

Table 1.

Baseline weights for WT and Hsp70-KO mice

	WT	Hsp70-KO	P-value
liver weight (g)	1.1 ± 0.04	1.1 ± 0.03	0.93
body weight (g)	26.2 ± 0.6	26.6 ± 1.2	0.43
LW/BW	1.4 ± 0.6%	1.3 ± 0.2%	0.18

Abbreviations: Hsp70-KO = heat-shock protein 70, knockout mice; WT = wildtype 129 mice; LW/BW = liver to body weight ratio

Figure 1. Reduced liver regeneration after PHx in Hsp70−/− mice.

PHx was performed in WT and Hsp70−/− mice and regeneration was assessed using measurements of LW/BW at various post-operative time-points. Boxes represent mean LW/BW at each time-point (6 mice/group) and bars indicate standard error. P values were obtained using unpaired Student’s t-tests between WT and Hsp70−/− at each time-point.

Figure 2. No difference between WT and Hsp70−/− in degree of post-operative liver injury or synthetic function.

Plasma was collected from WT and Hsp70−/− mice and sent to the institutional core lab facility for alanine aminotransferase (ALT) and total bilirubin testing.

Attenuated liver growth in Hsp70−/− is associated with reduced cellular proliferation

To test whether lower LW/BW ratios in Hsp70−/− mice reflected decreased cellular proliferation, sections of livers from Hsp70−/− and WT mice were stained for Ki-67, a nuclear marker for cellular proliferation, and evaluated by quantitative image-analysis. Pre-operative nuclear Ki-67 expression was the same in both Hsp70−/− and WT mice (1.7% ± 0.3% vs. 1.2% ± 0.1% positive nuclei, p=0.17). Peak Ki-67 expression occurred at 48 hours, but was 20% lower in Hsp70−/− vs. WT mice (48 ± 3% vs. 69 ± 5% positive nuclei, p<0.01). Area under the curve (AUC) across all measured timepoints was lower in in Hsp70−/− vs. WT mice (60.4 vs. 90.5) (Fig. 3A). Lower levels of Ki-67 expression in Hsp70−/− vs. WT mice were also observed at 24 hours in genotype-blinded review by a pathologist (Fig. 3B), though this difference did not reach statistical significance using automated image analysis (1.5% ± 0.4% vs. 5.8% ± 2.3% positive nuclei, p=0.09). Ki-67 expression remained elevated to the same degree in both groups at 96 hrs (10.6% ± 2.7% vs. 11.5% ± 1.5% positive nuclei, p=0.69). Parallel sections stained for hemotoxylin and eosin at 24 and 48 hrs post-PHx showed normal parenchymal architecture, with no signs of hepatocellular damage or necrosis, and no differences between WT and Hsp70−/− mice (Fig. 3C). These studies suggest that cellular proliferation in mice following PHx peaks on POD2 and remains increased through POD4, but that this response is blunted when inducible Hsp70 is absent.

Figure 3. Hsp70−/− mice exhibit reduced cellular proliferation following PHx.

(A) Quantitative image analysis showed significantly less nuclear staining for Ki-67 in Hsp70−/− mice at 48 hrs post-hepatectomy. (B) Differences are visualized in representative sections of Ki67-stained liver tissue at 24 hrs. (C) No significant histological differences were noted on H&E staining for the adjacent sections.

Reduced TNF-α expression in Hsp70−/− mice after PHx

Inflammatory cytokines and chemokines play critical roles in the early stages of murine liver regeneration.² Several such proteins, especially TNF-α and IL-6, are released in response to Hsp70 paracrine activity in other biological contexts.¹⁰ We therefore questioned whether Hsp70 contributes to the release of cytokines/chemokines in early liver regeneration, and whether the lower regeneration seen in Hsp70−/− mice was due to a loss of this function. We undertook a Luminex screening of sera for cytokine/chemokine expression in WT and Hsp70−/− mice during the immediate period after PHx. Only 4 of the 20 proteins assayed were detectable: IL-6, IL-10, IL-12 and CXCL1 (also known as KC). IL-10 and IL-12 were not significantly elevated at 4 hours in either genotype. IL-6 and CXCL1 levels were significantly elevated, but no differences were found in Hsp70−/− vs. WT groups (IL-6: 450.2 ± 223.5 vs. 330.0 ± 129.3 pg/mL, p=0.67) (CXCL1: 843.7 ± 353.1 vs. 1376 ± 395.2, p=0.37) (Fig. 4A).

Figure 4. Serum and liver cytokine/chemokine expression in Hsp70−/− mice following PHx.

(A) A Luminex-based assay was used to screen serum from Hsp70−/− and WT mice (N=3/group) for cytokine/chemokine expression following PHx. 16 of the 20 factors measured were below detection thresholds. Levels at 4 hours for IL-6, IL-10, IL-12 and CXCL1 are shown. (B) Gene expression was assayed from liver biopsies at baseline and during early liver regeneration using RT-PCR. Levels of TNFα were significantly lower in Hsp70−/− in comparison to WT at 4 hrs. RT-PCR comparisons were performed with Student’s t-test and a sample size of 8 mice/group (4 hrs) or 3 mice/group (0 hrs).

To explore whether lack of Hsp70 affected intrahepatic cytokine/chemokine production, we next evaluated gene expression levels of TNF-α, IL-6, CXCL1 and HGF in Hsp70−/− and WT liver samples. Expression was measured both before surgery and at 4 hrs post-PHx using qPCR (Fig. 4B). Pre-operative levels were not significantly different in Hsp70−/− vs. WT, though they appeared slightly higher in Hsp70−/− for TNFα (7.6 ± 3.8 vs. 1.7 ± 0.6 fold, p=0.20) and CXCL1 (9.0 ± 3.7 vs. 1.1 ± 0.05 fold, p=0.10). In both groups, all measured factors were upregulated at 4 hrs relative to pre-operative levels. At 4 hrs, Hsp70−/− and WT mice had comparable expression levels for IL-6 (15.1 ± 4.6 vs. 11.9 ± 3.2 fold, p=0.60), CXCL1 (104.0 ± 21.0 vs. 135.7 ± 19.2 fold, p=0.32), and HGF (1.9 ± 0.1 vs. 2.1 ± 0.3 fold, p=0.43), whereas TNF-α expression was significantly lower in Hsp70−/− mice (8.9 ± 1.6 vs. 26.2 ± 5.4 fold, p=0.01).

Reduced endotoxin responses in Hsp70−/− mice

Although the phenotype of HSP70−/− and WT mice appeared similar pre-PHx, we were interested in assessing whether the macrophage populations of these mice were comparable in terms of basal and LPS-induced levels of expression of TNF-α and related cytokines. We found that the basal levels of TNF-a, IL-1b and IL-6 mRNA were similarly very low in the livers and spleens of Hsp70−/− and WT mice (Fig. 5). However, whilst considerable induction of all 3 cytokines in both tissues was observed at 3 hours post-LPS, the level of cytokine mRNA in each case was significantly decreased (p<0.01) in Hsp70−/− vs. WT mice (Fig. 6). Likewise, while flow cytometric analysis of splenic macrophage and DC populations in Hsp70−/− vs. WT mice showed only modest differences in the proportions of Ly6C+/CD11b+ cells, LPS administration resulted in decreased TNF-α production by splenic CD11b+ macrophages and CD11c+ DC in Hsp70−/− vs. WT mice (Fig. 6).

Figure 5. Impaired cytokine mRNA production in Hsp70−/− mice following LPS injection.

Cytokine mRNA expression levels in Hsp70−/− and WT mice (3/group) were analyzed by qPCR using pre- and 3 hours post-injection of LPS (1 mg/kg). Levels were normalized to 18S and are shown as relative expression (mean ± SD, **p<0.01).

Figure 6. Impaired macrophage and DC production of TNF-a by Hsp70−/− mice following LPS injection.

Flow cytometric analysis of macrophage and DC populations in spleens of Hsp70−/− and WT mice (3/group) pre- and 3 hours post-injection of LPS (1 mg/kg). Representative flow data are shown at left, with the proportion of labeled cells indicated in each panel; histograms at right show pooled percentage data (mean ± SD, *p<0.05, or n/s, not significant).

HSP70 expression is upregulated during early liver regeneration in humans

We tested the expression of HSP70 related genes in a human PHx setting by assessing RNA transcripts in human LDLT donor livers using Affymetrix Human Gene 1.0 ST Array GeneChips. In a cohort of 23 human LDLT donors, HSP expression was significantly induced during the first hours of liver regeneration (Table 2). Transcripts from 39 molecules from 4 different HSP families were upregulated in the POST biopsy compared to the PRE (18 HSP40; 9 HSP70; 9 HSP90; 3 “other”). The two stress-inducible members of the HSP70 family, HSPA1A and HSPA1B, were both found to be upregulated following hepatectomy (fold increases 1.50 and 1.68 respectively). Two other HSP70 family members were notable for having more pronounced inductions than any of the other HSP genes, particularly HSPA13 (fold increase 3.55) and HSPA5 (fold increase 2.68). HSPA13 is a constitutively expressed, non-inducible member of the HSP70 family with ubiquitin-related functions, and has been implicated in Alzheimer’s disease and epilepsy.^16,17 HSPA5 is a form of HSP70 specific to the endoplasmic reticulum with critical housekeeping functions at baseline.¹⁸ With respect to clinical outcomes in the LD donor cohort, all patients went on to full recovery of liver function, with no cases of peri-operative morbidity or mortality.

Table 2.

Upregulated gene expression for heat-shock proteins in human liver regeneration

Family	Symbol	Gene name	Fold-change*	p-value
HSP40	DNAJB11	DnaJ (Hsp40) homolog, subfamily B, member 11	2.59	< 1E-07
	DNAJB9	DnaJ (Hsp40) homolog, subfamily B, member 9	2.58	< 1E-07
	DNAJC2	DnaJ (Hsp40) homolog, subfamily C, member 2	2.33	< 1E-07
	DNAJB1	DnaJ (Hsp40) homolog, subfamily B, member 1	2.22	< 1E-07
	DNAJC3	DnaJ (Hsp40) homolog, subfamily C, member 3	1.89	1.00E-07
	DNAJA1	DnaJ (Hsp40) homolog, subfamily A, member 1	1.80	< 1E-07
	DNAJC10	DnaJ (Hsp40) homolog, subfamily C, member 10	1.67	< 1E-07
	DNAJA3	DnaJ (Hsp40) homolog, subfamily A, member 3	1.42	< 1E-07
	DNAJB4	DnaJ (Hsp40) homolog, subfamily B, member 4	1.42	4.76E-04
	DNAJA2	DnaJ (Hsp40) homolog, subfamily A, member 2	1.41	< 1E-07
	DNAJB6	DnaJ (Hsp40) homolog, subfamily B, member 6	1.39	< 1E-07
	DNAJC11	DnaJ (Hsp40) homolog, subfamily C, member 11	1.38	< 1E-07
	DNAJC1	DnaJ (Hsp40) homolog, subfamily C, member 1	1.29	3.49E-05
	DNAJC16	DnaJ (Hsp40) homolog, subfamily C, member 16	1.25	2.80E-06
	DNAJC21	DnaJ (Hsp40) homolog, subfamily C, member 21	1.22	2.41E-05
	DNAJC7	DnaJ (Hsp40) homolog, subfamily C, member 7	1.19	4.96E-04
	DNAJC5	DnaJ (Hsp40) homolog, subfamily C, member 5	1.18	1.19E-05
	DNAJB6	DnaJ (Hsp40) homolog, subfamily B, member 6	1.17	8.93E-04
HSP70	HSPA13	heat shock protein 70kDa family, member 13	3.55	< 1E-07
	HSPA5	heat shock 70kDa protein 5 (glucose-regulated protein, 78kDa)	2.68	< 1E-07
	HSPA8	heat shock 70kDa protein 8	1.73	< 1E-07
	HSPA4	heat shock 70kDa protein 4	1.71	< 1E-07
	HSPA1B	heat shock 70kDa protein 1B	1.68	3.00E-07
	HSPA14	heat shock 70kDa protein 14	1.52	< 1E-07
	HSPA1A	heat shock 70kDa protein 1A	1.50	1.14E-05
	HSPA4L	heat shock 70kDa protein 4-like	1.44	4.08E-05
	HSPA9	heat shock 70kDa protein 9 (mortalin)	1.35	< 1E-07
HSP90	HSP90AB1	heat shock protein 90kDa alpha (cytosolic), class B member 1	1.76	< 1E-07
	HSP90AB3P	heat shock protein 90kDa alpha (cytosolic), class B member 3	1.69	< 1E-07
	HSP90AA6P	heat shock protein 90kDa alpha (cytosolic), class A member 6	1.56	1.00E-07
	HSP90AA1	heat shock protein 90kDa alpha (cytosolic), class A member 1	1.43	3.00E-07
	HSP90AA2	heat shock protein 90kDa alpha (cytosolic), class A member 2	1.43	2.50E-06
	HSP90B1	heat shock protein 90kDa beta (Grp94), member 1	1.42	1.00E-07
	HSP90AB4P	heat shock protein 90kDa alpha (cytosolic), class B member 4	1.38	< 1E-07
	HSP90AB2P	heat shock protein 90kDa alpha (cytosolic), class B member 2	1.30	2.24E-05
	HSP90B3P	heat shock protein 90kDa beta (Grp94), member 3	1.24	2.28E-05
Other	HSPH1	heat shock 105kDa/110kDa protein 1	2.30	< 1E-07
	HSPD1	heat shock 60kDa protein 1 (chaperonin)	1.46	< 1E-07
	HSPE1	heat shock 10kDa protein 1 (chaperonin 10)	1.25	1.57E-04

^*Fold change = expression in post-hepatectomy biopsy / expression in pre-hepatectomy biopsy

DISCUSSION

Over 80 years of investigation have passed since Higgins and Anderson first described liver regeneration in a rodent hepatectomy model, yet the most proximal components initiating this process remain unclear.¹⁹ Identifying these molecular triggers has important significance for patients recovering from liver resection - whether due to cancer, transplant donation, ischemic injury or trauma - as impairments in liver regeneration can lead to organ failure, and because no clinical process to augment liver regeneration currently exists. Several unique features make the Hsp response an excellent candidate for an early initiating event in liver regeneration. Hsp proteins are known to respond to ATP depletion, which has been shown to occur immediately following liver resection.^3,20,21 Intracellular Hsp proteins perform cellular housekeeping roles at baseline, and are therefore available in great abundance without the need for a priming step. Indeed, stress-related spillage of Hsp into the extracellular milieu has been proposed as a paracrine damage signal.²² Extracellular Hsp molecules are capable of inducing the inflammatory cytokines IL-6 and TNF-α, both of which are considered integral to liver regeneration.^2,10,22 Accordingly, in this report, we hypothesized that Hsp responses contribute to the release of cytokines during the early phases of liver regeneration. We demonstrated that: (1) mice lacking inducible Hsp70 have diminished liver regeneration, cell proliferation, and TNF-α expression; and (2) Hsp gene expression is upregulated in early human liver regeneration.

Our findings are consistent with the limited prior data that exists relating Hsp70 to liver regeneration. The first study reported inductions of mRNA and/or protein expression for Hsp27, Hsp60, Hsp70 and Hsp90 at various post-operative time-points following PHx in mice.²³ To test whether or not these expression changes were required for complete liver regeneration, the authors treated the mice with quercetin, a nonspecific flavanoid inhibitor of Hsp synthesis. They found LW/BW was modestly decreased (statistics not reported) for treated vs. untreated mice at 48 hours post-hepatectomy. The other prior study examined the effects of a known Hsp90 inhibitor (17-dimethylaminoethylamino-17-demethoxygeldanamycin; 17-DMAG) on liver regeneration in the murine PHx model.²⁴ Importantly, 17-DMAG not only inhibits Hsp90, but also potentiates Hsp70 expression. In the absence of stimuli, Hsp90 sequesters heat-shock factor 1 (Hsf1), an Hsp70 transcription factor, and renders it inactive. Chemical inhibition of Hsp90 leads to Hsf1 release and Hsp70 transcription. Contrary to the authors’ initial hypothesis, liver regeneration was ultimately not worsened by 17-DMAG treatment; rather, 17-DMAG led to increased post-PHx liver weight and cellular proliferation at 24 and 48 hours respectively, as well as higher VEGF expression. These results suggest that increased Hsp70 function, secondary to 17-DMAG treatment, correlates with increases in liver regeneration. Our data showing that Hsp70 loss is associated with diminished regeneration are therefore complementary with both previous studies.

We noted that mice lacking Hsp70 had reduced TNF-α expression in the early hours following PHx (Figure 4B), suggesting that Hsp70 contributes to the cytokine responses involved in liver regeneration. A non-traditional role for Hsp proteins, aptly termed “chaperokine” activity, has been well-described in other contexts.¹¹ Thus, Hsp70 was found to have paracrine signaling capabilities, including the activation of human monocytes through cell-surface receptor CD14, release of intracellular calcium and the subsequent induction of TNF-α, IL-1b and IL-6 expresssion.¹⁰ These findings were supported by two other studies noting similar chaperokine functions of additional Hsp proteins. Thus, Basu et al demonstrated that purified Hsp proteins (Hsp70, Hsp90) from mouse liver were capable of inducing cytokine expression (IL-1b, IL-12, TNF-α) in peritoneal macrophages.²² Likewise, Ohashi et al found that purified extracellular human Hsp60 led to expression of TNF-α and nitric oxide (NO) in murine bone-marrow derived macrophages through toll-like receptor 4.²⁵ All of these studies identified the macrophage as the target cell for Hsp chaperokine activity, raising the possibility that Hsp signalling also induce Kupffer cells, which have already been implicated in the cytokine expression that occurs during early liver regeneration. While our phenotypic studies of macrophage subpopulations did not show major differences between Hsp70−/− and WT mice, Hsp70 deletion was accompanied by impaired liver and splenocyte cytokine production upon exposure to LPS (Fig. 5). Likewise, Hsp70−/− macrophage and DC populations showed decreased LPS-induced TNF-α production when compared to responses by WT mice (Fig. 6). Hence, Hsp70 appears vital to TNF-α induction by macrophages in liver and lymphoid tissues subject to inflammatory stimuli, as well as following PHx.

The current study has several limitations. In terms of project design, the human and animal findings are both observational, and therefore cannot test for true causal relationships – those proposed are based on published literature. The human data are also limited by the nature of the study subjects. Pre-transplant selection ensures that LDLT donors are healthy individuals. Results from LDLT donors may therefore not be applicable to other patients whose livers regenerate, but who are systemically ill at baseline. Such examples include liver resection for cancer or trauma, transplantation of partial allografts in LDLT or split-liver donation, or acute hepatic injury following toxic or metabolic insults. With respect to the animal model, Hsp70−/− mice contained mutations that are stable and global, i.e. inducible Hsp70 was absent from birth in all body tissues. It is therefore possible that our observed differences were related to unperceived developmental effects of this genotype. Our data are further limited by certain technical drawbacks. Since liver specimens were analyzed as whole tissue, they included hepatocytes and all associated cells, including resident macrophages (Kupffer cells), biliary endothelium and vasculature. We therefore cannot specify which cell type(s) is/are responsible for our observed findings in liver. Of additional technical concern, the majority of the cytokines/chemokines assayed with Luminex were below the detection threshold. These results should be interpreted as inconclusive rather than negative.

In summary, Hsp proteins were upregulated in human liver regeneration, most notably those in the Hsp70 family, and mice lacking Hsp70 had decreased liver regeneration and TNFα expression. These data suggest a mechanism in which Hsp70 contributes to early liver regeneration via potentiation of TNFα expression. Future studies should aim to test whether increased extracellular Hsp70 is sufficient to stimulate this mechanism, and whether it can reconstitute function in the knockout model. If so, strategies to improve liver regeneration in surgical patients could potentially include increasing circulating Hsp70 levels, which could be accomplished through direct administration, Hsp90 inhibition, or remote ischemic preconditioning.

Abbreviations

A2ALL

Adult-to-Adult Living Donor Liver Transplantation

FGF

fibroblast growth factor

GM-CSF

granulocyte-macrophage colony stimulating factor

GRITS

Genomics and Regeneration in the Transplant Setting

Hsp70

heat shock protein-70

IL

interleukin

IFN-γ

interferon gamma

CXCL10

IP-10

CXCL1

KC/GRO-alpha

LW/BW

liver to body weight ratio

LDLT

living donor liver transplantation

MIP-1α

macrophage inflammatory protein-1α

MCP-1

monocyte chemoattractant protein

MIG

monokine induced by interferon-γ

NEF

nuclear exchange factor

NF-κB

nuclear factor kappa B

PHx

partial hepatectomy

POD

post-operative day

TNF-α

tumor necrosis factor-alpha

VEGF

vascular endothelial growth factor