Hepatocyte-like cells reveal novel role of SERPINA1 in transthyretin amyloidosis

Transthyretin amyloidosis (ATTR) is a rare, progressive disease caused by mutations of transthyretin (TTR). ATTR is thought to result from the formation of amyloid fibrils, primarily affecting the peripheral autonomic nervous system and the heart (Conceição et al., 2016). More than 100 disease-causing mutations of the TTR gene have been described. TTR displays extensive β-sheet structures prone to self-oligomerization (Blake and Serpell, 1996). TTR is a tetrameric protein and it is assumed that kinetic instability of mutant proteins favors fibril formation (Foss et al., 2005). TTR synthesis and secretion into the blood predominantly takes place in the liver, while minor sites of synthesis (<5%) include the choroid plexus, retinal pigment epithelium, pancreas, placenta and the Schwann cells (Buxbaum et al., 2014; Cavallaro et al., 1990; Schreiber, 2002). TTR functions as a plasma protein responsible for the transport of thyroxine (T4) and the retinol-binding protein (RBP4), which acts as a carrier for retinol (vitamin A) (Monaco et al., 1995; Neumann et al., 2001). TTR^−/− mice, however, do not show significant signs of disease and are fertile (Wei et al., 1995).

The impact of the protein quality control (PQC) system has been associated to the pathogenesis of amyloidotic diseases, e.g. in amyotrophic lateral sclerosis (ALS), Alzheimer's and Parkinson's disease (Daturpalli et al., 2013; Park et al., 2018; Wang et al., 2014). Several lines of evidence suggest that PQC, operative during the synthesis of TTR in the hepatocyte, is also an important determinant of ATTR. In a pioneering work, biophysical and secretion assays suggested the contribution of endoplasmic reticulum (ER)-assisted folding by cell type-specific chaperones (Sekijima et al., 2005). TTR mutations ATTRD18G and ATTRA25T, both associated with late-onset ATTR of the central nervous system, are found at low concentrations in the blood (Hammarström et al., 2003; Sekijima et al., 2003). For ATTRD18G, low secretion efficiency has been observed in HEK cells, suggesting that ER retention of misfolded TTR results in the induction of the unfolded protein response (UPR), including upregulation of the ER chaperone glucose-regulated protein 78 (GRP78) (Sato et al., 2007). More recently, proteome analysis (2D-DIGE) following expression of the TTR mutation ATTRL55P in yeast revealed the induction of chaperones that mediate protein folding, including members of the HSP70, peptidyl-prolyl-cis-trans isomerase (PPIase) and ubiquitin-like modifier (SUMO) families (Gomes et al., 2012). Several TTR mutations were shown to be subjected to ER-associated degradation (ERAD) pathways (Sato et al., 2012). Multi-dimensional protein identification technology (MudPIT) profiling of adipose tissue has suggested the overexpression of several chaperones in ATTR patients (Brambilla et al., 2013). Of note, proteome analyses of plasma from ATTRV30M patients indicated altered expression of several extracellular chaperones (da Costa et al., 2015).

The protease inhibitor SERPINA is also highly expressed in the liver and strongly secreted into blood (Topic et al., 2012). SERPINA1 deficiency results in a genetic disorder resulting in damage of the liver and the lung due to uncontrolled neutrophil elastase and non-functional SERPINA1 in the liver (Radlovic et al., 2014). Direct interaction of TTR and SERPINA1, however, has not yet been described.

Whereas previous studies analyzed whole liver, human serum, adipose tissue or a variety of standard cell culture lines, such as HEK293 cells, to assess the impact of PQC on gene expression, we took advantage of hepatocyte-like cells (HLCs) differentiated from induced pluripotent stem cells (iPSCs). Gene expression of chaperones related to amyloidosis was explored by real-time qPCR analysis in HLCs with the intrinsic genetic background of ATTR patients, i.e. in HLCs derived from ATTR patients (ATTR-HLCs). TTR knockdown in HLCs was employed to study the relation of TTR and individual chaperone gene expression. We identified SERPINA1 as a highly regulated chaperone in ATTR-HLCs. Further analyses revealed a physical interaction of SERPINA1 with TTR as well as its interference with TTR aggregation, suggesting that SERPINA1 represents a novel target for therapy of ATTR.

In this study, ATTR-HLCs were used to assess chaperone gene expression. HLCs from healthy individuals (H-HLCs) served as control. The characteristics of the four ATTR cell lines and the four cell lines derived from healthy individuals are given in Table S1. We used a protocol for reprogramming and differentiation that has been shown to be very robust in terms of variability (Niemietz et al., 2016). After reprogramming of primary patient cells, the cells displayed typical iPS cell colony growth (Fig. 1A), high immunofluorescence staining of pluripotent stem cell markers OCT4, NANOG, SSEA-4, and TRA-1-60 (Fig. 1B), high mRNA levels of embryonic stem cell-specific markers OCT4, NANOG and SOX2 (Fig. 1C), and expected expression levels in the trilineage hPSC Scorecard™ Assay (Fig. 1D) suggesting that typical iPSCs have been derived from ATTR patients and healthy individuals. After a 14 day differentiation protocol of iPSCs, the cells showed their characteristic polygonal shape and high nuclear to cytoplasmic ratio that is typical of HLCs (Fig. 1E), overall high immunofluorescence staining of albumin, HNF4A and TTR (Fig. 1F), high glycogen storage (Fig. 1G), and high expression levels of the hepatocyte marker genes serum albumin (ALB), apolipoprotein A1 (APOA1), ATPase copper transporting Beta (ATP7B), nuclear constitutive androstane receptor (NR1I3, also known as CAR) and transferrin (TF) (Fig. 1H). These data indicate that phenotypes, function as well as expression levels of hepatocyte marker genes are almost identical in iPSCs and HLCs derived from ATTR patients and healthy controls, when cells were subjected to the reprogramming and differentiation protocols. Importantly, variation was low between expression of hepatic marker genes in HLCs derived from H-HLCs and ATTR-HLCs when using at least two clones per line (±s.e.<1.1) (Fig. S1).

Since the relation of TTR to chaperone gene expression was of high importance to our study, we further assessed whether overall TTR expression might differ between ATTR-HLCs and H-HLCs. TTR mRNA and TTR protein secreted into the cell culture medium by HLCs did not differ between both groups (Fig. 2A,B). Of note, TTR mRNA levels observed in HLCs were similar to those in primary human hepatocytes, suggesting that HLCs model human hepatocytes (Niemietz et al., 2016). We also inspected the different forms of TTR by western blot analyses and did not observe differences between both groups of HLCs (Fig. 2C), suggesting that any deviation that might be observed for chaperone gene expression between ATTR-HLCs and H-HLCs was not biased by the level of TTR expression.

Having verified that TTR appears to be synthesized and secreted in a similar manner and magnitude in both groups of HLCs, RT-qPCR analyses of chaperone genes were performed. In total, 39 genes were investigated (Table S2). Selection of genes was based on PubMed searches for genes reported to be associated with protein folding, degradation, ubiquitylation and translational arrest in amyloidotic dieseases, such as ATTR, amyotrophic lateral sclerosis (ALS), Alzheimer's disease, amyloid light-chain (AL) amyloidosis and amyloid A (AA) amyloidosis. All genes were first validated in human hepatoma HepG2 cells and found to be robustly expressed (data not shown). The expression of the chaperone genes was categorized according to a threshold of fold-change (FC) that takes into account the variability of hepatocyte marker expression (FC<±2.5). A FC≥±5 (two times that of the mean hepatic marker gene variability) was chosen to indicate significance of chaperone gene expression between groups. Of note, differences in individual chaperone gene expression in the cell lines were low (±s.e.<1.3) (Fig. S2). Five chaperone genes, i.e. SERPINA1, fibrinogen alpha chain (FGA), fibrinogen beta chain (FGB), alpha-2-macroglobulin (A2M) and heat shock protein beta-1 (HSPB1), were found to differ in at least two ATTR-HLC lines, whereas 34 genes showed expression levels below the threshold (Fig. 3). Expression levels of SERPINA1 and FGA differed in three ATTR-HLC lines. Upregulation was found for FGA (3 ATTR-HLC lines), A2M (2 ATTR-HLC lines) and FGB (2 ATTR-HLC lines). Downregulation was observed for HSPB1 (2 ATTR-HLC lines). SERPINA1 displayed upregulation in one ATTR-HLC line and downregulation in two ATTR-HLC lines. However, variation in SERPINA1 expression was very low in the four H-HLC lines suggesting that variability due to different donors to generate HLCs is generally low (Fig. S3). Of note, four of the genes (SERPINA1, FGA, A2M and FGB) belong to a group of seven chaperones that are predominantly located extracellularly (da Costa et al., 2015). In contrast, 32 chaperones that have been reported to predominantly show intracellular localization did not exceed a FC≥±5, indicating different expression levels in ATTR-HLCs as compared to H-HLCs (Fig. 3).

Chaperone gene expression in HLCs could be associated to the overall level of TTR expression. We used small interfering RNAs (siRNA) and antisense oligonucleotides (ASO) for knockdown of TTR expression in the HLCs (Niemietz et al., 2016). As expected, TTR downregulation was almost identical in ATTR-HLCs and H-HLCs resulting in an mRNA knockdown of 75% to 85% and a protein knockdown of 67% to 85% (Fig. S4). The expression of the five chaperones (SERPINA1, FGA, A2M, FGB and HSPB1) that were found to be differently regulated in at least two ATTR-HLC lines (Fig. 3) were examined for the effect of TTR knockdown. First, the correlation coefficient of TTR mRNA before and after knockdown was determined in H-HLCs and ATTR-HLCs (Fig. 4A,B). Whereas no correlation (r<0.1) was observed for HSPB1, a high correlation (r>0.7) was observed for the other chaperone genes prior TTR knockdown in H-HLCs and ATTR-HLCs. However, after TTR knockdown, the correlation coefficient significantly (r>0.5) dropped in ATTR-HLCs. Inspection of the chaperone mRNA level after TTR knockdown revealed a significant increase of SERPINA1 in ATTR-HLCs, whereas no induction was observed for H-HLCs (Fig. 4C), and the other chaperone genes were only marginally affected by TTR knockdown in both HLC groups. Patient-specific analysis of TTR mRNA expression in the four ATTR-HLC lines indicated an upregulation of SERPINA1 mRNA in three out of four patient-specific HLC lines (Fig. 4D). Secretion of SERPINA1 from HLCs after TTR knockdown was also assessed. No significant change in SERPINA1 levels was observed after TTR knockdown for H-HLCs, whereas SERPINA1 protein levels were elevated in three of four ATTR-HLCs (Fig. 4E), suggesting that – in ATTR-HLCs – SERPINA1 represents a chaperone that is inversely correlated to expression of TTR.

High levels of SERPINA1 and TTR are found in human serum. However, a physical interaction between SERPINA1 and TTR has not been reported. We assessed interaction of SERPINA1 and TTR by co-immunoprecipitation using cell culture supernatants derived from HLCs. First, expression of SERPINA1 was determined by western blot analysis of HLC cell culture supernatants. A single protein band of ∼55 kDa was observed in the supernatants of ATTR-HLCs and H-HLCs, which is similar to that in human plasma samples (Fig. 5A). Next, HLC cell culture supernatants were incubated with anti-TTR antibody for pull-down using immunobeads, followed by western blot analysis to detect co-precipitated SERPINA1. A SERPINA1-specific band was observed for cell culture supernatants derived from ATTR-HLCs and H-HLCs (Fig. 5B). Of note, SERPINA1 could not be precipitated from HLC supernatants when non-specific antibody was used for pull-down. Densitometric quantification of the SERPINA1 co-immunoprecipitates relative to the loading control revealed that 37.4±8.6% of SERPINA1 was pulled-down, suggesting that significant amounts of SERPINA1 physically interact with TTR. However, no differences in the amount of precipitated SERPINA1 was observed when cell culture supernatants of ATTR-HLCs and H-HLCs were used, suggesting that both wild-type and mutant TTR are bound to SERPINA1. To assess the interaction of SERPINA1 and TTR using an alternative method, a hybrid ELISA was employed (Fig. 5C). Positive values of absorbance relative to controls were observed when cell culture supernatants of ATTR-HLCs were used (OD₄₅₀ ≈1.6). Moreover, analysis of human plasma also resulted in above-background TTR detection values. Omission of anti-SERPINA1 antibody or the use of mouse sera containing human TTR and murine SERPINA1 did not produce above-background values (data not shown), indicating the specificity of the assay.

We further explored the nature of interaction between SERPINA1 and TTR by addressing TTR aggregation – an intrinsic event of ATTR. To establish an in vitro TTR aggregation assay, a previously described method employing the acidic denaturation of wild-type TTR (Klabunde et al., 2000) was used with modifications. After acidic denaturation of human TTR, the soluble and insoluble fractions, i.e. supernatant and pellet, were assessed by centrifugation followed by western blotting of TTR. After 72 h at pH 4.6, high molecular forms (HMFs) of TTR were significantly increased in the insoluble fraction (Fig. 6A,B). In contrast, such forms were not observed without incubation at 37°C (0 h) or at neutral pH, suggesting that treatment at low pH for several hours is a prerequisite for the formation of TTR HMFs. To further assess the specificity of the assay, we asked whether Tafamidis, a known stabilizer of TTR that inhibits TTR fibril formation (Chen et al., 2016), interferes with TTR aggregation. Pre-incubation with Tafamidis significantly reduced formation of TTR HMFs in a dose-dependent manner (Fig. S5), suggesting that the in vitro assay is suitable to detect interference of TTR aggregation. While pre-incubation with recombinant SERPINA1 did not fully inhibit the formation of HMFs present in the insoluble fraction, almost the same amount of HMFs was repeatedly observed in the soluble fraction (Fig. 6C,D). In contrast, recombinant albumin used as control did not interfere with formation of TTR HMFs, suggesting that SERPINA1 specifically hampers the conversion of soluble TTR into insoluble HMFs. Densitometric quantification revealed that addition of SERPINA1 significantly increased the relative amount of HMFs by ∼2-fold in the soluble fraction as compared to untreated control (0.86±0.06 and 0.37±0.04, respectively) (Fig. 6C,D).

In this study, we analyzed the gene expression of genes that have been previously associated with chaperone function, to assess the role of the PQC system for aberrant TTR expression in the hepatocyte. An iPSC-based cell model was used to determine the expression levels in hepatic cells. Of note, in contrast to liver-derived patient cells, the cells studied here were not subjected to secondary regulation by other organs or factors present in the blood. In our cohort of chaperone genes chosen for qRT-PCR analysis of HLCs, we observed that expression of several genes was significantly different in ATTR-HLCs as compared to H-HLCs. Notably, chaperones predominantly associated to extracellular location were affected in ATTR-HLCs, whereas genes primarily involved in ERAD and UPR pathways were mostly expressed at levels found in cells that had been derived from healthy individuals. Both last-mentioned systems have been described to fail detection of TTR mutants – i.e. ATTR-L55P and ATTR-V122I – that show high thermodynamic stability resulting in high secretion efficiency (Chen et al., 2014; Sekijima et al., 2005). Accordingly, ATTRV30M, which we investigated here in two ATTR-HLC lines, had been classified to have similar thermodynamic stability as wild-type TTR (Sato et al., 2007; Sekijima et al., 2005). In addition, ATTRG47A and ATTRR34T, the other two amyloidogenic mutations studied, share central clinical parameters with ATTRV30M (González-Duarte et al., 2013; Patrosso et al., 1998). Our observation, therefore,indicates that transcription mediated through ERAD and UPR pathways is not significantly affected in the presence of TTR mutants with a high thermodynamic stability.

Microarray analyses of liver biopsy samples, the current gold standard to assess differently regulated hepatic gene expression in ATTR, indicated a modulation of ER-associated proteins in patients carrying ATTRV30M (Norgren et al., 2014), e.g. gene expression of FK506 binding protein 2 (FKBP2) was significantly downregulated. In our study, a disparate regulation of this gene was not observed. However, apart from the different processing events of cells and RNA that were used used to determine gene expression in the two approaches, other factors, such as age of patients, influence of blood-borne growth factors and course of disease may affect levels of gene expression. HLCs derived from iPSC have been previously used for gene expression studies, also involving a few chaperone genes (Leung et al., 2013). The gene encoding heat shock protein family A (Hsp70) member 1A (HSPA1A) was one chaperone that was found to be slightly overexpressed (∼2-fold). We found HSPA1A to be differently expressed in two ATTR-HLCs (FC<−2.7). We might have missed the biological significance of this and other minor mRNA dysregulations of chaperone genes as we have chosen a high internal threshold (FC<±5) to indicate significance.

In our study, SERPINA1, FGA, A2M, FGB and HSPB1 were found to be highly expressed in ATTR-HLCs as compared to H-HLCs. Of these, HSPB1 is the only chaperone that is predominantly found intracellularly. Gene expression was found to be downregulated in two ATTR-HLCs, while two other patient-specific cell lines that displayed minor downregulation did not exceed a threshold of FC±5 (Fig. 3). HSPB1 is known to induce sequestration of beta-amyloid and is induced in the brains of mice transgenic for the human amyloid precursor protein (Ojha et al., 2011). Thus, HSPB1 may be overrepresented in neuronal cells exposed to amyloid fibrils (Santos et al., 2008) but may have similar functions in hepatocytes, although TTR amyloid deposition is not observed in the liver.

Of the chaperones identified in this study, fibrinogen is the only chaperone that has previously been reported to directly interact with TTR (da Costa et al., 2011; Tang et al., 2009b). The chaperone activity of FGA is based on the isoform fibrinogen-420 (Tang et al., 2009a), and glycation of FGA results in a loss of function (Fonseca et al., 2016). Moreover, FGA can cause amyloid deposition as a consequence of genetic mutation (Yazaki et al., 2018). Also, cross-seeding by FGA has to be considered as one possible mechanism by which amyloidic proteins spread, mostly worsening disease (Sakai et al., 2017; Verma et al., 2018). However, aggregation-prone proteins have also been shown to inhibit crosslinking of amyloidogenic proteins. Interestingly, this role has been described for monomeric TTR, which has been shown to interfere with beta-amyloid oligomerization (Garai et al., 2018; Li et al., 2013). It must, however, be taken into account that beneficial and disease-promoting elements can be closely related.

A2M has been identified to inhibit amyloid fibril formation of aggregation-prone proteins in other diseases, such as Alzheimer's disease (Yerbury et al., 2009). Of note, FGA, FGB and A2M encode for acute-phase proteins, i.e. proteins whose plasma concentration increase or decrease in response to inflammation (Jain et al., 2011), suggesting that upregulation of these genes is a characteristic of ATTR (Suenaga et al., 2017). At least two patient-specific ATTR-HLC cell lines exhibited upregulation of FGA, FGB and A2M. In line, previous studies also report an induction of the three chaperones in plasma samples of ATTRV30M patients (da Costa et al., 2015), corroborating the validity of our HLC model. Overexpression of these chaperones in our cell model, therefore, indicates that such regulation is due to the intrinsic synthesis of mutant TTR in the hepatocyte, rather than to a secondary regulative response. It can be assumed that FGA, FGB and A2M work synergistically, enabling hepatocytes to cope with amyloidogenic TTR. Although expression of FGA and FGB is correlated to result in functional fibrinogen, we can only speculate on the role of A2M to contribute to a complex cluster of chaperones targeting protein aggregates (Wyatt et al., 2012; Yerbury et al., 2009; Zsila, 2010). Although in our approach several chaperones were identified to be associated with ATTR pathogenesis, a rather small and heterogeneous patient group, encompassing three different mutations and early/late onset of disease, was studied. While our data suggest that chaperone expression differs in hepatic cells of ATTR patients and of healthy individuals, a genotype- and phenotype-specific analysis needs a higher number of patients.

SERPINA1 was identified as a chaperone in amyloidoses, and is part of amyloid aggregates in Alzheimer’s disease and ALS (Howlett and Moore, 2006; Zsila, 2010). In ATTR, SERPINA1 was found to be upregulated in plasma of ATTR patients (da Costa et al., 2015). Our findings indicate upregulation of SERPINA1 mRNA in one ATTR-specific hepatic cell line and downregulation in two other cell lines, while one cell line did not show disparate expression levels as compared to healthy individuals. To our knowledge decreased levels of SERPINA1 have not been reported for ATTR; however, Alzheimer’s patients show low SERPINA1 liquor levels (Hansson et al., 2009). As our ATTR patient cohort was low, the course of SERPINA1 regulation in hepatocytes cannot be proposed. However, SERPINA1 represents one of the most-regulated genes in our cohort of hepatic cell0073, suggesting that it is highly responsive to aberrant TTR expression in hepatocytes.

The important role of SERPINA1 in ATTR is emphasized again in our study by analysis of TTR knockdown. SERPINA1 expression was inversely correlated with TTR levels after knockdown, suggesting a mutual dependence of both genes. Intriguingly, investigations of serum TTR and SERPINA1 levels in cells of patients suffering various diseases indicates opposed levels of expression (Bag et al., 2014; Fatima et al., 2017; Raju M. et al., 2016). Although the exact molecular regulation of opposite expression levels remains to be unraveled, IL-6 driven induction of gene expression following inflammation might play a role. It is, therefore, tempting to speculate that the process of inflammation in ATTR modulates disease (Gonçalves et al., 2017). Furthermore, our data indicate for the first time that TTR directly interacts with SERPINA1. Although we did not detect any difference in the binding of SERPINA1 to wild-type or mutant TTR in our immunoprecipitation assay, subtle differences might exist in vivo, depending on the TTR mutant expressed by ATTR patients.

A protocol involving acid-mediated TTR aggregation was used in our study and resulted in high-molecular forms (HMFs) of TTR as observed after western blot analysis. Since the detection of HMFs depended on glutaraldehyde treatment prior to SDS gel electrophoresis, we assume that HMFs are highly labile. Staining of HMFs with Congo Red failed, possibly due to interference with glutaraldehyde (data not shown). Our data indicate that SERPINA1 interferes with TTR aggregation, suggesting that the chaperone can hamper a conversion of the soluble form into insoluble aggregates. Aggregation into self-assembled fibrils of SERPINA1 has been reported together with concurrent inhibition of these oligomers by SERPINA1 fragments (Janciauskiene et al., 1995). Several proteins, including serum amyloid P-component (SAP), apolipoprotein, complement binding factors and clusterin, have been found in amyloid deposits of patients (Azevedo et al., 2011; Klein et al., 2011), suggesting that interaction of TTR with serum proteins is common. Although binding of serum proteins to TTR may take place at different steps of amyloid formation and has diverse functional consequences, it has previously been proposed that clusterin preferentially binds to TTR monomers, resulting in the blocking of the first molecular events (Greene et al., 2015). However, as shown in our study, clusterin was not differently regulated in HLCs of ATTR patients. In the presence of the TTR tetramer stabilizer Tafamidis, we observed a specific pattern of HMFs that differed from the HMF pattern after addition of SERPINA1. Thus, we hypothesize that SERPINA1 binds to both wild-type TTR and mutant TTR, and reduces fibril formation during a phase other than that of TTR tetramer stabilization. By contrast, low levels of SERPINA1 might be unfavorable, as suggested by studies on ATTR patients with SERPINA1 deficiency. An early onset of disease after liver transplantation was reported (Guttmann et al., 2017).

The PQC system might represent an alternative therapy approach to target amyloidosis. Promising studies, e.g. for Alzheimer's disease, focus on proteins of the BRICHOS family (Cohen et al., 2015; Willander et al., 2012,b). It has previously been suggested that domains proSP-C and Bri2 can delay fibrillation of Aß40 and Aß42 (Willander et al., 2012,b). Proteins of the heat shock family have been demonstrated to efficiently regulate fibril length of α-synuclein, e.g. in Parkinson's disease (Gao et al., 2015; Nillegoda et al., 2015). For ATTR, it has been shown in cell culture that induction of ATF6 results in reduced secretion of aggregation-prone TTR, possibly impeding pathogenic fibril deposition (Chen et al., 2014; Shoulders et al., 2013). Although our work here adds several chaperones – most importantly SERPINA1 – to the growing search for other therapy strategies to treat ATTR, we were unable to decipher the exact biological mechanisms the identified chaperones use for pathogenesis.

In order to determine chaperone gene expression, we used iPSC-derived cell lines. As noted earlier, iPSC-based approaches can have shortcomings due to incomplete reprogramming and differentiation, most prominently when comparing different donors (Kajiwara et al., 2012; Ortmann and Vallier, 2017). Although we addressed variability by using identical protocols for iPSC generation and hepatic differentiation and an overall low variability of marker gene expression – including TTR and SERPINA1 – was observed, a bias due to the heterogeneity of cell population cannot be excluded. Within the limitations of our approach, our data suggest that SERPINA1 has a role in ATTR pathogenesis, possibly by inhibiting the formation of large TTR aggregates. Our data also indicate that a concerted modulation of – mostly extracellular – chaperone genes is induced in ATTR patients; this might be part of a complex proteostasis network, resulting in an adaptive response to aberrant TTR expression (van Oosten-Hawle and Morimoto, 2014).

All samples were obtained in accordance with ethics committee of the Universitätsklinikum Münster. Participants were informed regarding the use of data and informed consent was obtained in writing from all individuals enclosed in the study.

12-well cell culture plates (Greiner Bio-One, Kremsmünster, Austria) were coated with 0.1% gelatin (Merck Millipore, Darmstadt, Germany). Briefly, urine was centrifuged at 400 g for 10 min and cell pellets were washed with 50 ml of cold PBS (Sigma-Aldrich, St. Louis, MO), 1% penicillin–streptomycin (Pen–Strep; Thermo Fisher Scientific, Waltham, MA). After a second centrifugation, cell pellets were resuspended in UC medium [DMEM/F-12 (Life Technologies, Carlsbad, CA) supplemented with 10% FCS (Thermo Fisher Scientific), 1% Pen–Strep, 1×MEM non-essential amino acid solution (NEAA, Sigma-Aldrich), 1× GlutaMAX-I (Life Technologies), 0.1 mM 2-mercaptoethanol (Life Technologies) and SingleQuot Kit CC-4127 REGM (Lonza, Basel, Switzerland)]. Cells were resuspended in 1 ml of UC medium and incubated at 37°C with 5% CO₂ for 24 h. Half of the medium was changed daily. At confluence, cells were expanded for two passages (p2). Typically – and independently from the initial urine volume – 4×10⁶ cells were obtained after 3 weeks of cell culture.

Approximately 0.5×10⁶ of urine-derived cells at p2 were transfected with 1 µg of pCXLE-hOCT3/4-shp53-F, pCXLE-hSK and pCXLE-hUL (Addgene plasmids #27077, #27078 and #27080, respectively) by using the Amaxa Basic Nucleofector Kit for Primary Mammalian Epithelial Cells (VPI-1005, Lonza, Basel, Switzerland). Subsequently, cells were transferred to Matrigel^® matrix (Corning, Corning, NY) coated 6-well cell culture plates (Greiner Bio-One) and cultured for 24 h in UC medium. At day 1, UC medium was replaced with serum-free complete cell culture medium mTeSR-1 (Stemcell Technologies, Vancouver, BC, Canada) supplemented with 1% Pen–Strep. iPSC colonies were subcultured every 7 days by using 1 U/ml Dispase (Stemcell Technologies).

iPSC lines showing stable growth characteristics were transferred into single-cell suspensions using Accutase solution (Sigma-Aldrich). Typically, one 6-well cell culture plate with 4–5×10⁶ cells was used for hepatic differentiation. After 24 h, culture medium was replaced for 3 days with DMEM/F12, 100 ng/ml recombinant activin-A (PeproTech Germany, Hamburg, Germany), 100 ng/ml fibroblast growth factor-2 (FGF2, PeproTech Germany) and 50 ng/ml recombinant human Wnt3a (R&D Systems, Minneapolis, MN) using increasing concentrations of KnockOut SR Xenofree CTS (KSR, Life Technologies) of 0.2–2.0%. The next 8 days, cells were cultured in DMEM/F12 supplemented with 10% KSR, 1 mM NEAA, 1 mM L-glutamine, 1% dimethyl sulfoxide (Sigma-Aldrich) and 100 ng/ml hepatocyte growth factor (HGF, Peprotech). To induce hepatic maturation, cells were treated for the final 3 days with DMEM/F12 supplemented with 10% KSR, 1 mmol/l NEAA, 1 mM L-glutamine and 0.1 µM dexamethasone (Sigma-Aldrich).

Following fixation with 4% paraformaldehyde (Electron Microscopy Sciences) and permeabilization with 0.5% Triton X-100 (Thermo Fisher Scientific), cells grown in a 12-well plate (1–2×10⁶ cells) were blocked with 3% bovine serum albumin solution (BSA, Sigma-Aldrich) and stored at 4°C until staining. Primary and secondary antibodies were diluted in 1% BSA solution. Anti-albumin antibody (ab2406, Abcam, Cambridge, UK), anti-TTR antibody (A0002, Dako, Glostrup, Denmark), anti-HNF4a antibody (sc-6556, Santa Cruz Biotechnology, Dallas, TX) and anti-NANOG antibody (sc-33759, Santa Cruz Biotechnology) were used at 1:100. Anti-Oct4 antibody (60093PE, Stemcell Technologies), anti-SSEA-4 antibody (60062PE, Stemcell Technologies) and anti-TRA-1-60 antibody (60064AD, Stemcell Technologies) were used at 1:200. Antibodies against Alexa Fluor 488 (A11001, Invitrogen, Carlsbad, CA), Alexa Fluor 568 (A11057, Invitrogen) and Alexa Fluor 594 (A11012, Invitrogen) were used at 1:500. Samples were incubated with primary antibody for 2 h, and with secondary antibody for 1 h at 4°C. Photographs were taken using an Olympus CKX41-X10 microscope with cellSens Standard 1.11 imaging software.

Cell pellets derived from one 6-well cell culture plate were used to isolate whole RNA using the RNeasy Mini Kit (Qiagen, Hilden, Germany) and 1 µg of RNA was applied for first-strand synthesis using the SuperScript™ III First-Strand Synthesis SuperMix for qRT-PCR kit (Invitrogen). Of the reverse transcriptase product 1 µl was mixed with Takyon ROX SYBR Master Mix Blue dTTP (Eurogentec, Lüttich, Belgium) and 150 nM of primers (Table S2). qPCR was analyzed on the ABI Prism 7900 HT Sequence Detection System (Applied Biosystems, Foster City, CA). Data were analyzed by SDS 2.4 software (Applied Biosystems). Ct values were normalized to GAPDH and evaluated using the comparative Ct method (2^−ΔΔCt).

Embryoid bodies were established as previously described (Niemietz et al., 2016). 1×10⁶ cells were used to generate embryoid bodies. Briefly, iPSC clusters were cultured for 7 days in 6-well suspension culture plates (Greiner Bio-One) according to the manufacturer's protocol (TaqMan^® hPSC Scorecard™). The trilineage differentiation potential was examined using the 384-well hPSC Scorecard™ Panel (Life Technologies) followed by evaluation using the Thermo Fisher Cloud hPSC Scorecard Analysis software.

Glycogen synthesis of HLCs was assessed at day 14 of hepatic differentiation by periodic acid-Schiff (PAS) reaction using the system protocol (395B-1KT, Sigma-Aldrich).

HLCs at day 14 of hepatic differentiation were transferred to Opti-MEM (Gibco). siTTR1 (Alnylam Pharmaceuticals, Cambridge, MA) was added at 0.3 µM to the medium. TTR-ASO (IONIS Pharmaceuticals, Carlsbad, CA) was mixed at 1.5 µM with 2.5 µl Lipofectamine 2000 Reagent (Life Technologies). Scrambled oligonucleotides were used as control. Cell culture supernatants were harvested daily for 3 days.

Lysates derived from a 6-well plate were prepared using RIPA buffer (60 mM tris-HCl, 150 mM NaCl, 2% Na-deoxycholate, 2% Triton X-10, 0.2% SDS and 15 mM EDTA) in the presence of protease inhibitors (Roche, Complete Mini, EDTA-free) and separated on 10–12.5% SDS polyacrylamide gels. For blotting Amersham Protran Premium 0.45 nitrocellulose (GE Healthcare Life Sciences, Buckinghamshire, UK) was used. After overnight blocking in 5% non-fat dry milk (AppliChem, Darmstadt, Germany) blots were incubated in Tris-buffered saline (TBS)/0.05% Tween 20 (Sigma-Aldrich). Anti-TTR antibody (A0002, Dako) and anti-SERPINA1 antibody (843285, R&D Systems) were used at 1:1000 in 0.05% TBS-T with 0.5% non-fat milk for 1 h. Anti-goat HRP (ab97110, Abcam) and anti-rabbit HRP (ab6721, Abcam) were used at 1:10,000. Amersham ECL Western Blotting Detection Reagent (GE Healthcare Life Sciences) was applied. All blots were quantified by using ImageJ 1.50i analysis software.

ELISA of TTR was performed as reported elsewhere (Coelho et al., 2013). Briefly, Nunc 96-well MaxiSorp™ plates (44240421, Thermo Fisher Scientific) were coated overnight with anti-TTR antibody (A0002, Dako) in 50 mM carbonate/bicarbonate buffer (C3041-50CAP, Sigma-Aldrich) at 4°C. After washing, samples were diluted in 1× powerblock (HK085-5K, Biogenex, Fremont). Cell culture supernatants were used at 1:10 and human plasma samples were used at 1:19,600. For standard curve calculation, TTR (P1742-5MG, Sigma-Aldrich) was applied. Detection was by ab9015 (Abcam; at 1:2500) and alkaline-phosphate conjugation by A5187-1ML (Sigma-Aldrich). Colorimetric detection was performed by p-Nitrophenyl phosphate addition (N2770-50SET, Sigma-Aldrich). Absorbance was determined at 405 nm. Human SERPINA1 DuoSet ELISA (DY1268, R&D Systems) was used according to the manufacturer's instruction. Cell culture supernatants were used at 1:4000 and in 5% Tween 20 in PBS. TMB substrate (555214, BD Biosciences, Franklin Lakes, NJ) was used for colorimetric detection. Absorbance was determined at 450 nm. Hybrid ELISA was performed with goat anti-human SERPINA1 antibody (843285, R&D Systems) for capture and with sheep anti-human TTR antibody (ab9015, Abcam) for detection.

Human TTR (P1742, Sigma-Aldrich) was reconstituted in 1× powerblock (BioGenex, Fremont, CA) and incubated under acidic conditions to induce TTR aggregation. 1 µg of TTR was mixed with 1× volume of 200 mM acetate buffer containing 100 mM KCl and 1 mM EDTA (pH 4.6). Samples were incubated at 37°C for 72 h prior centrifugation at 20,000 g, 4°C for 30 min. The supernatant was carefully removed from the pellet and both fractions were treated with 2.5% glutaraldehyde solution (340855, Sigma-Aldrich). Samples were subjected to TTR western blotting thereafter.

Cell culture supernatant (15 µl) derived from 6-well plates was precleared with A/G Plus-Agarose beads (sc2003, Santa Cruz Biotechnology). Beads were mixed with 2 µg of anti-TTR antibody (A0002, Dako) and incubated with precleared samples in non-denaturing RIPA buffer (20 mM Tris HCl pH 8, 137 mM NaCl, 1% Triton X-100, 2 mM EDTA) and protease inhibitors while shaking at 4°C for 24 h. The next day, samples were washed 5× with PBS, centrifuged at 1000 g for 30 s and subjected to 12.5% PAGE.

Statistical analyses were performed by using Kruskal–Wallis one-way ANOVA and Student's t-test using SPSS 25 software. Data are given as mean±standard error (±s.e.m.). Correlation of two genes was determined by using Pearson's equation followed by Fisher's z-transformation.

We are grateful to O. Nadzemova for excellent technical assistance.