Targeting A-kinase anchoring protein 12 phosphorylation in hepatic stellate cells regulates liver injury and fibrosis in mouse models

Trans-differentiation of hepatic stellate cells (HSCs) to activated state potentiates liver fibrosis through release of extracellular matrix (ECM) components, distorting the liver architecture. Since limited antifibrotics are available, pharmacological intervention targeting activated HSCs may be considered for therapy. A-kinase anchoring protein 12 (AKAP12) is a scaffolding protein that directs protein kinases A/C (PKA/PKC) and cyclins to specific locations spatiotemporally controlling their biological effects. It has been shown that AKAP12’s scaffolding functions are altered by phosphorylation. In previously published work, observed an association between AKAP12 phosphorylation and HSC activation. In this work, we demonstrate that AKAP12’s scaffolding activity toward the endoplasmic reticulum (ER)-resident collagen chaperone, heat-shock protein 47 (HSP47) is strongly inhibited by AKAP12’s site-specific phosphorylation in activated HSCs. CRISPR-directed gene editing of AKAP12’s phospho-sites restores its scaffolding toward HSP47, inhibiting HSP47’s collagen maturation functions, and HSC activation. AKAP12 phospho-editing dramatically inhibits fibrosis, ER stress response, HSC inflammatory signaling, and liver injury in mice. Our overall findings suggest a pro-fibrogenic role of AKAP12 phosphorylation that may be targeted for therapeutic intervention in liver fibrosis.


Introduction
Hepatic stellate cells (HSCs) constitute approximately 5-8% of the normal liver and are major sites for vitamin A storage in the body (1). During chronic liver injury, HSCs acquire a pro-fibrogenic phenotype or activated state that is critical in the liver's response to injury (2). HSC activation causes increased production of extracellular matrix (ECM) components such as collagens and α-smooth muscle actin (α-SMA). Persistent injury leads to fibrosis due to abnormal accumulation of ECM (3). HSC pathways that cause fibrogenic responses in the liver can be targeted for therapeutic intervention.
Collagen maturation and secretion is facilitated by the endoplasmic reticulum (ER)-resident chaperone, heat shock protein 47 (HSP47) along with other ER foldases such as BIP/GRP78 (4,5). Under normal physiological conditions, HSP47 is expressed at low levels in the liver (6) and other organs such as lung, heart, kidney (7). Fibrogenic stimulation by carbon tetrachloride (CCl4) or bile duct ligation (BDL) in mice and human liver fibrosis is associated with induction of HSP47 expression (6,8,9). The induction in HSP47 correlates with increased collagen secretion from activated HSCs during liver fibrosis. Therefore, silencing HSP47 to inhibit collagen production is an appealing option for reversing fibrosis (10). However, because HSP47 also plays a chaperoning function in the healthy liver and other organs, the collateral effects of its therapeutic silencing should be investigated (10). Apart from its function as a collagen chaperone, a recent interactome study identified HSP47 as a binding partner for an unfolded protein response (UPR) sensor protein, inositol-requiring enzyme 1 alpha (IRE1α) (5). HSP47 activates IRE1α oligomerization and phosphorylation by displacing its regulator, BIP, thereby triggering the UPR response during ER stress (5). Whether triggering of UPR signaling by HSP47-IRE1α interaction and BIP displacement in HSCs may enhance folding of pro-fibrogenic proteins such as collagen is unclear so far. But it is generally accepted that HSCs exhibit ER stress and UPR signaling in response to liver injury stimuli (11).
It has been demonstrated that the scaffolding ability of AKAP12 is altered by its phosphorylation (17,19,22).
Prephosphorylation of AKAP12 by PKC suppresses its interaction with PKC itself and increases PKC activity (23).
Phosphorylation of AKAP12 at a PKC phosphorylation site (S507/515) prevents the sequestration of CCND1 by AKAP12 leading to its nuclear translocation, allowing cell cycle progression (19,21). AKAP12 phosphorylation by cyclin-dependent kinase 1 (CDK1) at a threonine residue (T766) enhances the recruitment of the polo-like kinase (PLK1) in human glioblastomas to ensure efficient mitotic progression (13). Even though phosphorylation is known to regulate AKAP12's scaffolding activities, the functional impact of its phospho-modifications on liver disease has not been evaluated. We previously demonstrated that HSC activation during liver injury was associated with an induction in phospho-AKAP12 (24). In this work we demonstrate that specific AKAP12 phosphorylation events in HSCs regulate its scaffolding activity towards the collagen chaperone, HSP47. HSC-specific CRISPR-editing of AKAP12's phospho-sites preserves the AKAP12-HSP47 scaffold, reduces HSP47's collagen-chaperoning activity, dramatically lowering overall collagen content and liver injury during carbon-tetrachloride (CCl4)-induced liver fibrosis. AKAP12 phospho-modulation directed towards HSCs regulates HSP47-IRE1α interaction, thereby controlling UPR signaling in HSCs. Furthermore, AKAP12 phospho-site modulation in HSCs suppresses overall ER stress in the fibrotic liver. Our data supports a previously unidentified function of AKAP12 and its phosphomodification in regulating the outcome of liver fibrosis in animal models.

Results
Expression, phosphorylation, and scaffolding activity of AKAP12 is altered in CCl4-treated mouse liver and human liver fibrosis The expression of AKAP12 protein was decreased in livers of CCl4-treated mice by 14% compared to oil controls without a change in Akap12 mRNA ( Figure 1A). CCl4 treatment induced the expression of HSC activation marker, α-SMA by 4.5-fold compared to control ( Figure 1A, figure 1-source data 1). As evidenced by proximity ligation assay (PLA), the phosphorylation of AKAP12 was induced in desmin-positive HSCs of CCl4 livers by 5.3-fold compared to control ( Figure 1B, figure 1-source data 2). AKAP12 staining judged by Image J quantification (materials and methods) was decreased in CCl4-treated liver by 16% compared to control ( Figure 1B, figure 1 source data 2) consistent with the western blot result ( Figure 1A). The interaction of AKAP12 with HSP47 was inhibited by 54% despite a 3.9-fold increase in overall HSP47 levels in CCl4 livers compared to control ( Figure 1C, figure 1-source data 3). A human liver fibrosis tissue array containing 16 liver fibrosis tissues and 11 normal tissues was stained with PLA probes for AKAP12 and HSP47 to detect their interaction. The interaction between AKAP12 and HSP47 was inhibited by 64% in human liver fibrosis tissue compared to normal ( Figure 1D, figure 1-figure supplement 1). This was associated with a 20% decrease in total AKAP12 staining and a 3.8-fold increase in HSP47 staining in liver fibrosis compared to normal ( Figure 1D).

CRISPR-directed editing of AKAP12's activation-responsive phospho-sites enhances AKAP12's scaffolding activity and inhibits HSC activation
The phospho-peptide map of AKAP12 protein from Day 7 culture-activated human or mouse HSCs was compared to that of Day 0 quiescent HSCs or normal hepatocytes. A peptide region containing 5 S/T phospho-sites exhibited increased phosphorylation in Day 7 activated HSCs but not in Day 0 HSCs or hepatocytes (Table 1, table   supplement 1). These activation-responsive phospho-sites were conserved in mouse and human (Table 1). Day 5 activated human HSCs were transfected with CRISPR small guide RNA (sgRNA) and donor RNA (Supplementary table S1) to delete the 5 AKAP12 phosphorylation sites by homology-directed repair (HDR) as described under materials and methods. Genomic DNA PCR from CRISPR edited (HDR) cells using deletion-specific primers (Supplementary table S1) resulted in a 261 bp amplicon that was not amplified in WT cells or cells treated with SaCas9 (staphylococcus aureus CRISPR-associated protein) alone (Figure 2A, original gel shows 4 experiments).
The interaction between AKAP12 and HSP47 in CRISPR-edited HSCs (HDR) was induced by 2.5-fold compared to WT cells ( Figure 2B, original blot developed with anti-mouse IgG is shown in figure 2-source data 1). This was associated with a 40% decrease in α-SMA levels, demonstrating that AKAP12 phospho-site editing inhibited HSC activation ( Figure 2B). The overall level of HSP47 decreased by 25% whereas the level of AKAP12 protein remained unchanged after HDR ( Figure 2B). Deletion of phospho-sites in mouse HSCs resulted in a 422 bp deletion-specific amplicon (Figure 2-figure supplement 1-source data 1). Like human HSCs, mouse HSCs also exhibited increased AKAP12-HSP47 interaction after AKAP12 phospho-site editing (figure 2-figure supplement 1source data 2). Reversal of HSC activation by AKAP12 editing was determined by examining vitamin A auto fluorescence (25). Cultured human HSCs at Day 0 exhibited strong vitamin A autofluorescence that was reduced in Day 5 activated HSCs ( Figure 2C, 3 independent experiments are shown). AKAP12 editing in Day 5 HSCs restored the loss of vitamin A fluorescence compared to Day 5 HSCs or Day 5 HSCs+SaCas9 alone ( Figure 2C).
HSP47 is an ER-resident chaperone (4). A weak PLA signal of AKAP12-HSP47 interaction co-localized with the ER marker, calreticulin in activated (WT) HSCs ( Figure 2D, left panel). However, upon CRISPR-editing (HDR) a strong AKAP12-HSP47 PLA signal co-localized with calreticulin in the ER ( Figure 2D, left panel). We examined whether HSP47's collagen-chaperoning activity was regulated by AKAP12 phospho-site editing. Our results show that the collagen-HSP47 PLA signal strongly co-localized in the ER of activated HSCs (WT) ( Figure 2D, right panel).
CRISPR-editing of AKAP12 (HDR) reduced the collagen-HSP47 interaction significantly by 65% compared to WT cells ( Figure 2D, right panel). Individual experiments are shown in figure 2-source data 2.

PKCα phosphorylates AKAP12 and inhibits its interaction with HSP47
Kinase-prediction software were used to predict that out of the 5 AKAP12 activation-responsive phospho-sites, two serines (S687/S688) were strongly predicted substrates of PKCα kinase with a consensus of [S/T]-X-R/K whereas one threonine (T675) could not be assigned a kinase (supplementary table S2). S676/S678 sites were also PKCα sites but shared consensus sites with calmodulin kinase (CAMK). The overall confidence of prediction for the S676/S678 sites was less than that of S687/S688 sites. In vitro kinase assay followed by phostag™ gel analysis revealed that phosphorylation of AKAP12 was significantly enhanced in the presence of active PKCα enzyme compared to kinase negative controls ( Figure 3A). Mutations of AKAP12 S676/S678 to alanine modestly reduced the phosphorylation of biotinylated recombinant AKAP12 whereas S687A/S688A mutation dramatically suppressed the phostag™ shift of AKAP12 ( Figure 3A). The mutation seemed to completely inhibit the phospho-band. Since other phosphorylation events could also cause the shift, we repeated the experiment to see whether this complete suppression was reproducible. In an additional experiment ( Figure 3-source data 1) we observed that the S687/S688A mutation suppressed but did not always wipe out the phospho-shift. Also, in some experiments we observed the -kinase control had a faint phospho-signal. The recombinant protein produced by rabbit reticulocyte lysates in an in vitro translated system may have baseline phosphorylation as reported in the manufacturer's protocol (TNT® Coupled Transcription/Translation system, Promega). Direct binding was observed between biotinylated AKAP12 and HSP47 in a recombinant system in the absence of active PKCα ( Figure 3B, figure 3source data 2). Presence of PKCα inhibited the interaction between AKAP12 and HSP47 ( Figure 3B). To evaluate whether phosphorylation of AKAP12 by PKCα in HSCs would regulate AKAP12's scaffolding activity, cells were treated with PKCα siRNAs (A or B). Silencing PKCα by 74% with siRNA-A increased AKAP12-HSP47 interaction by 3-fold whereas a 90% knockdown caused by siRNA-B enhanced AKAP12-HSP47 interaction by 8-fold compared to negative control siRNA ( Figure 3C, figure 3-source data 3). HSP47 levels remain unchanged by siRNA treatments ( Figure 3C).
In vivo gene editing of the Akap12 region corresponding to its activation-responsive phospho-sites in HSCs of mouse liver.
The Akap12 exon 3 contains sequences corresponding to the activation-responsive phospho-sites of AKAP12 protein. To perform gene editing of this region specifically in HSCs of mouse liver, two different CRISPR HDR approaches were used ( Figure 4A). A PDEL donor was used to delete the AKAP12 phospho-sites whereas each S or T phospho-site was mutated to A using a PMUT donor. Two unique sgRNAs specific for the region around the phospho-sites along with the donor ( Figure 4A, Supplementary table S1) were cloned into AAV vectors ( Figure 4B, left panel). To perform CRISPR editing in HSCs of mouse liver, the SaCas9 enzyme was cloned into AAV vector under control of two different HSC-specific promoters (Glial fibrillary acidic protein, GFAP or Lecithin retinol acyltransferase, LRAT) (26,27). AAV vectors were injected into mice during oil or CCl4 administration according to the plan in Figure 4B, right panel. To evaluate the HSC specificity of GFAP-SaCas9 mediated CRISPR (CR) editing compared to that of an empty vector (EV) control (materials and methods), genomic DNA of HSCs or hepatocytes isolated from livers of oil+EV, oil+CR, CCl4+EV, CCl4+CR groups was subjected to multiplex PCR with PDEL forward and reverse primers and a PDEL deletion-specific primer (Supplementary table S1). Oil+EV or CCl4+EV HSCs or hepatocytes gave a 298 bp amplicon in this multiplex PCR. Oil+CR or CCl4+CR groups resulted in 298 bp WT and 256 and 154 bp mutated amplicons due to complementarity with the deletion-specific primer ( Figure   4C). PCR with deletion-specific primers did not amplify the 256 or 154 bp mutant region in hepatocytes, indicating that CRISPR-editing using an HSC promoter-specific SaCas9 occurred in HSCs but not hepatocytes ( Figure 4E).
HSC specificity of CRISPR was also evaluated by immunostaining of Sacas9 enzyme with HSC marker, desmin or hepatocyte marker, albumin. The GFAP-driven SaCas9 enzyme strongly co-localized with desmin-positive HSCs ( Figure 4D) but not with albumin-positive hepatocytes in the liver ( Figure 4F). CCl4 exposure increased the overall numbers of desmin-positive HSCs ( Figure 4D) due to increased activation and proliferation (28). A specific primer to detect PMUT could not be designed, hence PMUT specificity and efficiency was tested along with PDEL by next generation amplicon sequencing (NGS) using a 298 bp amplicon from HSCs or hepatocytes of GFAP-SaCas9 CRISPR livers ( Figure 4G). On-target and off-target base changes were analyzed by comparing the target read sequences to the reference sequence of WT Akap12 amplicon as described under materials and methods. For the PDEL CRISPR, Oil+CR HSCs exhibited 30% mutant reads compared to the total reads whereas CCl4+CR HSCs exhibited 60% mutant reads compared to total ( Figure  The CRISPR deletion efficiency using LRAT-saCas9 in HSCs of oil+CR group was 45% whereas that of the CCL4+CR group was 30% of total reads (figure 4-figure supplement 1, Supplementary table S3).

Phospho-editing of AKAP12 regulates liver injury and fibrosis in the CCl4 mouse model
At gross level, CCl4 administration for five weeks reduced the body weight of mice by 20% and increased the liver to body weight ratio by 1.25-fold compared to oil ( Figure 5A). AKAP12 phospho-editing by GFAP-SaCas9 in normal mice (oil+CR) did not alter body or liver weight compared to oil+EV ( Figure 5A). However, AKAP12 phospho-editing in CCL4 mice (CCl4+CR) normalized CCL4+EV-mediated alterations in body weight and liver/body weight to that of oil+EV levels ( Figure 5A). Histologically, control mice (oil) had a normal hepatic cord pattern around the central vein whereas fatty vacuolar changes and disorganized hepatic lobular structure with centrilobular fibrosis was observed in CCl4 livers ( Figure 5B) as referenced previously (29). This was associated with an 8-13-fold induction in liver injury as measured by ALT/AST levels ( Figure 5C). AKAP12 phospho-editing by PDEL or PMUT in control mice (oil+CR) did not affect the normal histology of the liver or the levels of ALT/AST ( Figure 5B, C). AKAP12 phospho-editing by PDEL or PMUT in CCl4 mice (CCl4+CR-PDEL or PMUT) dramatically reduced the CCl4-induced histological distortions and decreased the ALT/AST level by 75-80% compared to CCl4+EV ( Figure 5B, C). H&E staining for 6 PDEL experiments is shown in figure 5-source data 1. LRAT-SaCas9 directed PDEL-CRISPR also resulted in higher body weight, lower liver/body weight ratio and suppression of CCl4-induced histological changes like that of GFAP-SaCas9 ( Figure 5-figure supplement 1).

Phospho-editing of AKAP12 regulates AKAP12's HSP47-scaffolding activity, HSC activation, HSP47's collagen-chaperoning activity and collagen production in the CCl4 mouse model
The AKAP12-HSP47 scaffold was reduced in livers of CCl4+EV mice compared to oil controls ( Figure 6A  phospho-editing dramatically inhibited CCl4-mediated HSC activation as evidenced by a drop in α-SMA levels ( Figure 6A). In conjunction with restoration of AKAP12-HSP47 scaffold, the increased interaction between collagen and HSP47 upon CCl4 exposure was inhibited by AKAP12 phospho-editing ( Figure 6C). AKAP12 PDEL or PMUT phospho-editing also inhibited the increase in collagen mRNA levels caused by CCl4 exposure ( Figure 6C). Coimmunoprecipitation of collagen with HSP47 antibody yielded non-specific bands at positions above the collagen position in all samples including IgG control. The original uncropped blot is shown in figure 6-source data 3. PLA staining showed that the AKAP12-HSP47 scaffold was localized with desmin-positive HSCs under normal (oil) conditions ( Figure 6D). A drop in AKAP12-HSP47-desmin co-localization was observed upon CCl4 exposure that was restored by AKAP12 PDEL or PMUT phospho-editing ( Figure 6D). Picosirus red staining of CCl4 livers showed increased collagen deposition that was substantially reduced when mice were administered AKAP12 phosphoediting vectors ( Figure 6E, top panel). Sirius red staining for 6 PDEL experiments is shown in figure 6-source data 4. The hydroxyproline content of collagen was increased 2.4-fold in CCl4 livers compared to oil+EV and normalized by AKAP12 PDEL. PMUT phospho-editing inhibited CCl4-mediated induction but did not completely normalize hydroxyproline content to oil+EV or oil+CR levels ( Figure 6E, bottom panel).

HSC-specific phospho-editing of AKAP12 regulates the ER stress response
To determine how HSC-specific AKAP12 phospho-editing reduced overall liver injury and modulated collagen mRNA levels upon CCl4 exposure, we performed proteomics analysis of HSCs and livers isolated from oil+EV, oil+CR, CCl4+EV or CCl4+CR groups to compare the molecular changes under these conditions. Proteomics analysis revealed alterations in several proteins in CCl4 HSCs as well as total liver that were regulated by AKAP12 HSC-specific phospho-editing (Supplementary table S5). Ingenuity pathway analysis (IPA) of these proteins identified two top scoring pathways, the ER stress response and UPR, that were significantly dysregulated by CCl4 and were normalized upon AKAP12 phospho-editing (supplementary table S5 Figure 7B). However, even though the proteomics analysis showed inhibition of CCl4-induced BIP levels by AKAP12 phospho-editing ( Figure 7A), we could not confirm this effect by western blotting ( Figure 7B). Since BIP is a known collagen chaperone (5), we examined its interaction with collagen in our CRISPR model. BIP exhibited increased interaction with collagen in the CCl4+EV HSCs compared to oil+EV HSCs and AKAP12 phospho-editing strongly inhibited the BIP-collagen interaction in HSCs ( Figure 7B). IRE1α, a UPR component that binds to HSP47 and becomes phosphorylated during ER stress (5), exhibited increased interaction with HSP47 in CCL4 HSCs that was inhibited by AKAP12 phospho-editing ( Figure 7B). The IRE1α-HSP47 interaction was further confirmed in desmin-positive HSCs of the CRISPR model by PLA staining (Figure 7-figure supplement 2). CCl4-mediated IRE1α phospho-activation (S724 phosphorylation) was strongly inhibited by AKAP12 phospho-editing without a change in total IRE1α levels ( Figure 7B). Furthermore, two pathways, P38MAPK and SMAD2/3 that are known to be induced in HSCs through IRE1α activation (30) were also suppressed by AKAP12 phospho-editing ( Figure 7B). The proteome of CCl4-exposed livers exhibited increased ER stress and UPR signaling components that were modulated by AKAP12 HSC-specific phosphoediting ( Figure 7A, supplementary table S5). BIP levels by western blotting were induced in CCl4 livers and inhibited by AKAP12 phospho-editing, confirming the proteomics result ( Figure 7C). Like the proteomics data, we did not find any change in total IRE1α expression. However, phospho-activated IRE1α was suppressed by AKAP12 phospho-editing in total liver ( Figure 7C). Since ER stress induces inflammatory signals in different systems (11), we examined whether the HSCs from our CRISPR mouse model exhibited altered inflammatory signaling upon AKAP12 phospho-modulation. Out of the known HSC cytokines, we found the pro-inflammatory cytokine, IL-17, IL-6 and IL-β to be strongly induced in CCl4-HSCs whereas AKAP12-phospho-edited HSCs suppressed their expression ( Figure 7D). On the other hand, an anti-inflammatory cytokine, IL-10 was suppressed in HSCs by CCl4 administration, and its expression was restored by AKAP12 phospho-editing ( Figure 7D). To examine whether ER stress modulation within activated HSCs was transmitted to other liver cell types, we evaluated crosstalk between HSCs and hepatocytes in a co-culture system where AKAP12 was CRISPR-edited. Co-culture with activated HSCs induced the ER stress response markers BIP and induced IRE1α phosphorylation in hepatocytes compared to coculture with quiescent HSCs ( Figure 7E). Co-culture with activated HSCs in which AKAP12 was phospho-edited (CR) reduced the ER stress signal in hepatocytes compared to activated HSCs alone whereas hepatocytes cocultured with quiescent HSCs with CR did not exhibit any alteration in ER stress markers compared to WT ( Figure   7E). Original blots for figure 7 are shown in Figure 7-source data 1 to 4.

Discussion
In fibrotic mouse and human livers, HSCs exhibit increased AKAP12 phosphorylation and decreased AKAP12 scaffolding activity towards the collagen chaperone, HSP47. By mapping the phosphorylation events that are altered upon activation of human or mouse HSCs, we have demonstrated that phosphorylation of specific S or T residues of AKAP12 is triggered during HSC activation. Hence, we named these sites as activation-responsive phospho-sites. Out of the 5 activation-responsive phospho-sites, 4 serine residues were confirmed as PKCα substrates. Mutagenesis analysis on recombinant AKAP12 showed that the S687 and S688 were stronger PKCα substrates compared to S676 and S678 because their mutations drastically suppressed phosphorylation. We further observed that phosphorylation of AKAP12 by PKCα suppressed direct binding between AKAP12 and HSP47. Confirming this recombinant data, we observed that silencing PKCα in HSCs induced the binding between AKAP12 and HSP47. AKAP12's known scaffolding activities towards CCND1, PLK1 and PKCα that were identified previously are regulated by its phosphorylation (13, 16 21). We therefore evaluated the role of site-specific phosphorylation in modulating AKAP12's scaffolding functions in HSCs.
Using a CRISPR-based gene editing approach, we deleted AKAP12's phosphorylation sites in culture-activated human or mouse HSCs and observed an enhancement in AKAP12's interaction with HSP47, a strong inhibition of HSC activation (judged by α-SMA levels) and restoration of the quiescent marker, vitamin A that is suppressed in activated HSCs (25). HSP47 resides in the ER (4) and since AKAP12 interacted with HSP47, we evaluated whether it co-localized with HSP47 in the ER and whether the AKAP12-HSP47 scaffold in the ER was affected by CRISPRediting its phosphorylation sites. The AKAP12-HSP47 scaffold was induced in the ER upon AKAP12 phosphoediting. HSP47's chaperoning activity towards collagen is highly induced during HSC activation and this allows increased maturation and secretion of collagen (4). Since AKAP12 binds to HSP47, we examined whether this interaction regulated HSP47's collagen chaperoning function. The ER of activated HSCs stained strongly for the collagen-HSP47 scaffold but AKAP12 phospho-site editing diminished the collagen scaffolding activity of HSP47.
Our findings suggest that lack of AKAP12 activation-responsive phosphorylation quenches HSP47's collagen chaperoning activity and prevents HSC activation.
HSC activation is a hallmark of liver fibrosis. The fact that enhanced phosphorylation of AKAP12 at its activationresponsive phospho-sites promotes HSC activation fueled our hypothesis that site-specific AKAP12 phosphorylation may be involved in promoting liver fibrosis in animal models. To address this hypothesis, we designed CRISPR-AAV vectors to perform gene editing of AKAP12's activation-responsive phospho-sites specifically in HSCs of mouse liver. This was achieved by expressing the CRISPR causing enzyme, SaCas9 under control of the HSC-specific promoters, GFAP or LRAT (26,27). Both GFAP and LRAT specifically expressed SaCas9 in HSCs, but GFAP-driven SaCas9 was increased in activated HSCs compared to normal HSCs whereas the reverse was observed with LRAT-Cas9. GFAP promoter activity is induced during HSC activation (31) whereas LRAT expression is known to be suppressed (32). This might have been responsible for the different effects of these two promoters. The AAV particles of serotype 6 were used because AAV6 efficiently transduces activated HSCs in the CCl4 mouse model (33). HSC-specific gene editing of AKAP12 was performed by deleting the DNA region corresponding to the 5 phospho-sites (PDEL). AKAP12 phospho-site editing by this PDEL mechanism strongly inhibited HSC activation, enhanced the AKAP12-HSP47 scaffold, and suppressed the collagenchaperoning activity of HSP47 leading to decreased collagen production in the liver. To confirm the involvement of AKAP12 phosphorylation at these residues in promoting pro-fibrogenic phenotype, we inhibited phosphorylation at these sites by CRISPR-mediated editing of the S/T residues to A (PMUT). The overall editing efficiency caused by PMUT was lower than that of PDEL in activated HSCs from CCl4-exposed livers. Despite the less efficiency of PMUT editing, it was effective in suppressing the fibrogenic response in the liver, supporting an important role of AKAP12 phosphorylation in regulating the outcome of liver fibrosis. As opposed to fibrotic livers, CRISPR editing in HSCs of normal liver did not alter the molecular identity of the liver. Since normal HSCs do not exhibit phosphorylation of AKAP12 at the activation-responsive phospho-sites, they appear to be unaffected by modulating these sites. This control data reiterates the fact that increased AKAP12 phosphorylation caused by HSC activation has pro-fibrogenic effects.
Apart from suppression of fibrotic parameters, we observed that AKAP12 phospho-modulation in HSCs inhibited collagen mRNA levels and globally suppressed liver injury. Since inhibition of collagen transcription and overall liver injury may not be due to AKAP12's ability to regulate HSP47's collagen-chaperoning activity, we searched for additional mechanisms of action of phospho-AKAP12. We performed proteomics analysis in HSCs from our CRISPR model and total liver from the same to identify molecular signals altered by AKAP12 phospho-editing. In HSCs, we identified BIP/GRP78, a regulator of the IRE1α branch of UPR signaling and a known collagen chaperone in the ER (5,34). Interestingly, a recent interactome study identified HSP47 as a binding partner for IRE1α (5). IRE1α is an ER transmembrane kinase that is kept in inactive state by its binding to BIP. HSP47 activates IRE1α oligomerization and phosphorylation by displacing BIP and triggering the UPR response during ER stress (5). The functional effect of the HSP47-IRE1α interaction on UPR signaling and collagen folding during fibrogenic stimulation in HSCs is undescribed so far. But IRE1α activation caused by ER stress inducers in HSCs is known to enhance collagen transcription as well as collagen protein expression through activation of p38MAPK and SMAD pathways (30). HSCs exhibit ER stress and UPR signaling in response to liver injury stimuli (11,35). In fact, ER stress appears to be both a cause and effect of HSC activation (11,36). Since phospho-edited AKAP12 interacted with HSP47 in the ER of HSCs, we wondered whether HSP47-mediated UPR signaling might be regulated by AKAP12. We found that IRE1α-HSP47 interaction (5) was enhanced in CCl4 HSCs and so were downstream pathways known to be enhanced by IRE1α activation in HSCs (phospho-P38MAPK and SMAD2/3) (30).
Interestingly AKAP12 phospho-editing suppressed HSP47's UPR-activating function by quenching the CCl4mediated IRE1α-HSP47 interaction in HSCs that further inhibited IRE1α phospho-activation preventing downstream P38MAPK and SMAD signaling in these cells. Another component of the UPR signaling we found from proteomics was BIP. BIP is a collagen chaperone that also inactivates IRE1α under basal conditions (5).
During ER stress, HSP47 displaces BIP from IRE1α, activating IRE1α-mediated UPR signaling (5). Whether this HSP47-mediated BIP displacement promotes BIP's activity as a collagen chaperone during HSC activation or liver fibrosis is unclear so far. We hypothesized that AKAP12 by virtue of its increased phosphorylation and lack of scaffolding towards HSP47 may regulate the BIP-IRE1α-HSP47 axis and promote BIP's collagen chaperoning function. Indeed, we observed increased interaction of BIP with collagen in HSCs of CCL4 livers that was suppressed by AKAP12 phospho-editing. We could not find any interaction between AKAP12 and BIP in HSCs but speculate that loss of AKAP12-HSP47 scaffolding leading to increased HSP47-IRE1α interaction might have released BIP from the IRE1α sites and favored BIP-collagen scaffolding.
Enhanced protein secretion is associated with ER stress and UPR signaling in activated HSCs and is crucial for processing of inflammatory proteins and ECM components upon pro-fibrogenic stimulation (11). Studies in liver and other systems support the role of ER stress in promoting inflammatory signaling (37,38). Also, inflammatory proteins have a less well described role in promoting ER stress and UPR signaling (38). ER stress is therefore both a cause and consequence of inflammatory signaling (38). Cytokines such as IL-1β are known to be induced in activated HSCs through ER stress (39). Other cytokines known to be expressed by HSCs, IL-17 and IL-6 (40,41) are prone to modulation by ER stress (42,43). The anti-inflammatory and anti-fibrotic cytokine, IL-10 expressed by HSCs (44) was recently shown as a target of ER stress in macrophages (45). IL-6 and IL-1β are mediators of ER stress in the liver (46). In pancreatic beta cells, IL-1β is known to induce ER stress in a nitric oxide-dependent manner (47). The anti-inflammatory effect of IL-10 has been shown to block ER stress in intestinal epithelial cells (48). Since our data on AKAP12 suggests that it regulates ER stress pathways in HSCs, we tested whether known inflammatory signals linked to ER stress were also regulated by AKAP12. We found the pro-inflammatory cytokines, IL-17, Il-1β and Il-6 to be induced in CCl4-HSCs whereas AKAP12-phospho-edited HSCs exhibited a strong suppression of these cytokines. On the other hand, the anti-inflammatory cytokine, IL-10 was suppressed in HSCs by CCl4 administration, and its expression was restored by AKAP12 phospho-editing. Literature suggests that inflammatory molecules and UPR signaling may contribute to increased collagen transcription during liver fibrosis.
Pro-inflammatory IL-6 signaling induces collagen transcription (49) and loss of anti-inflammatory signals such as IL-10 inhibit it (50). The IRE1α-directed UPR also induces collagen transcription through increased p38MAPK and SMAD2/3 signaling (30). Since AKAP12 phospho-editing suppressed IRE1α-directed UPR signaling through its association with HSP47 and regulated ER stress-linked cytokines expressed in HSCs, these factors may have contributed to the overall drop in collagen mRNA levels.
Since ER stress/UPR signaling plays a role in enhancing liver injury and the ER stress inducer, tunicamycin is known to induce ALT/AST levels (51), we examined whether AKAP12 HSC-specific editing regulated the liver ER stress response. We found dysregulation of ER stress and UPR-associated components in total liver of CCl4 mice (BIP and other ER foldases such as protein disulfide isomerases, PDIA1 and PDIA6) that were regulated by HSCspecific AKAP12 phospho-editing. Induction of BIP expression in the liver was normalized by AKAP12 phosphoediting. Although the total IRE1α levels were unchanged by CCl4, IRE1α phospho-activation was inhibited by AKAP12 HSC-specific phospho-editing. These results suggest that controlling the ER stress response/UPR signaling within HSCs during pro-fibrogenic stimulation also modulates the same in the whole liver. The phenomenon of ER stress being communicated from stressed cells to other cells within a tissue has been reviewed in the context of cells that produce large amounts of proteins such as immune cells (37). It has also been published that ER stress invokes liver fibrosis primarily due to ER stress within HSCs due to their activation (36). Since hepatocytes are known to be sensitive to CCl4-mediated ER stress (52), we examined whether crosstalk between activated HSCs and hepatocytes in a co-culture system promoted the ER stress response in hepatocytes and whether AKAP12 regulated this crosstalk. Modulating HSC activation through AKAP12 regulated the ER stress response in hepatocytes in culture. Since we observed regulation of ER stress-linked inflammatory cytokine production from HSCs of AKAP12 CRISPR edited livers, we think that inflammatory cytokines from HSCs might transmit ER stress to the whole liver and AKAP12 provides a mode to control these effects during fibrogenesis. A schematic of our findings and their implications in cellular crosstalk during fibrogenesis are summarized in Figure   7F.
In summary, we have identified AKAP12 as a scaffolding partner of HSP47 in normal HSCs that controls HSP47's collagen chaperoning activity and its interaction with UPR signals in HSCs. Site-specific phosphorylation of AKAP12 occurs during HSC activation and this modification inhibits its interaction with HSP47. This induces HSP47's collagen chaperoning activity, collagen production and HSP47's interaction with UPR signaling proteins upon pro-fibrogenic stimulation. Blocking AKAP12 phospho-modification inhibits HSC activation, collagen production, fibrosis as well as overall liver injury possibly via modulation of the ER stress response and inhibition of ER stress-linked inflammatory signals. Structural studies to identify how AKAP12's activation-responsive phospho-sites interact with HSP47 will facilitate design of small molecules to block AKAP12 phosphorylation and enhance its HSP47 scaffolding activity. Since AKAP12 phospho-modification is not evident in normal HSCs but is induced upon pro-fibrogenic stimulation, AKAP12 phosphorylation may be utilized as a druggable target in liver fibrosis.

Primary cell isolation and culture
Primary human HSCs purchased from ScienCell, Incorporation, CA) were cultured on plastic dishes for 6 hours (day 0) or further cultured till activation (day 5 or day 7). Mouse HSCs or hepatocytes were isolated from 3-4 months old C57BL/6 mice according to our previously established protocols (24). Mouse HSCs were culture-activated on plastic dishes like human HSCs.

Phospho-peptide mapping
AKAP12 was immunoprecipitated from HSCs or hepatocytes using an AKAP12 antibody-conjugated protein A/G column (Thermo Scientific). The AKAP12 beads were submitted to Applied Biomics, CA for phospho-peptide mapping. Tryptic peptides were enriched for phospho-peptides and processed for detection of a phospho-site by mass spectrometry. Phosphorylated residues were confirmed by mass spectrometry peak showing the neutral loss of phosphate that was detected from peak shifts on MS/MS spectrum (Table 1, table supplement

CRISPR gene editing in cultured HSCs
CRISPR-Cas9 mediated gene editing at the AKAP12 gene locus (exon 3) to delete the region of its activationresponsive phospho-sites was performed by homology-directed repair (HDR). A 22-bp small guide RNA sequence (sgRNA) upstream of a protospacer adjacent motif (PAM-5'-GTGGAT-3') recognized by saCas9 (PAM consensus-NNGRRT where N=any nucleotide, R=A or G) (53), was designed and synthesized using the Edit-R CRISPR system (Horizon Discovery, Colorado) (human guide sequence, supplementary table 1). The CRISPR design tool was used to determine the sgRNA whose sequence is unique compared to the rest of the genome to avoid offtarget effects. A donor RNA to delete the phospho-region was designed and synthesized using the Edit-R HDR donor designer system (Horizon) (human PDEL HDR donor, supplementary table 1). The sgRNA was stabilized by 2'-O-methyl nucleotides and phosphorothioate linkages in the backbone on both the 5′ and 3′ ends and the HDR donor was stabilized by phosphorothioate linkages on both ends to improve functionality during transfection.
Cultured cells were co-transfected with a commercially available plasmid, AAV6-GFAP-saCas9, containing the SaCas9 gene under control of the GFAP promoter (Vector Biolabs, PA), sgRNA and HDR donor RNA using the DharmaFECT Duo Transfection Reagent that allows co-transfection of RNA and DNA (Horizon). Cells with transfection reagent alone or SaCas9 plasmid alone+transfection reagent were used as controls. CRISPR designs for mouse HSCs were performed as above for human with mouse guide sequence #1 and mouse PDEL HDR donor (Supplementary table 1). After 48-72 hours of transfection, genomic DNA from human or mouse HSCs was amplified by multiplex PCR using two primers to amplify the region around the deletion site and a third deletionspecific primer to detect HDR-mediated gene editing.

Carbon-tetrachloride (CCl4) injection in mice
12-week-old C57BL/6 male mice were injected intraperitoneally with CCl4 (HPLC grade, Cat#270652, Sigma, diluted 1:3 in mineral oil) or mineral oil (control) at 1 ul/gram body weight bi-weekly for 5-weeks. All procedures for the care and use of mice were approved by the Institutional Animal Care and Use Committee at Cedars-Sinai Medical Center (CSMC).

CRISPR gene editing in mice
HDR-based gene editing in control or CCl4 mice was performed according to the scheme in figure 4A and B. Two 22-bp sgRNA sequences upstream of a saCas9 PAM (53), were designed using the Edit-R CRISPR system (Horizon Discovery). Off-target analysis for the two sgRNA was performed using the algorithm from the Benchling [Biology Software-(2022) retrieved from https://benchling.com] (supplementary table S6). The two sgRNA sequences were cloned into a single AAV6 vector under the control of a U6 promoter by the cloning service available from Vector builder Inc., IL. An AAV6 vector containing a non-targeting sgRNA was used as an empty vector control (EV). The sequence corresponding to a PDEL or PMUT donor with 500 bp flanking either side of the target region was cloned into a separate AAV6 vector. The PAM sequence in these donors was mutated to prevent re-cleavage by SaCas9 after HDR. The AAV6-GFAP-SaCas9 vector (Vector Biolabs) was used for HSC-specific gene editing. In addition, another AAV6-LRAT-SaCas9 vector was prepared by cloning the mouse LRAT promoter (Accession ID: NM_023624) upstream of SaCas9 (Vector Builder). AAV6 particles of the sgRNA construct, EV construct, PDEL/PMUT donors and GFAP/LRAT-SaCas9 were purified using Vector builder's AAV production service. For each viral vector, titer was determined by real-time PCR using primers specific for the AAV inverted terminal repeats (ITR). A titer of 1-2X10 13 genome copies (GC)/ml was achieved for each AAV. All vectors tested negative for mycoplasma contamination. EV or sgRNA vectors along with PDEL or PMUT donors and GFAP or LRAT SaCas9, were injected into tail vein of mice at 10 11 GC/vector in a volume of 100 ul PBS. Viral vectors were injected into oil or CCl4 mice during the 2 nd and 4 th week of oil or CCl4 administration ( Figure 4B). The HSC specificity of CRISPR was determined by SaCas9 immunofluorescence as described under the immunostaining section. The efficiency of CRISPR editing in HSCs and hepatocytes of gene-edited livers was evaluated by next generation amplicon sequencing. A 298 bp PCR product was amplified from genomic DNA using primers that recognized regions upstream and downstream of the site of AKAP12 deletion or mutation. Amplicons were purified from gels and submitted to Azenta Life Sciences Inc., CA. for performing next generation amplicon sequencing. Briefly, illumina adaptor sequences (FW: 5'-ACACTCTTTCCCTACACGACGCTCTTCCGATCT-3',REV: 5'-GACTGGAGTTCAGACGTGTGCTCTTCCGA TCT-3' were added to the amplicons and sequenced by Azenta illumina platform sequencers. The wild type and mutant or deletion mutant reads were counted from each sample and the efficiency of editing was the percentage of edited reads (PDEL or PMUT) versus the total reads.
Frequencies of on-target and off-target base changes were analyzed by comparing the target reads to reference reads corresponding to the WT Akap12 amplicon between the two sgRNA sequences ( Figure 4A). Within this region, any mismatches other than PDEL or PMUT were considered as off-targets. The mismatches to the reference were observed mainly outside the target region at a frequency of 5% or less ( Figure 4G).

Human tissue array
The human tissue array (Cat# XLiv086-01) in the form of paraffin-embedded tissues was purchased from the human tissue biorepository, US Biolabs Inc. MD. Arrays were stained by immunostaining as described below.

Real-time RT-PCR
Total RNA from cells or tissues was reverse transcribed to cDNA using M-MLV reverse transcriptase (Nxgen).
CDNA was subjected to quantitative RT-PCR using TaqMan probes for mouse Akap12 and the housekeeping gene, Gapdh (mouse) (Life technologies) (24). The PCR profile was: initial denaturation: 95°C for 3 minutes, 45 cycles: 95°C, 3 seconds; 60°C, 30 seconds. The cycle threshold (Ct value) of the target genes was normalized to that of control gene to obtain the delta Ct (Ct). The Ct was used to find the relative expression of target genes according to the formula: relative expression= 2 -Ct , where Ct= Ct of target genes in experimental condition -Ct of target gene under control condition.

Co-immunoprecipitation and western blotting
Total protein extract was processed for immunoprecipitation by incubating 200 µg of pre-cleared protein with with 2 µg of antibody as we described previously (24). Immunoprecipitated protein was processed for western blotting as previously published (24) and developed with Clean-blot IP detection reagent (HRP) (Thermo Scientific, IL).
Antibodies used for western blotting are listed in supplementary table S7.

Vitamin A autofluorescence
UV-excited autofluorescence of human HSCs was captured by fluorescence microscopy using a Keyence BZ-X710 inverted fluorescent microscope (Itasca, IL) as we described previously (24).

Site-directed mutagenesis
An expression vector (pReceiver-WG16) containing the human AKAP12 gene under control of the T7 promoter was purchased from Genecopoiea, MD and mutated at AKAP12's activation-responsive sites (S/T to A mutations) using the QuikChange II® site-directed mutagenesis Kit (Stratagene, CA) as we described previously (54).

In vitro kinase assay, phostag analysis and binding studies
The T7-AKAP12 vector or its mutants were in vitro translated using the non-radioactive TNT® Coupled Transcription/Translation system containing rabbit reticulocyte lysate (RRL) and a biotin-lysyl tRNA according to the manufacturer's instructions (Promega, WI) to incorporate biotin label into the translated AKAP12 protein.
Biotinylated AKAP12 was purified from the RRL components using a biotin-antibody column. Biotinylated AKAP12 or its mutants (5 ul) were used as a substrate for phospho-PKCα in a 25 µl in vitro kinase reaction using 100 ng of active recombinant PKCα enzyme (Millipore-Sigma, MA), 5 µl of a lipid activator (Millipore-Sigma; 20mM MOPS, pH 7.2, 25mM β-glycerolphosphate, 1mM sodium orthovanadate, 1mM dithiothreitol, 1mM CaCl2), 3 µl of Mg 2+/ ATP cocktail (Millipore-Sigma, 20mM MOPS, pH 7.2, 25mM β-glycerophosphate, 5mM EGTA, 1mM Na3VO4, 1mM dithiothreitol, 75mM MgCl2, and 0.5mM ATP) and 2.5 ul of 20 mM Hepes-NaOH buffer, pH7.6. The reaction was carried out at 30°C for 2 hours. The kinase reaction was run on a zinc phostag gel containing 15 uM phostag gel (Fujifilm Wako Chemicals, VA) to separate phosphorylated form of AKAP12 from its unphosphorylated counterparts as we described earlier (54). Membranes were probed with streptavidin-HRP (supplementary table   S7) to detect biotinylated AKAP12. Biotin antibody was conjugated to protein A/G plus agarose columns using a coupling buffer according to the crosslink immunoprecipitation kit (Thermo Scientific) followed by binding of recombinant biotinylated AKAP12. The columns were treated with recombinant HSP47 protein the absence or presence of active PKCα enzyme. Bound proteins were eluted from the washed column using elution buffer from the crosslinking immunoprecipitation kit (Thermo Scientific) and run on gels along with biotinylated AKAP12 as input and antibody-bound protein A/G beads as IgG controls. Blots were incubated with HSP47 antibody followed by Clean-blot IP detection. Reverse IP was done by following the same protocol using HSP47 antibody columns treated with biotinylated AKAP12 followed by detection with streptavidin-HRP. Recombinant HSP47 input was purchased from Prospec protein specialists, NJ.

Duolink proximity ligation assay and immunostaining procedures
For immunocytochemical procedures, cells were fixed with paraformaldehyde and then permeabilized with triton-X 100 before antibody staining. For immunohistochemical analysis, tissues were de-paraffinized and antigen retrieval was performed using the citrate-based antigen unmasking solution (Vector laboratories, CA). For phosphodetection using proximity ligation assay (PLA), primary AKAP12 or phospho-serine (PSer) antibodies (supplementary table 6) were directly conjugated to PLA minus or plus complementary oligonucleotide arms (PLA minus, Catalog no. DUO92010; PLA plus, Catalog no. DUO92009, Millipore-Sigma) according to our previously published protocol (24). To examine protein-protein interactions in cells or tissues, samples were incubated with the antibodies for the interacting targets at 4°C overnight (AKAP12-HSP47, HSP47-collagen). After washing the unbound antibodies, samples were further incubated overnight with secondary antibodies (rabbit or mouse) that were bound to PLA plus or minus complementary probes (Millipore-Sigma, supplementary table S7). The PLA probes were ligated when the proteins were in proximity due to their interaction giving a fluorescent signal as we previously reported (24). To evaluate the localization of interacting partners, co-immunostaining of the PLA signals was done with HSC (desmin) or subcellular compartment (calreticulin ER) marker antibodies. Marker antibodies were detected by Alexa fluor® green rabbit or mouse secondary antibodies (Abcam, supplementary table 6). Colocalization of SaCas9 with desmin or albumin markers in liver tissue was detected by Alexa fluor® secondary antibodies (supplementary table S7). AKAP12 expression in tissues was detected using the mouse HRP/DAB detection immunohistochemistry kit (cat # ab64264, abcam).

Histopathological examination
Liver sections fixed with 10% neutral formalin were processed for paraffin embedding, sectioning, hematoxylin and eosin (H&E) and picrosirius red staining (collagen) using the services provided by the liver histology core of the University of Southern California research center for liver diseases (NIH grant P30 DK048522).

Hydroxyproline measurement
The hydroxyproline content of tissue was measured following the protocol from the hydroxyproline assay kit (Cell Biolabs Inc., CA). Briefly, 10 mg of liver tissue was homogenized, and acid hydrolysis was done with 12N HCl.
Hydrolyzed samples were treated with chloramine T to convert the hydroxproline to a pyrrole. Ehrlich's reagent or 4-(Dimethylamino) benzaldehyde added to the pyrrole reacted with it to produce a chromophore whose absorbance could be read at 540-560 nm. The content of hydroxyproline in the tissue sample was determined by comparison to a hydroxyproline standard from the kit that was processed like the unknown sample.

ALT/AST measurement
ALT and AST levels from plasma of mice were measured with the ALT and AST colorimetric activity assay kits (Cayman Chemical, MI). ALT activity was measured by monitoring the rate of NADH oxidation in a coupled reaction using lactate dehydrogenase (LDH). The NADH to NAD+ oxidation caused a decrease in A340 nm absorbance.
The rate of decrease (A340/min) is directly proportional to the ALT activity. AST activity was measured by the rate of NADH oxidation in the presence of malate dehydrogenase. NADH to NAD+ conversion caused a decrease in A340 nm absorbance. Lactate dehydrogenase was added to the AST reaction to prevent interference from endogenous pyruvate in the plasma. The A340/min for both ALT and AST were converted to units/L by dividing the A340 values by the NADH extinction coefficient and multiplying by the sample dilution factor as per the protocol instructions (Cayman).

Proteomics analysis
Total protein from liver or HSCs was subjected to mass spectrometry-based proteomics analysis by the services of Poochon proteomics solutions, MD. The Nanospray LC/MS/MS analysis of tryptic peptides for each sample was performed sequentially with a blank run between each two sample runs using a Thermo Scientific Orbitrap Exploris peptide length was specified to be five amino acids. The precursor mass tolerance was set to 15 ppm, whereas fragment mass tolerance was set to 0.05 Da. The maximum false peptide discovery rate was specified as 0.05.
The resulting Proteome Discoverer Report contains all assembled proteins with peptides sequences and peptide spectrum match counts (PSM#). The PSM count is a measure of the abundance of the protein.

Statistical analysis
Western blotting data was quantified by densitometry of blots using the ImageJ software (NIH). PLA staining and immunofluorescence data was analyzed in a blinded manner by two individuals and quantified using ImageJ according to published protocols (55). Scatter bars showing individual experimental points and their means were plotted using GraphPad Prism 9.3.0, GraphPad software. Biologically independent replicates combined from at least three individual experiments were represented as mean ± standard error (Mean±S.E.). Statistical analysis was performed using two-tailed Student's t-test for paired comparisons and one-way ANOVA (GraphPad Prism) for comparing differences between multiple groups. Significance was defined as P<0.05. Actual p value for each comparison is listed in figure legends.                  The human tissue arrays containing 11 normal livers and 16 liver fibrosis tissues was stained with PLA probes to detect AKAP12-HSP47 interaction and Alexa fluor® probes to detect HSP47 (green) or AKAP12 (red) as described under methods. A representative staining from this complete panel and quantitation by Image J is shown in figure 1D.