Pathological changes of renal epithelial cells in mice transgenic for the TT virus ORF1 gene

TT virus (TTV) was isolated in 1997 from the serum of a patient with post-transfusion hepatitis of unknown aetiology (Nishizawa et al., 1997 ). This virus is a non-enveloped, negative-stranded, circular DNA (3·8 kb) virus that belongs to the family Circoviridae (Miyata et al., 1999 ). It has been reported that the titre of TTV DNA in the liver was between 10- and 100-fold higher than that in the serum and that the circular, double-stranded form of TTV DNA, the intermediate of replication, was found in the liver, indicating that TTV replication occurs in the liver (Okamoto et al., 2000b ). Since TTV DNA was detected frequently in the sera of patients with post-transfusion non-AG hepatitis and TTV DNA titres in these patients correlated with levels of serum alanine aminotransferase (ALT) (Nishizawa et al., 1997 ), it was suggested that TTV might be involved in the pathogenesis of post-transfusion non-AG hepatitis (Tuveri et al., 2000 ).

In addition to blood transfusion transmission, TTV seems to be non-parentally transmitted via the faecaloral route, as TTV DNA has been detected in both faeces (Okamoto et al., 1998 ) and saliva (Ross et al., 1999 ). TTV DNA was found also in breast milk, indicating post-natal transmission of TTV (Kazi et al., 2000 ). These routes of TTV transmission could contribute to the high prevalence of TTV infections in the general population worldwide. Recent epidemiological studies have shown that TTV infection is very frequent, with a high number of asymptomatic individuals. The high rate of asymptomatic TTV infection in the population has suggested that TTV probably does not cause hepatitis (Allain, 2000 ; Nishiguchi et al., 2000 ). However, TTV has an extremely wide divergence in the sequence of its genome (Hijikata et al., 1999 ) and is classified into at least 16 subtypes, according to differences in the genome sequence of more than 30% from one another (Okamoto et al., 1999 ). Therefore, it is possible that some of the genotypes may be capable of causing hepatitis (Okamoto et al., 1999 ; Okamura et al., 2000 ; Sugiyama et al., 2000 ). Moreover, TTV-related virus (Hijikata et al., 1999 ; Khudyakov et al., 2000 ) and TTV-like mini virus (Takahashi et al., 2000 ) were discovered recently, suggesting that these viruses, which were not distinguished previously from TTV, may be involved in the development of the disease.

The TTV genome has a GC-rich region that contains a potential stemloop structure, which is highly conserved among TTV genotypes. This region appears to be involved in virus replication and acts as the origin of replication (Miyata et al., 1999 ). It was reported that there are five open reading frames (ORFs) in the TTV genome, although we have found an additional ORF, ORF6, by a computer search (Fig. 1A). Some of the transcripts of TTV were spliced in a process by which ORF2 was linked to either ORF4 or ORF5, tentatively designated ORF2-4 and ORF2-5, respectively (Kamahora et al., 2000 ). However, the roles of each ORF gene product have not been clarified yet.

(20K): Fig. 1. (A) Structure of the linearized TTV genome. Six ORFs, deduced from a computer search, are shown. ORF2 links to ORF4 or ORF5 by splicing, resulting in ORF2-4 or ORF2-5, respectively. The positions of each ORF in the TTV genome are depicted. (B) The BamHIBamHI fragments containing each TTV ORF were used for microinjection.

While the epidemiological relation of TTV with hepatitis has been studied extensively, the significance of TTV infection to extrahepatic diseases remains to be elucidated. In this study, we produced mice transgenic for the TTV ORF1 gene and found that these mice developed pathological changes, not in the liver, but in the kidney. Transgene construction.
The cDNAs of TTV ORFs 1, 2-4, 2-5, 3 and 6, were amplified by PCR from the plasmid pTV-TRM1-1 (Okamoto et al., 1999 ). The TRM1 strain (DDBJ, accession no. AB038340) is a genotype 1a TTV isolated from a patient with hepatitis C virus-associated hepatocellular carcinoma (Nishizawa et al., 1999 ). The positions of the TTV ORFs are shown in Fig. 1(A). The primer sequences used for PCR are shown in Table 1. For efficient translation, six nucleotides preceding the initiation codon within the forward primers were replaced by the Kozak sequence (Kozak, 1989 ). The ORF2-4 fragment was constructed as described below. First, fragments of ORFs 2 and 4 were amplified by PCR using the ORF2 forward and ORF2-4 reverse primers and the ORF4 forward and reverse primers, respectively. The fragments were then mixed and used as the templates for second-round PCR using the ORF2 forward and ORF4 reverse primers to produce the ORF2-4 fragment. Since 30 bp of the 5' terminus of the ORF2-4 reverse primer and the ORF4 forward primer are designed to be complementary to each other, these two templates annealed to each other at the complementary region, resulting in the connection of ORF2 to ORF4. ORF2-5 was constructed in the same way. PCR was performed under the following conditions: one cycle at 94 °C for 2 min, 20 cycles of 94 °C for 30 s, 50 °C for 30 s and 72 °C for 45 s and 1 cycle at 68 °C for 7 min. The amplified fragments of the TTV ORF genes were sequenced and inserted into the XhoI site of pCXN2, comprising the cytomegalovirus (CMV) enhancer, the chicken β-actin promoter and the rabbit β-globin poly(A) signal (a gift from Junichi Miyazaki, Osaka University, Japan) (Niwa et al., 1991 ). To isolate the vector sequence, the BamHIBamHI fragments were cut from each pCXN2-TTV ORF and used for microinjection. For cloning the spliced TTV ORF1, 100 ng kidney mRNA from an LT1 mouse was reverse transcribed (50 °C for 30 min) and amplified by PCR (1 cycle at 94 °C for 2 min, 35 cycles of 94 °C for 30 s, 53 °C for 30 s and 68 °C for 45 s and 1 cycle at 68 °C for 7 min) using the Titan One Tube RTPCR kit (Roche) and the TTV ORF1 forward and reverse primers.

Table 1. PCR primer sequences used to clone the TTV ORF genes

Production of transgenic mice.
The BamHIBamHI fragments from each pCXN2-TTV ORF were electrophoresed in a 0·8% agarose gel and purified using the GeneClean II kit (BIO 101). The fragments were dissolved in 10 mM TrisHCl and 0·25 mM EDTA (pH 7·5) at a concentration of 500 copies/pl and microinjected into the male pronuclei of fertilized C3H/HeN mouse eggs. Genotyping was performed by dot-blot hybridization and PCR using DNA extracted from mouse tails. Mice were kept under specific-pathogen-free conditions in an environmentally controlled clean room at the Laboratory Animal Research Centre, Institute of Medical Science, University of Tokyo, Japan. Experiments were conducted according to the Institutional Ethical and Safety Guidelines for Animal Experiments and Safety Guidelines for Gene Manipulation Experiments.

RNA analysis.
Total RNA was extracted using the acid guanidium thiocyanatephenolchloroform method (Chomczynski & Sacchi, 1987 ). mRNA was purified from total RNA using the QuickPrep Micro mRNA Purification kit (Pharmacia). For Northern blot analysis, 2 µg mRNA was electrophoresed in a 1·2% agarose gel containing 50% formaldehyde and blotted onto a Gene Screen Plus membrane (Du Pont NEN). After UV cross-linking, the membrane was hybridized with a ³²P-labelled DNA probe at 42 °C overnight. Autoradiograms were developed and the radioactivity of each band was quantified using the BAS 2000 BioImage analyser (Fuji).

Biochemical analysis of mouse sera.
To determine the levels of serum albumin (Alb), serum creatinine (Cr), blood urea nitrogen (BUN) and total serum cholesterol (T-chol), 10 µl of sera from 4-week-old wild-type and LT1 mice and a 2-month-old Snd15 mouse were applied onto DRI-CHEM slides (Fuji) and measured using the DRI-CHEM 5500V (Fuji).

Histological analysis.
The kidneys from 4-week-old wild-type and LT1 mice and a 2-month-old Snd15 mouse were fixed in 10% formalinPBS, paraffin-embedded and sectioned at a thickness of 2 µm, according to standard procedures. The sections were stained with haematoxylin and eosin (H&E) or periodic acidSchiff (PAS). For electron microscopy, tissues were fixed with 1% glutaraldehyde in PBS, embedded in Epon and sliced at a thickness of 70 nm. The ultra-thin sections were stained with uranyl acetate and lead citrate. All samples were examined under a JEOL 1200EX electron microscope.

Immunohistochemistry.
Paraffin-embedded sections (2 µm) were immunostained against proliferating cell nuclear antigen (PCNA) using the Rabbit ImmunoCruz Staining system (Santa Cruz Biotechnology), according to the manufacturers instructions. In brief, the deparaffined sections were incubated with 2·5 µg/ml of rabbit anti-PCNA polyclonal antibody (Santa Cruz Biotechnology) in 5% normal goat serum in PBS for 1·5 h at room temperature. Then, the sections were incubated with biotinylated secondary antibody and subsequently incubated with streptavidinhorseradish peroxidase. The sections were counterstained with methyl green.

Production of transgenic mice
Initially, we attempted to produce transgenic mice that carried multiple TTV ORF genes. DNA fragments of TTV ORFs 1, 2-4, 2-5, 3 and 6 (Fig. 1A), respectively, were inserted into a mammalian expression cassette containing the CMV enhancer, the β-actin promoter and the β-globin poly(A) signal (Fig. 1B). For efficient translation, six nucleotides preceding the ATG codon of the transgenes were replaced with the Kozak sequence. These transgenes were mixed and injected into 838 fertilized mouse eggs. Eight transgenic mice, Fst04, Fst09, Fst16, Fst27, Fst46, Fst54, Fst59 and Fst67, were produced among the 84 mice born. These transgenic mice carried the transgene(s) in various combinations (Table 2). Because Fst54 was sterile and died at 2 months of age, the mouse could not be studied further. Fst04, Fst09 and Fst16 could transmit the transgenes to the next generation but the other transgenic mice could not. Since the transgenes of Fst09 were integrated into separate sites of the chromosomes, Fst09 generated two sublines, LT1 and LT2. LT1 mice carried the ORF1 gene and LT2 mice carried the ORF2-4 and ORF2-5 genes. The total copy number of the transgenes in LT1 and LT2 mice was two and three, respectively. All of the transgenic mice, except for those of the LT1 subline, did not show any apparent phenotype. However, LT1 mice carrying the ORF1 gene showed growth retardation with a reduced average body weight (Fig. 2A, C). These mice gradually accumulated ascites (Fig. 2B) and died before reaching 5 weeks of age (Fig. 2D), implying that the TTV ORF1 gene had a pathogenic effect. In order to confirm this effect, we carried out a second trial to produce other lines of transgenic mice carrying the TTV ORF1 gene. The DNA fragment of TTV ORF1 was injected into 1002 fertilized mouse ova. Three transgenic mice, Snd15, Snd66 and Snd75, were obtained among the 121 mice born (Table 2). These three transgenic mice were mosaic for the TTV ORF1 gene and did not transmit it to the next generation. Snd66 and Snd75 grew normally, but Snd15 showed growth retardation, although it did not develop ascites. The body weights of LT1 mice (7·2±0·8 g) and Snd15 (12·0 g) at 4 weeks of age were significantly smaller than those of wild-type littermates (14·2±1·2 g). Because growth retardation was observed in two independent transgenic mice, a toxic effect of TTV ORF1 gene expression to mouse growth was suggested. The low ratio (less than 10%) of transgene transmission implied that expression of the ORF1 gene affects embryonic development (Table 2).

Table 2. Production of transgenic mice

(37K): Fig. 2. Characteristics of TTV ORF1 transgenic mice. (A) A growth-retarded LT1 mouse (right) and a non-transgenic littermate (left) at 4 weeks of age. (B) Development of severe ascites in the LT1 mouse. (C) Body weights of LT1 mice (n=4) and Snd15 at 4 weeks of age. (D) Survival curve of LT1 mice (squares, n=4) and non-transgenic littermates (circles, n=4). All LT1 mice died before reaching 5 weeks of age.

Expression level and splicing of the TTV ORF1 gene transcript
Transgene expression was analysed by Northern blot hybridization. LT1 mice and Snd15, which showed growth retardation, expressed the ORF1 gene. However, no expression of the ORF1 gene was detected in other transgenic mice, which carried the ORF1 gene but did not show any apparent phenotype (Table 2). This result also suggested that ORF1 gene expression was linked to toxicity. In the following studies, we analysed further the pathological changes of LT1 mice and Snd15. The highest level of transgene expression was detected in the kidneys of LT1 mice (Fig. 3A, B). On the other hand, in Snd15, a high level of transgene expression was observed in the brain and a medium level of expression was observed in the kidney. The expression levels in the kidney of the Snd15 mouse were approximately 20-fold lower than that in the kidneys of LT1 mice, suggesting that the difference in the pattern of expression resulted in no accumulation of ascites in the Snd15 mouse.

(39K): Fig. 3. (A) Expression of TTV ORF1 in an LT1 mouse and Snd15. Northern blot analysis using mRNAs from transgenic and wild-type mice was performed as described in Methods. The β-actin probe was used as an internal control. (B) Relative expression levels of the spliced form of the TTV ORF1 gene. The radioactivity of each band in (A) was quantified by an Image analyser and normalized using the levels of the internal control. The results are shown as relative levels. (C) The splicing scheme of the TTV ORF1 gene transcript. Positions of the splicing donor and acceptor sites in the TTV genome are shown (arrows). Boxes represent the coding regions for the spliced form of ORF1. The untranslated regions (wavy lines) and the intron (straight line) are indicated. (D) Amino acid sequence deduced for the spliced form of TTV ORF1. The protein has an arginine-rich domain at the N-terminal end and a serine-rich domain at the C-terminal end. Arginine and serine residues are indicated in bold.

Although the size of the TTV ORF1 gene transcript was expected to be 3·1 kb, transcripts of 1·1 kb in size were detected (Fig. 3A). This result suggested that the 1·1 kb transcript was a spliced product. To confirm this, the 1·1 kb transcript was isolated from the kidney of an LT1 mouse, cloned using RTPCR and sequenced. The result revealed that the 1·1 kb transcript was spliced and joined nt 575 with nt 2431. The newly generated ORF encoded a putative product of 143 aa (DDBJ accession no. AB055937) (Fig. 3C, D).

The donor and acceptor sites of the 1·1 kb transcript were identical to those of ORF2-5 (Fig. 1A and Fig. 3C). The reading frame of the spliced form of ORF1 was switched from frame 1 to frame 2 at the splicing junction. The putative protein encoded by the spliced form of ORF1 was highly basic, with an arginine-rich domain at the N-terminal end and a serine-rich domain at the C-terminal end (Fig. 3D).

Renal failure of LT1 mice
Since LT1 mice showed an accumulation of ascites, we suspected kidney function failure in these mice and, thus, measured the levels of serum Cr, serum Alb, BUN and T-chol. As shown in Table 3, the levels of serum Cr and BUN from LT1 mice were significantly higher than those of wild-type mice. This suggested that the renal function of LT1 mice was impaired, resulting in uraemia. The level of BUN in the Snd15 mouse was also high, but 6-fold less than that seen in LT1 mice. In LT1 mice, the level of serum Alb was very low, while it was normal in the Snd15 mouse. Although proteinuria could not be determined precisely because of poor urination, an approximate level of proteinuria was estimated using urinalysis paper. The level of proteinuria in LT1 mice was at least 3-fold higher than that seen in wild-type mice, but the level of proteinuria in the Snd15 mouse was normal (data not shown). Since nephrotic syndrome is characterized by massive proteinuria and hypoalbuminaemia, these results indicated that LT1 mice developed nephrotic syndrome. This diagnosis was supported by finding high levels of serum T-chol in LT1 mice, which is another feature of nephrotic syndrome. On the other hand, serum ALT levels in LT1 mice and Snd15 were normal, suggesting that these transgenic mice did not develop hepatitis (data not shown). In fact, the livers of LT1 mice and Snd15 were examined under light microscopy and no abnormality was observed (data not shown).

Table 3. Biochemical analysis of transgenic mouse sera

Morphological changes and abnormal proliferation of renal epithelial cells
Histological changes were examined microscopically. In LT1 mice, severe abnormalities were observed in the kidney (Fig. 4B), where very high levels of ORF1 gene expression were detected. Many cysts filled with proteins were observed in the dilated lumen of the uriniferous tubule, especially in the distal tubule. The morphological study of capsule cells of Bowmans capsule revealed that they were very flat in wild-type mice (Fig. 4D), while the capsule cells of LT1 mice were enlarged, with large vacuoles that resulted in the occupation of Bowmans space (Fig. 4E).

(145K): Fig. 4. Histopathology and immunohistochemistry of the kidneys from transgenic mice. (A) A normal kidney from a wild-type mouse. (B) Cyst formation in a kidney from an LT1 mouse. (C) An apparently normal kidney from the Snd15 mouse. (D) Normal glomeruli in a kidney from a wild-type mouse. (E) Abnormal renal corpuscles and degenerated renal tubules in a kidney from an LT1 mouse. (F) Enlarged capsule cells of the renal corpuscles in a kidney from the Snd15 mouse. (E, F) Some of the abnormal capsule cells (arrows) and vacuolar degenerated distal tubule cells (arrowheads) are indicated. (GI) Immunostaining with PCNA to detect proliferating cells in a kidney from a wild-type mouse (G), an LT1 mouse (H) and the Snd15 (I) mouse. Some of the PCNA-positive cells in the uriniferous tubule (arrows) or glomerulus (arrowheads) are indicated. (A)(C) Magnification, x4; H&E-stained. (D)(F) Magnification, x20; PAS-stained. (G)(I) Magnification, x20; methyl green-counterstained.

A similar enlargement of capsule cells was also found in the Snd15 mouse. None of the capsule cells in Snd15 formed squamous epithelia, but formed simple cuboidal epithelia similar to the proximal tubule cells (Fig. 4F). In both LT1 mice and Snd15, vacuolar degeneration was found in some of the proximal tubular cells (Fig. 4E, F). For both transgenic mice, no abnormalities were found in other parts of the kidney, including the collecting tubes, calyx and the renal papilla, as well as a number of glomeruli. In addition, no inflammatory cells were observed in the kidney. These results suggest that expression of TTV ORF1 affects the morphology of renal epithelial cells.

It is accepted generally that excessive proliferation of renal epithelial cells is involved in the development of various human kidney diseases. Because of this, we examined the proliferation of renal cells using PCNA as a proliferating cell marker. In the kidney cortex of LT1 mice and Snd15, some of the nuclei in distal and proximal tubular cells were stained deeply when stained with an anti-PCNA antibody (Fig. 4H, I). The number of PCNA-positive tubular epithelial cells was 10- and 3-fold higher in LT1 mice and Snd15, respectively, when compared with wild-type mice (negative control, data not shown). A result of particular interest is that 1 out of 10 glomeruli from the Snd15 mouse had PCNA-positive cells (Fig. 4I), which were not found in the glomeruli of LT1 and wild-type mice.

Electron microscope analysis of renal epithelial cells
In order to investigate the kidney changes in more detail, electron microscope analysis was performed on LT1 and wild-type mice. The results showed that the number of mitochondria and microvilli in the proximal tubule cells from LT1 mice was decreased significantly, compared with normal levels in wild-type mice (compare Fig. 5B with A). Mitochondria were distributed diffusely in the cytoplasm of the proximal tubule cells. There were also fewer endocytotic vesicles in the cytoplasm of proximal tubular cells as compared to those in the wild-type mouse. In addition, necrotic profiles were observed in some of the proximal tubular cells from the LT1 mouse. These observations suggest that the function of the proximal tubule cells is impaired in LT1 mice. As shown in Fig. 5(D), the glomeruli of the LT1 mouse showed severe abnormalities. The profile of podocytes was extremely swollen and appeared immature. The normal shape of a mature podocyte is very flat (Fig. 5C), like a capsule cell. The foot processes in the LT1 mouse were completely absent. Electron-dense deposits, which denote immune-complex deposition, were not detected in the glomeruli of the LT1 mouse.

(180K): Fig. 5. Electron microscope study on a kidney from a transgenic mouse in comparison to that from a wild-type mouse. (A, B) Proximal tubules of a wild-type mouse (A) and an LT1 (B) mouse. In the LT1 mouse, the number of mitochondria and microvilli is significantly decreased. (C, D) Glomeruli of a wild-type (C) and an LT1 (D) mouse. Drastic changes of glomerulus of the LT1 mouse are seen. The profiles of podocytes are extremely swollen and appear immature. The foot process is lacking. P, podocyte; E, vascular endothelial cell; BS, the space of Bowmans capsule. Some of the foot processes (arrows) and basement membrane (arrowheads) are indicated. Bar, 2 µm.

In this study, we produced mice transgenic for the TTV ORF genes in various combinations and found that mice expressing the ORF1 gene showed morphological changes in renal epithelial cells. The transcript of ORF1 was spliced and encoded a putative 143 aa protein with an arginine-rich domain at the N-terminal end and a serine-rich domain at the C-terminal end. All of the LT1 mice that expressed the ORF1 gene at high levels in the kidney developed renal failure resembling that of nephrotic syndrome and died before reaching 5 weeks of age. Many cysts were observed in the dilated lumen of the uriniferous tubule in LT1 mice. The podocytes and capsule cells in LT1 mice were extremely swollen and appeared to be immature. The foot processes of podocytes in LT1 mice were completely absent. Since the foot process is essential for glomerular function and acts as a filtration barrier (Farquhar, 1975 ), the lack of podocyte foot processes in LT1 mice could be a likely cause of renal failure and ascites. Progression of these symptoms seems to be the cause of early death in LT1 mice.

Another transgenic mouse, Snd15, which presented a lower level of ORF1 gene expression (1/20 compared with LT1 mice), also showed similar, but milder, changes in the renal epithelial cells. All of the capsule cells in the Snd15 mouse were enlarged abnormally and filled Bowmans space. Since the normal morphology of the podocyte and capsule cell changes from a cuboid shape to a very flat shape during the maturation process, the enlarged shape of the capsule cells in the Snd15 mouse also suggests that these cells are immature. In both LT1 mice and Snd15, vacuolar degeneration was confirmed in some of the proximal tubular cells. Furthermore, the number of proliferating cells in the renal epithelia was increased significantly in both of these mouse types, as determined by the number of PCNA-positive cells. This observation suggests that immature cells are increased in this transgenic mouse, because cells that have completed terminal differentiation tend to stay at the G₀ phase and proliferative activity is low relative to that of immature cells. These results suggest that the expression of the TTV ORF1 gene in mice affects the maturation of renal epithelial cells in a dose-dependent manner. Since two independent founder mice and their derivatives developed similar renal epithelial cell abnormalities, it is clear that these abnormalities are not caused by the effects of the transgene insertion into the host chromosome, but, rather, by the toxic effects of the ORF1 gene product.

It is known that immunological mechanisms are involved in the development of various kidney diseases. The electron microscope study showed that the electron-dense deposits, which denote immune-complex deposition, were not found in the glomeruli of LT1 mice. Immune-complex deposition in the glomeruli was also examined by immunofluorescent staining and no significant difference between the transgenic and wild-type glomeruli was observed (data not shown), suggesting that immunological mechanisms were not involved in the generation of renal abnormalities in our transgenic mice. This is consistent with the observations from histological analysis in which no inflammatory cell infiltration was found in the kidneys of these transgenic mice.

The most prominent characteristic of LT1 mice is a dysplasia of renal epithelial cells, including podocytes, capsule cells and proximal tubule cells, while the other parts of the kidney and other tissues were histologically normal. During the development of the kidney, podocytes, capsule cells and distal and proximal tubule cells arise from common precursor cells called induced metanephrotic mesenchymal cells (Horster et al., 1999 ). After induction by the uretic bud, the precursor cells aggregate and transmit themselves to the epithelial lineage, which differentiates terminally to podocytes, capsule cells and distal and proximal tubule cells. This transmission is known as mesenchymalepithelial transmission (MET). The results presented here suggest strongly that expression of the TTV ORF1 gene impairs specifically the processes of terminal differentiation of precursor cells to podocytes, capsule cells and distal and proximal tubule cells after MET and keeps them at the immature stage. The abnormalities of the Snd15 mouse were also restricted to the cells derived from the precursor cells.

Since TTV was isolated originally from a patient with post-transfusion hepatitis of unknown aetiology, most pathoepidemiological studies of TTV have focused on the association of this virus with hepatitis. There are only a few reports that have investigated the correlation between TTV and kidney disease, although it is well known that the co-prevalence of TTV infection was significantly greater in haemodialysis patients (Campo et al., 2000 ; Gallian et al., 1999 ; Maggi et al., 1999 ; Martinez et al., 2000 ). Therefore, we think it is important to investigate the potential effects of TTV infection on the progression of the disease in haemodialysis patients and patients with kidney disease.

Kamahora et al. (2000) examined the transcriptional profile of TTV in COS-1 cells and detected 3·0, 1·2 and 1·0 kb mRNAs corresponding to ORF1, ORF2-4 and ORF2-5 mRNA, respectively. These mRNAs were transcribed from a single promoter and were spliced alternatively. All of the mRNAs were observed in the bone marrow cells of a patient with acute myeloblastic leukaemia (Okamoto et al., 2000a ). The splicing sites of the spliced form of ORF1 found in our transgenic mice are identical to those of ORF2-5 and the complete sequence encoded by the spliced form of ORF1 lies completely within the 1·0 kb TTV mRNA, which is detected in COS-1 cells and in leukaemia patients. Because of the lack of reliable antibodies against TTV proteins, it has not been shown that the spliced product of ORF1 nor the other products of the other TTV genes are produced in infected cells.

The putative product encoded by the spliced form of ORF1 has an arginine-rich domain at the N-terminal end and a serine-rich domain at the C-terminal end. The arginine-rich domain of the putative product shows 63% identity to protamine, which binds to chromosomes and tightly condenses them (Wouters-Tyrou et al., 1991 ). The positive charge of the arginine-rich domain in protamine is essential to bind and neutralize the negative charges of the DNA strands (Shiffman et al., 1978 ). It is accepted generally that the binding of protamine inactivates gene transcription by the process of chromosome condensation, because the condensed chromosome tends to prevent transcriptional factors from accessing the target sites (Mahlknecht & Hoelzer, 2000 ). We speculate that the putative product binds to the host chromosomes using the arginine-rich domain and then alters the secondary structure of the DNA strands around the binding sites. As a result, the regulation of gene transcription around the binding sites may be affected. Some of the genes affected by the putative product may be important for the differentiation of renal epithelial cells, thus resulting in the phenotypes observed in our mice.

Although the association between TTV infection and hepatitis has been suspected from prior epidemiological studies (Okamoto et al., 1999 ; Okamura et al., 2000 ; Sugiyama et al., 2000 ), in our transgenic mice, the liver was histologically and serochemically normal. Nevertheless, this does not exclude the possibility that the putative product of the spliced form of the TTV ORF1 gene is involved in the development of liver disease. This is because TTV ORF1 has been expressed from the early embryonic stage in our mice and, therefore, immunological tolerance to the transgene product may be established in our transgenic mice already. Further studies are required to elucidate the pathogenicity of TTV.

Taken together, we have shown that expression of the TTV ORF1 gene affects the differentiation of renal epithelial cells. The results of this study may provide new insights for the investigation of TTV pathogenicity.

The authors thank Dr Hiroshi Sagara for his excellent technical assistance in the electron microscope study and also for his valuable discussion. We also thank Dr Mutsumi Inaba for his technical support. We are grateful to the members of our laboratory for their helpful insights and animal care. This work was supported by grants from the Ministry of Education, Science, Sports, Culture and Technology of Japan.

Footnotes

^b Present address: Division of Molecular Virology, Institute for Genetic Medicine, Hokkaido University, Hokkaido, Japan.