Introduction

Transthyretin (TTR)-associated amyloidosis (ATTR) is an autosomal dominant disorder caused by a point mutation in the TTR gene. According to the online registry for hereditary amyloidosis mutations (http://www.amyloidosismutations.com), 138 TTR mutations are associated to human amyloidosis. TTR is a 127-amino acid, 55-kDa protein composed of four identical, non-covalently associated subunits [1, 2]. TTR serves as a transport molecule for thyroxine (T4) and retinol-binding protein 4 (RBP4). The process of TTR amyloidogenesis involves rate-limiting dissociation of the TTR tetramer, followed by partial unfolding of monomers to yield non-fibrillar aggregates, protofibrils, and mature amyloid fibrils [3,4,5]. Structural studies of the TTR tetramer yielded several compounds that bind to the T4 hormone pocket of TTR, consequently stabilizing TTR and inhibiting fibril formation [6,7,8,9,10,11].

To gain insight into the pathogenesis of ATTR, several groups have generated transgenic mice that carry TTR genes with various mutations, such as Met30, Ser10/Met30, Pro55, or Ser84 [12,13,14,15,16,17,18]. Using the mouse metallothionein promoter or the human homologous TTR promoter, we produced transgenic mice in which amyloid deposition was observed in similar tissues as in human autopsy cases, except for its absence in the peripheral and autonomic nervous systems [15, 18]. Teng et al. [16] have also reported amyloid deposition in transgenic mice carrying the wild-type human TTR gene. By contrast, transgenic mouse strains with amyloidogenic Pro55 develop non-fibrillar TTR deposition, but fail to develop amyloid deposits in a 129S1/Sv background [12]. Later, Reixach et al. [19] demonstrated that heterotetramers comprising mouse and human subunits are kinetically more stable than those of human homotetramers and are considered to be inhibitory to dissociation and subsequent amyloid formation. Sousa et al. [12] reported the presence of amyloid fibrils in Pro55 transgenic mice on a TTR-null background.

Although liver transplantation is a promising approach for preventing disease progression, its utility of liver transplantation is limited by restricted donor availability, patient eligibility, and the risk of adverse events [20,21,22,23,24,25,26]. Several compounds that bind to the T4 hormone pocket were shown to stabilize the TTR tetramer thereby inhibiting fibril formation. Two TTR stabilizers, such as Tafamidis and Diflunisal, are now in clinical trials, and Tafamidis was approved in the European Union in 2011. Although both drugs seem to slow disease progression, no therapy can reverse the disease course [27,28,29,30]. Thus, there is an urgent medical need for new treatments that can halt or reverse the progression of amyloidosis and that are effective across all genotypes and disease stages.

Animal models that reproduce the pathology of human TTR-mediated familial amyloidosis are required to assess the in vivo efficacy of drug candidates. We previously produced humanized mice (Ttr hTTRVal30 and Ttr hTTRMet30) carrying either human normal (hV) or mutant (hM) TTR cDNA at the mouse Ttr locus using a Cre-mediated recombination system [31]. To test the stabilizing activity of drugs in vivo, we think it is essential to mimic the physiologic condition of human serum, where human TTR binds to human RBP4. Thus, we also created a humanized mouse (Rbp4 hRBP4) carrying the human RBP4 (hR) cDNA at the mouse Rbp4 locus using the same strategy [32,33,34]. We then crossed these lines to generate mice humanized mice with a normal (Ttr hTTRV30/hTTRV30 :Rbp4 hRBP4/hRB4, abbreviated as hV/hV:hR/hR) or patient genotype (Ttr hTTRV30/hTTRM30 :Rbp4 hRBP4/hRB4, abbreviated as hV/hM:hR/hR). Here we examined amyloid deposition in these double-humanized mice compared with conventional transgenic mouse strains on a wild-type (Ttr +/+:Tg[6.0hTTRMet30], abbreviated as +/+:Tg) or knockout Ttr background (Ttr−/− :Tg[6.0hTTRMet30], abbreviated as −/−:Tg).

Materials and methods

Conventional Transgenic Mouse Strains

Wild-type (+/+:Tg) or knockout Ttr background (−/−:Tg) hemizygous mice for the 6.0hTTRMet30 gene were used as control. These mice contained ~30 copies of the 6.0hTTRMet30 gene [35].

Construction of the Targeting/Replacement Vectors and the Isolation of Targeted/Replaced ES Cells

The construction of the targeting vector and the isolation of targeted ES cells for the mouse Ttr locus were achieved using a previously described method [31]. Briefly, a 3-kb 5′ homologous region upstream of the ATG codon and a 6.7-kb 3′ homologous region downstream of the ATG codon were ligated into a p71neoP cassette containing lox71-PGK-neo-loxP-polyA, to produce the TTR71neoPTTR construct that comprised 5′ Ttr homologous region-lox71-PGK-neo-loxP-polyA-3′ Ttr homologous region. Finally, a diphtheria toxin A (DT-A) fragment with an MC1 promoter was ligated to the 5′-end of TTR71neoPTTR to produce the targeting vector (Fig. 1a). The targeting vectors were introduced into TT2 ES cells [36] derived from an F1 embryo obtained from a mating between C57BL/6 and CBA mice (Charles River, Inc., Yokohama, Japan). To generate the replacement vector, we first made a cassette that contained a rabbit β-globin intron II, a hV cDNA or a hM cDNA, and a puromycin resistance gene with the phosphoglycerate kinase promoter (PGKpuro) flanked by two Flp recognition target (FRT) sequences. Then, the cassette flanked by lox66 and loxP sites was inserted into pSP73 (Promega, Tokyo, Japan) (Fig. 1a). The ES cell lines with the Ttr-targeted null allele were coelectroporated (Bio-Rad Gene Pulser at 400 V and 125 μF) with 20 μg of replacement vector plasmid and 20 μg of pCAGGS-Cre plasmid [37] to produce ES clones with the replaced allele.

Fig. 1
figure 1

Creation of targeted and replaced alleles. a Ttr locus. Homologous recombination between the wild-type allele and the targeting vector resulted in the creation of a targeted null allele carrying the PGK-neo gene flanked by lox71 and loxP. In the second step, the targeted clones were electroporated with the replacement vector containing hV or hM cDNA flanked by lox66 and loxP. Site-directed recombination occurred between lox71/lox66 and loxP/loxP, resulting in the creation of the replaced allele, Ttr hV or Ttr hM. b Rbp4 locus. Homologous recombination between the wild-type allele and the targeting vector yielded a targeted null allele carrying the PGK-neo gene flanked by lox71 and loxP. In the second step, targeted clones were electroporated with the replacement vector containing hRBP4 ORF flanked by loxKMR3 and loxP. Site-directed recombination between lox71/loxKMR3 and loxP/loxP resulted in the creation of the replaced allele, Rbp4 hR

Full size image

The construction of the targeting vector and the isolation of targeted ES cells for the mouse Rbp4 locus were achieved using a previously described method (Fig. 1b) [32, 34]. Briefly, a 2.8-kb 5′ homologous region upstream of the ATG codon and a 6.7-kb 3′ homologous region downstream of the ATG codon were ligated into a p71neoP cassette containing lox71-PGK-neo-loxP-polyA, to produce the RBP471neoPRBP4 construct (5′ Rbp4 homologous region-lox71-PGK-neo-loxP-polyA-3′ Rbp4 homologous region). Finally, a diphtheria toxin A (DT-A) fragment with an MC1 promoter was ligated to the 3′-end of 3′ Rbp4 homologous region to produce the targeting vector. The targeting vectors were introduced into TT2 ES cells [36]. To construct the replacement vector, we first made a cassette that contained a human RBP4 cDNA and a puromycin resistance gene with the PGK promoter (PGKpuro) flanked by two FRT sequences. Then, the cassette flanked by loxKMR3 and loxP sites was inserted into pSP73 (Promega, Tokyo, Japan). The ES cell lines with the Rbp4-targeted null allele were coelectroporated (Bio-Rad Gene Pulser at 400 V, 125 μF) with 20 μg of replacement vector plasmid and 20 μg of pCAGGS-Cre plasmid [37] to produce ES clones with the replaced allele.

PCR and Southern Blot Analyses of Isolated Targeted ES Clones

The Ttr targeted ES clones were screened and confirmed by long PCR and Southern blot analyses using genomic DNA purified by phenol, chloroform and ethanol precipitation from ES cells. In the long PCR analysis, the primers, 2-20-1S (5′- GTAAGCAATCTTAGCCAGGCTCTCC-3′) and lox71-PR (5′-TATACGAACGGTATAGGTCCCTCGAC-3′), were used to detect the 5′ end of the targeted allele (3.0 kb). The primers mTtr6 (5′-TGGGCTGAGTCTCTCAATTCTG-3′) and neo-F (5′-AGAGGCTATTCGGCTATGAC-3′) were used to detect the 3′-end of the targeted allele (9.0 kb). For Southern blot analysis, DNAs from ES cells was digested with BamHI or XbaI and were analyzed for the presence of the targeted allele using a neo probe. Digested DNAs was electrophoresed on a 1.2% agarose gel and blotted onto nylon membranes (Hybond-N+; Amersham, Tokyo, Japan). After the membranes were cross-linked by exposure to ultraviolet light (UV Stratalinker 1800; Stratagene, La Jolla, CA, USA), hybridization was performed using a neomycin-specific probe prepared using a DIG DNA labeling and detection kit (Roche, Tokyo, Japan).

The Rbp4-targeted ES cell clones were confirmed by Southern blot analyses as described previously [34]. For detection of homologous recombination in the 5′ region, DNA from ES cells was digested with XhoI/SalI or MunI and the targeted allele was detected with a neo probe and a 5′ probe; for detection of homologous recombination in the 3′ region, AgeI/HpaI or SacII/HpaI digestion and a neo probe and 3′ probe were used, respectively.

PCR and Southern Blot Analyses for Genotyping of ES Clones or Mice

Genotypes of ES cells and mice at the Ttr locus were determined by PCR analyses and confirmed by Southern blot analysis using genomic DNA from ES cells or tails according to the previously described methods [31]. Briefly, in the PCR analysis, the primers mTtr A9-s (5′-CGTAGAGCGAGTGTTCCG-3′) and mTtr19 (5′-CAGCTGTTGCTATAGTAATTCCC-3′) were used to detect the wild-type mouse Ttr allele (864 bp); the primers neo-R (5′-CACCATGATATTCGGCAAGC-3′) and neo-F (5′-AGAGGCTATTCGGCTATGAC-3′) were used to detect the targeted allele (545 bp); and the primers A9-s and SP-A (5′-CAGTGTATATCATTGTAACC-3′) were used to detect the replaced allele (783 bp). For Southern blot analysis, 10 μg of DNA from mice with the targeted allele and replaced allele was digested with BamHI or XbaI and BamHI or BglII, and was analyzed using a neo-specific or puromycin-specific probe, respectively.

Genotypes of ES cells and mice at the Rbp4 locus were determined by PCR analyses and confirmed by Southern blot analysis using genomic DNA from ES cells or tails according to previously described methods [32, 34]. Briefly, in the PCR analysis, the primers neo-R (5′-CACCATGATATTCGGCAAGC-3′) and neo-F (5′-AGAGGCTATTCGGCTATGAC-3′) were used to detect the targeted allele (545 bp); the primers mRbp4 GSPF1 (5′-CTCGGCTCCGTCGCTCCACG-3′) and mRbp4 GSPR1 (5′-CCAGAGCCCAGAGAACTGAG -3′) were used to detect the wild-type mouse Rbp4 allele (403 bp) and the replaced allele (3421 bp). To detect homologous recombination in the 5′ region, DNA from ES cells was digested with XhoI/SalI or MunI and analyzed for the presence of the targeted allele using a neo and 5′ probes. To detect homologous recombination in the 3′ region, DNA from ES cells was digested with AgeI/HpaI or SacII/HpaI and analyzed for the presence of the targeted allele using neo and 3′ probes. To detect the replaced allele genomic DNA from ES cells was digested with BamHI or BglII and was used for Southern blot analysis with a puromycin-specific probe.

Generation of Mouse Ttr or Rbp4 Knockout Mice and Human TTR or RBP4 Knock-in Mice

Chimeric mice were produced by aggregation of ES cells with eight-cell embryos from ICR mice according to previously described methods [31]. Chimeric male mice were backcrossed to C57BL/6 females (Nippon Clea, Kanagawa, Japan) and mice from after the tenth generation were used in the following experiments. The experimental protocols that involved animals were approved by the Kumamoto University Ethics Committee for Animal Experiments (F25-329, F27-122), and all experiments were performed in accordance with the institute’s guidelines.

Serum hTTR, mRBP4 and hRBP4 levels

Commercial ELISA kits were used according to the manufacturers’ instructions to determine the serum concentrations of hTTR (KA0495, Abnova, Taipei, China), mRBP4 (Mouse Retinol-Binding protein 4 ELISA Kit SimpleStep, ab202404, Abcam, Tokyo, Japan) and hRBP4 (AG-45A-0035YEK-KI01, AdipoGen, San Diego, CA, USA). Elisa assay was performed when the male mice were at 3, 12, and 24 months of age.

Histochemical and Immunohistochemical Analysis

Male mice were used for histochemical and immunohistochemical analyses. Mice were killed by cervical dislocation at 12, 18, and 24 months of age. Tissues—brain, heart, kidneys, spleen, skeletal muscle, stomach, small and large intestines, and sciatic nerves—were excised, fixed in 10% neutral-buffered formalin, and embedded in paraffin. Paraffin sections were stained with hematoxylin and eosin. For histochemical demonstration of amyloid, the serial sections were stained with Congo Red according to Wright’s method. To detect the emerald green birefringence emitted from the amyloid deposits, a polarized microscope was used to examine the Congo Red stained sections.

For immunohistochemical demonstration of the major components of amyloid deposits, the serial sections were immunostained by the indirect immunoperoxidase method. The primary antibodies were rabbit anti-human TTR (diluted 1:500; Sigma-Aldrich, Tokyo, Japan) or goat anti-mouse serum amyloid A (SAA) (15 μg/ml, R&D Systems, Tokyo, Japan) polyclonal antibodies. The primary antibodies were detected with biotinylated anti-rabbit or anti-goat secondary antibody (diluted 1:200; Vectastain ABC Kit; Vector Laboratories, Burlingame, CA) and DAB detection kit (Ventana Medical Systems, Tucson, AZ) according to the manufacturer’s instructions. To obtain negative controls, the same procedure was done without the primary antibodies.

Gel Filtration Chromatography to Separate RBP4–TTR Complexes

Plasma (0.1 ml) diluted with same volume of buffer (100nmol/L Na2PO4-NaH2PO4, pH7.4) was loaded onto a Superdex 200 Increase 10/300 GL column (GE Healthcare, Tokyo, Japan) column connected to an LC-8020 HPLC system (Toso, Tokyo, Japan). Protein complexes were separated by passing phosphate-buffered saline through at 0.5 ml/min for 50 min at room temperature. Eluted proteins were detected by their absorbance at 280 nm and collected in 0.5-ml fractions for analysis by SDS–PAGE.

Statistical Analysis

Amyloid deposition was widely distributed in tissues. Thus, it is difficult to use ImageJ to quantify tissue deposition. Instead, we used simply pathological criteria to quantify tissue deposition. For example, no amyloid deposition was scored as 0; ±(deposition limited to the walls of small vessels) as 1; +(deposition in walls of small vessels and surrounding areas) as 2; ++ (moderate deposition in interstitium) as 3; and +++ (marked deposition in interstitium and parenchyma) as 4. Mann–Whitney U-test was used for the statistical analysis. P < 0.05 was considered to indicate a significant difference.

Results

Establishment of ES Clones and Mouse Strains with the Targeted Null Allele or Replaced Allele

The mouse lines carrying either hV or hM at the mouse Ttr locus have been described previously [31]. Mouse lines with hR at the mouse Rbp4 locus (humanized Rbp4 mice) were produced in a similar way as humanized TTR mice [32]. Briefly, the replaced allele was created in two steps. First, the targeted null allele was generated by disrupting the ATG exon, which is in exon 1 in the Ttr gene (Fig. 1a) and exon 2 in the Rbp4 gene (Fig. 1b), using a targeting vector containing a neomycin (neo) resistance gene flanked by lox71 and loxP sites. Second, the Cre-mediated site-specific introduction of hV, hM or hR cDNA was performed to produce the replaced allele (Fig. 1a, b).

These humanized mice were bred to produce two mouse strains: hV/hV:hR/hR and hV/hM:hR/hR. Mice carrying the humanized allele were bred to homozygosity [33]; heterozygous and homozygous mice were born at the expected from Mendelian ratio.

In addition, we used two other transgenic strains, +/+:Tg [38] and −/−:Tg that were established by mating +/+:Tg with Ttr knockout mice. These mice carry the hM gene harboring a 6.0-kb upstream region on the C57BL/6 background or the Ttr-null background, respectively.

Serum hTTR, mRBP4, and hRBP4 Levels

Four mice in each strain were used for the examination of serum hTTR, mRBP4, and hRBP4 levels. In +/+:Tg mice, mean (±SEM) hTTR serum levels at 3, 12, and 24 months of age were 149 ± 7.53, 148 ± 12.17, and 142 ± 16.19 μg/ml, respectively (Fig. 2a, left panel). In −/−:Tg mice, mean (±SEM) hTTR serum levels at 3, 12, and 24 months of age were 143 ± 5.12, 141 ± 9.39, and 131 ± 11.62 μg/ml, respectively (Fig. 2a, right panel). The corresponding values for hV/hV:hR/h mice were 6.62 ± 0.37, 6.40 ± 0.28, and 6.06 ± 0.49 μg/ml, respectively (Fig. 2b, left panel), while in hV/hM:hR/hR mice they were 5.77 ± 0.16, 5.94 ± 0.50, and 5.53 ± 0.64 μg/ml, respectively (Fig. 2b, right panel). In all mouse strains, there was no significant difference in mean hTTR serum levels among age groups. Although there was a trend of lower TTR levels in hV/hM:hR/hR mice than in hV/hV:hR/hR mice, the difference was not significant. Thus, the serum hTTR level in hV/hV:hR/hR mice or hV/hM:hR/hR mice was ~1/25 of that in +/+:Tg or −/−:Tg mice.

Fig. 2
figure 2

Serum levels of hTTR and hRBP4. a Serum hTTR levels in +/+:Tg and −/−:Tg mice. b Serum hTTR levels in hV/hV:hR/hR and hV/hM:hR/hR mice. c Serum mRBP4 levels in +/+:Tg and −/−:Tg mice. d Serum hRBP4 levels in hV/hV:hR/hR and hV/hM:hR/hR mice. There was no significant difference in mean hTTR, mRBP4, and hRBP4 serum levels among age groups

Full size image

In +/+:Tg mice, mean serum levels of mRBP4 at 3, 12, and 24 months of age were 25.2 ± 1.77, 25.1 ± 1.28, and 24.7 ± 1.06 μg/ml, respectively (Fig. 2c, left panel). The corresponding serum mRBP4 levels were 26.5 ± 0.69, 25.4 ± 0.89, and 24.9 ± 0.87 μg/ml, respectively, in −/−:Tg mice (Fig. 2c, right panel). In hV/hV:hR/hR mice, mean serum levels of hRBP4 at 3 months, 12 months and 24 months of age were 0.60 ± 0.05, 0.56 ± 0.04, and 0.55 ± 0.03 μg/ml, respectively (Fig. 2d, left panel). The corresponding mean serum hRBP4 in hV/hM:hR/hR mice were 0.46 ± 0.02, 0.47 ± 0.03, and 0.44 ± 0.03 μg/ml, respectively (Fig. 2d, right panel). In all mouse strains, there were no significant differences in serum mouse or human RBP4 levels among age groups. As we reported previously [32], the serum level of hRBP4 in hR/hR (Rbp4 hRBP4/hRBP4) mice was 7.8 ± 0.53 μg/ml. Thus, these data suggested that low serum hTTR levels resulted in low RBP4 levels, probably due to a low percentage of TTR-bound hRBP4 and glomerular filtration of non-TTR-bound RBP4 into urine. Again, there was a trend of lower RBP4 in hV/hM:hR/hR mice than in hV/hV:hR/hR mice, but this difference was not significant.

Physical Features

As the serum TTR and RBP4 levels were considerably low, we examined the mortality rate, body weight, food consumption, and clinical chemistry to verify the effects of gene modification. Among 15 mice in each group, 4, 3, 5, and 6 mice were found dead in the +/+:Tg, -/-:Tg, hV/hV:hR/hR, and hV/hM:hR/hR groups, respectively, before completion of the 24-month observation period. Thus, the number of dead animals was similar among the four strains. The cause of death was not investigated. The time course of the mean body weight gain is shown in Fig. 3. Although the mean body weight of hV/hV:hR/hR and hV/hM:hR/hR mice was slightly lower at some points, there was no difference at 24 months of age. Food consumption was measured for one week when mice were 6 months of age and was similar among the four mouse strains. In addition, the results of the clinical chemistry and hematology analyses of blood samples were similar among the four strains.

Fig. 3
figure 3

Time-courses of the mean body weight gain. Although the mean body weight of hV/hV:hR/hR and hV/hM:hR/hR mice was slightly lower at some points, there was no difference at 24 months of age among the four strains

Full size image

Onset and Degree of Amyloid Fibril Deposition

Amyloid deposition was not observed in the brain, liver, or spleen of all mouse strains. Amyloid fibrils were stained with the anti-TTR antibody, but not the anti-SAA antibody, suggesting that secondary amyloidosis did not occur under the conventional conditions of the mouse facility.

Amyloid deposition was observed only in the intestinal tract in +/+:Tg (Fig. 5a, e, i) and −/−:Tg mice (Fig. 5b, f, j). Amyloid deposition was first found in the alimentary tract at 12 months of age (Fig. 4, Table 1). The amount of amyloid deposits in the alimentary tract was increased with age (Fig. 4, Table 1). However, amyloid deposition in the heart and kidney was not found, even at 24 months of age, in these lines under the conventional conditions used in this study (Table 1). In general, there was less amyloid deposition in this experiment than that previously reported [15, 39].

Fig. 4
figure 4

Amounts of amyloid deposition. a At 12 months of age. Amyloid deposition in the alimentary tract was observed in +/+:Tg, −/−:Tg and hV/hV:hR/hR lines. b At 18 months of age. Amyloid deposition in the alimentary tract was observed in all lines. c At 24 months of age. The amounts of amyloid deposition in the alimentary tract was higher in the hV/hV:hR/hR and hV/hM:hR/hR lines than those in the +/+:Tg and −/−:Tg lines. Amyloid deposition in the heart was observed only in the hV/hV:hR/hR and hV/hM:hR/hR lines. Amyloid deposition in the sciatic nerve was observed only in the hV/hM:hR/hR line

Full size image
Table 1 Tissue distribution of TTR amyloid deposits
Full size table

Amyloid deposition was first found in the alimentary tract at 12 months of age in the hV/hV:hR/hR line but at 18 months of age in the hV/hM:hR/hR line (Table 1). Amyloid deposition in the hV/hV:hR/hR and the hV/hM:hR/hR lines at 24 months of age was shown in Fig. 5c, g, and k and Fig. 5d, h, and l, respectively. In these lines, the amount of amyloid deposits in the alimentary tract increased with age and was higher than that in the +/+:Tg and −/−:Tg lines (Fig. 4, Table 1). Amyloid deposition was first found in the stomach or small intestine and then in the large intestine. The major sites of amyloid deposition in the alimentary tract were similar to those previously described [15, 18]. In addition, amyloid deposition was observed in the heart in both hV/hV:hR/hR (Fig. 5m, n) and hV/hM:hR/hR (Fig. 5o, p) strains at 24 months of age, but there were more amyloid deposits in the hV/hM:hR/hR line than in the hV/hV:hR/hR line (Fig. 4). Surprisingly, amyloid deposition was found in the perineurium of the sciatic nerve of the hV/hM:hR/hR mice at 24 months of age (Fig. 5q–s; Table 1). We believe that this is the first case of amyloid deposition in the peripheral nerve in the model mouse, although non-fibrillar deposits were observed in Tg(6.0hTTRV30M):Hsf1 −/− (designated as TTR/HSF1) mice [40]. Amyloid deposition in the kidney was observed in only the wall of a small vessel in one hV/hV:hR/hR mouse. Taken together, there was more amyloid deposition in the hV/hV:hR/hR and hV/hM:hR/hR lines than in the +/+:Tg and −/−:Tg lines. These results were surprising because serum hTTR levels in the hV/hV:hR/hR and hV/hM:hR/hR lines were ~1/25 compared with those in the +/+:Tg and −/−:Tg lines. These results suggest that the kinetic and thermodynamic stability of the TTR:RBP4 complex is important for amyloid deposition that replicating the human disease.

Fig. 5
figure 5

Histochemical and immunohistochemical analyses. Histochemical and immunohistochemical findings at 24 months of age in the small intestine (a–l), the heart (m–p), and the sciatic nerve (q–s). The sections stained with Congo Red (a–d, m, o, and q) showed apple green birefringence under polarizing light (e–h, n, p, and r). Tissue sections were also stained with anti-human TTR (i–l, and s). Amyloid deposition was observed only in the intestinal tract in +/+:Tg (a, e, and i) and −/−:Tg mice (b, f, and j). By contrast, amyloid deposition was observed in the intestinal tract (c, g, and k) and heart (m and n) in hV/hV:hR/hR mice and in the intestinal tract (d, h, and l), heart, (o and p) and sciatic nerve (q–s) in hV/hM:hR/hR mice. Scale bar: 100 μm in a–p and 50 μm in q–s

Full size image

Assessment of TTR deposition by immunohistochemistry

We analyzed anti-human TTR-positive deposits by immunohistochemistry at 12, 18, and 24 months of age (Table 2). At 12 months of age, anti-human TTR-positive deposits was observed in the small intestine of all mice except hV/hM:hR/hR mice in which 30% of them showed TTR-positive deposits. In the heart and kidneys, 10–30 % of mice showed TTR-positive deposits, while no amyloid fibril deposits were observed. At 18 months of age, anti-human TTR-positive deposits was observed in the small intestine of all mice. In the heart and kidneys, 30–50 % of mice showed TTR-positive deposits, while no amyloid fibril deposits were observed. At 24 months of age, anti-human TTR-positive deposits was observed in the small intestine of all mice. In the heart and kidneys, 30–100 % of mice showed TTR-positive deposits. Thus, anti-human TTR-positive deposits were found in earlier age and higher percentage in all strains than amyloid fibril deposits.

Table 2 Anti-TTR antibody-positive and amyloid fibril tissue deposition
Full size table

Gel filtration chromatography

To analyze the kinetic and thermodynamic stability of the TTR:RBP4 complex, plasma samples from +/+, hV/hV:hR/hR and hV/hM:hR/hR mice were subjected to gel filtration chromatography, and column fractions were analyzed by western blotting. The majority of immunoreactive RBP4 and TTR eluted together in fractions 28–31 in the control mouse (Fig. 6). A small amount of immunoreactive RBP4 eluted after this primary peak in subsequent fractions 32–39 (Fig. 6). These secondary fractions represent non-TTR-bound RBP4 because no associated TTR immunoreactivity was observed (Fig. 6). The percentage of the TTR-RBP4 bound form in +/+ mice was 82.0 ± 2.2 (Fig. 6). However, most of the RBP4 in the hV/hV:hR/hR and hV/hM:hR/hR mice were not bound to TTR, indicating that the hTTR was also free. These results suggest that the increase in free hTTR accelerates amyloid deposition, even under conditions of low serum hTTR levels.

Fig. 6
figure 6

Gel filtration chromatography. Each fraction was analyzed by western blotting using anti-RBP4 and anti-TTR antibodies. The bound form of RBP4 and TTR was found in fractions 28–31, whereas the free form of RBP4 was found in fractions 32–39. The TTR–RBP4 complex was barely detected in the hV/hV:hR/hR and hV/hM:hR/hR lines

Full size image

Discussion

In this study, we analyzed amyloid deposition in two double-humanized mouse strains, hV/hV:hR/hR and hV/hM:hR/hR. The serum levels of hTTR and hRBP4 were much lower in the double-humanized mice than in conventional transgenic mice, Tg(6.0-hMet30), on the wild-type Ttr or the Ttr-null background. Nevertheless, amyloid deposition was more prominent in hV/hV:hR/hR and hV/hM:hR/hR mice than in conventional transgenic mice. In addition, amyloid deposition was also observed in hV/hV:hR/hR mice, carrying the normal human TTR gene. Amyloid deposition started later in hV/hM:hR/hR mice than in hV/hV:hR/hR mice, but hV/hM:hR/hR showed more amyloid deposition at 24 months of age. Furthermore, amyloid deposition was first observed in the sciatic nerve without any additional genetic change.

Previously, we reported no amyloid deposition in transgenic mice carrying the wild-type hTTR gene (designated as C57BL/6-Tg(7.2TTRVal30)) [14]. By contrast, non-fibrillar deposits were reported in mice transgenic for the wild-type hTTR gene (designated as B6D2F2-Tg(19.2TTRVal30)) by other investigators [16, 41]. Serum concentrations of hVal30 were 26–80 μg/ml in C57BL/6-Tg(7.2TTRVal30) mice and 1–3.5 mg/ml in B6D2F2-Tg(19.2TTRVal30) mice. As the serum levels of hVal30 in B6D2F2-Tg(19.2TTRVal30) mice were 10–100 times higher than those in C57BL/6-Tg(7.2TTRVal30) mice, the high Val30 expression was thought to be required for amyloid deposition of wild-type hTTR. However, the serum concentration of hTTR in our hV/hV:hR/hR mice was 6.62 ± 0.37 μg/ml, 10-fold lower than that in C57BL/6-Tg(7.2TTRVal30) mice and 100-fold lower than that in B6D2F2-Tg(19.2TTRVal30) mice. We also reported that the serum levels of human mutant TTR are correlated, to some extent, with the occurrence and degree of amyloid deposition in C57BL/6-Tg(7.2TTRVal30) mice [15]. However, amyloid deposition was more prominent in hV/hV:hR/hR and hV/hM:hR/hR mice than in +/+:Tg and −/−:Tg mice. Furthermore, we are the first to observe amyloid deposition in the sciatic nerve without additional genetic change, such as Hsf1 deficiency, in the hV/hM:hR/hR line.

As reported by other investigators [12, 13, 16], anti-TTR antibody-positive deposits were found in earlier age and higher percentage in all our mice than amyloid fibril deposits. Teng et al. demonstrated that both amyloid and nonamyloid deposits were intact human TTR monomers with no evidence of proteolysis by amino terminal amino acid sequence analysis and mass spectrometry [16]. Thus, the anti-TTR antibody-positive deposits may represent a preamyloid intermediate state.

In our models, the mouse Ttr and Rbp4 genes were replaced with the human TTR and RBP4 genes, respectively. Thus, there is no expression of mouse TTR. It is well known that heterotetramers of human and mouse TTR are present in the serum of these transgenic mice and that these heterotetramers are more stable than hTTR homotetramers [19]. In fact, amyloid deposition was accelerated by TTR-Leu55Pro on a Ttr-null background (designated as Ttr−/−:129S1/Sv-Tg(TTRPro55)] [12] or Ttr−/−:B6D2F2-Tg(TTRPro55) [13]). By contrast, Kohno et al. [35] demonstrated no differences in the onset, progression, or tissue distribution of amyloid deposition in Tg(6.0hTTRMet30) mice with or without the murine Ttr gene (designated as Ttr +/+:C57BL/6-Tg(6.0hTTRMet30) or Ttr −/−:Tg(C57BL/6–6.0hTTRMet30)). The serum levels of TTRPro55 in Ttr −/−:129S1/Sv-Tg(TTRPro55), Ttr −/−:B6D2F2-Tg(TTRPro55), and Ttr −/−:Tg(C57BL/6–6.0hTTRMet30) mice were 10–200, 20–35, and 240–340 μg/ml, respectively. These data suggest that amyloid deposition was affected by the presence or absence of mouse Ttr under conditions of low serum levels of human mutant TTR. Amyloid deposition may be facilitated by the absence of mouse TTR, probably because of the instability of human TTR tetramers. Interestingly, hTTRVal30 and hTTRMet30 tetramers are more unstable in mouse serum, when the serum human TTR levels were low [31]. Human TTR mRNA and TTR protein expression levels in the liver of homozygous hV/hV mice were approximately twice those in heterozygous + /hV mice. However, the serum human TTR levels were much lower in hV/hV mice than in + /hV mice. The low mean serum hTTR levels could be due to shunting to the ERAD pathway. In a previous study [31], we examined morphological abnormality by transmission electronmicroscopy and the localization of hTTR using immunofluorescent detection with anti-calnexin and anti-GM130 antibodies. We did not find any abnormality, suggesting that hTTR was secreted by normal pathway.

RBP4 may also influence the amyloid deposition. As we reported previously, the serum level of hRBP4 in Rbp4 hR/hR (hR/hR) mice was 7.4 ± 1.50 μg/ml [32], whereas the serum levels of hRBP4 levels in hV/hV:hR/hR and hV/hM:hR/hR mice were 0.60 ± 0.05 and 0.46 ± 0.02 μg/ml, respectively, indicating a considerable reduction in these mice with serum hTTR levels ~1/25 of normal. This decrease in RBP4 may be due to glomerular filtration of non-TTR-bound RBP4. As shown in Fig. 6, the hTTR-bound form of hRBP4 was barely detected, although ~80–90% of RBP4 is TTR bound under normal conditions [32, 42]. Thus, an increase in non-RBP4-bound TTR potentially accelerated amyloid deposition in our double-humanized mice. In fact, RBP4 can prevent TTR amyloid formation at acidic pH [43].

To the best of our knowledge, neither non-fibrillar nor fibrillary deposits have been observed in the peripheral and autonomic nervous systems of conventional TTR transgenic mouse models. Non-fibrillar deposits accompanied by the induction of pro-inflammatory cytokines, RAGE up-regulation, and NF-κB activation were first reported in the sciatic nerve, dorsal root ganglia (DRG), and autonomic ganglia of Tg(6.0hTTRV30M):Hsf1 −/− (designated as TTR/HSF1) mice [40]. This result suggested that disrupting the heat shock response would aggravate TTR deposition and that HSF1 is involved in famial amyloidotic polyneuropathy (FAP) pathogenesis as a defense mechanism against extracellular TTR deposits. In our hV/hM:hR/hR mice, we observed amyloid deposition in the perineurium of the sciatic nerve. However, no such inflammatory response was observed in our hV/hM:hR/hR mice. The inflammatory response might be important for neurodegeneration, but the physico-chemical properties of the TTR–RBP4 complex in the microenvironment surrounding the peripheral nerve or autonomic nerve may be important for non-fibrillar and fibrillary deposits.

We previously reported that amyloid deposition was observed in the heart and kidney of the +/+:Tg and −/−:Tg lines [15, 35]. However, this result was not reproduced in this experiment, perhaps because of different housing conditions, as amyloid deposition is greatly affected by environmental conditions [39]. The amount of amyloid deposits differs under various conventional conditions [44].

In human, serum levels of TTR decreased with aging [45,46,47]. Kohno et al. [35] examined the serum hTTR levels in +/+:Tg and −/−:Tg strains at 11, 14, and 18 months of age, but there was no significant difference among age groups. We also examined the serum levels of hTTR, mRBP4, and hRBP4 at 12 and 24 months of age. However, there was no significant difference among age groups in all mouse strains. It is known that the serum TTR level is a good biomarker for nutritional state [48, 49]. As experimental mice are kept in good housing conditions, the serum TTR levels may not change under such conditions.

Gel filtration assay revealed that most hRBP4 was free of hTTR. Free hRBP4 was easily excreted to urine due to its small protein size. However, serum RBP4 levels were 0.5–0.6 μg/ml and the ratio of RBP4 to TTR (5.7–7.5 ug/ml) was about 1:10 in our double-humanized mice. This ratio is similar to human status which is 1/5 to 1/10. Thus, we do not know why the hTTR/hRBP4 ratio was kept within normal range despite most hRBP4 was free of hTTR. Before gel filtration assay, the serum sample was diluted two times with phosphate buffer. This treatment may cause dissociation of TTR/RBP4 complex, especially when TTR and RBP4 concentration is low.

Taken together, the presence of human RBP4 instead of mouse RBP4 and the absence of mouse TTR might be responsible for the accelerated amyloid deposition in our hV/hV:hR/hR mice despite the low serum hTTR level. These data suggest that our models recapitulated the pathological findings in FAP more precisely. In addition, this model will be necessary for the appropriate evaluation of effective therapeutics. However, our mice does not represent human TTR status as far as the low serum hTTR level and high percentage of unbound TTR with RBP4 are concerned. In our double-humanized mice, mean serum RBP4 levels were 0.5–0.6 μg/ml and the ratio of RBP4 to TTR was ~1:10. This ratio is similar to that of the human situation (1/5–1/10). We wondered whether amyloid deposition occurred under such a low serum level of hTTR. RNA interference (RNAi) therapy is now under investigation. In this case, we expect very low level of serum hTTR and RBP4. So, the question is that can we stop amyloid deposition by lowering the hTTR expression. Serum level of hTTR in our mice is 1/25 of that in human. Thus, our data might be useful to consider the effect of RNAi therapy. The double-humanized mouse at the Ttr and Rbp4 loci, in which serum TTR and RBP4 levels are equivalent to those in humans, represents a promising approach to replicating more precisely the disease process.