Tau isoform expression and phosphorylation in marmoset brains

Tau is a microtubule-associated protein expressed in neuronal axons. Hyperphosphorylated tau is a major component of neurofibrillary tangles, a pathological hallmark of Alzheimer's disease (AD). Hyperphosphorylated tau aggregates are also found in many neurodegenerative diseases, collectively referred to as “tauopathies,” and tau mutations are associated with familial frontotemporal lobar degeneration (FTLD). Previous studies have generated transgenic mice with mutant tau as tauopathy models, but nonhuman primates, which are more similar to humans, may be a better model to study tauopathies. For example, the common marmoset is poised as a nonhuman primate model for investigating the etiology of age-related neurodegenerative diseases. However, no biochemical studies of tau have been conducted in marmoset brains. Here, we investigated several important aspects of tau, including expression of different tau isoforms and its phosphorylation status, in the marmoset brain. We found that marmoset tau does not possess the “primate-unique motif” in its N-terminal domain. We also discovered that the tau isoform expression pattern in marmosets is more similar to that of mice than that of humans, with adult marmoset brains expressing only four-repeat tau isoforms as in adult mice but unlike in adult human brains. Of note, tau in brains of marmoset newborns was phosphorylated at several sites associated with AD pathology. However, in adult marmoset brains, much of this phosphorylation was lost, except for Ser-202 and Ser-404 phosphorylation. These results reveal key features of tau expression and phosphorylation in the marmoset brain, a potentially useful nonhuman primate model of neurodegenerative diseases.


Edited by Paul E. Fraser
Tau is a microtubule-associated protein expressed in neuronal axons. Hyperphosphorylated tau is a major component of neurofibrillary tangles, a pathological hallmark of Alzheimer's disease (AD). Hyperphosphorylated tau aggregates are also found in many neurodegenerative diseases, collectively referred to as "tauopathies," and tau mutations are associated with familial frontotemporal lobar degeneration (FTLD). Previous studies have generated transgenic mice with mutant tau as tauopathy models, but nonhuman primates, which are more similar to humans, may be a better model to study tauopathies. For example, the common marmoset is poised as a nonhuman primate model for investigating the etiology of age-related neurodegenerative diseases. However, no biochemical studies of tau have been conducted in marmoset brains. Here, we investigated several important aspects of tau, including expression of different tau isoforms and its phosphorylation status, in the marmoset brain. We found that marmoset tau does not possess the "primate-unique motif" in its N-terminal domain. We also discovered that the tau isoform expression pattern in marmosets is more similar to that of mice than that of humans, with adult marmoset brains expressing only four-repeat tau isoforms as in adult mice but unlike in adult human brains. Of note, tau in brains of marmoset newborns was phosphorylated at several sites associated with AD pathology. However, in adult marmoset brains, much of this phosphorylation was lost, except for Ser-202 and Ser-404 phosphorylation. These results reveal key features of tau expression and phosphorylation in the mar-moset brain, a potentially useful nonhuman primate model of neurodegenerative diseases.
Tau is a microtubule (MT) 4 -associated protein expressed primarily in the axons of mammalian neurons (1)(2)(3)(4). Tau also constitutes a major component of the neurofibrillary tangles (NFTs) found in neurons of brains of Alzheimer's disease (AD) patients. Tau is a naturally unfolded protein composed of an N-terminal projection domain and C-terminal region containing MT-binding (MTB) repeats. There are six isoforms of tau, which are generated by alternative splicing of N-terminal insertions (exons 2/3) and the C-terminal second MTB repeat (exon 10) from a single gene. In particular, splicing out or in from exon 10 gives rise to three-repeat (3R) or four-repeat (4R) tau with different MT-binding activities. This splicing is regulated developmentally (5), and an imbalance between 3R and 4R tau is found in AD (3). However, the molecular mechanisms of this splicing and its significance are not fully understood.
The MTB and MT assembly activities of tau are regulated by phosphorylation (4,6,7). Tau in NFTs is hyperphosphorylated (8 -11), and aggregates of hyperphosphorylated tau are also detected in many neurodegenerative diseases, which are collectively referred to as "tauopathies" (2)(3)(4). The immunostaining of NFTs with several anti-phospho-tau antibodies is routinely employed to provide a final diagnosis of AD or other tauopathies (12,13), but the significance of tau phosphorylation for the development of AD or tauopathies is not yet clear. This is because of the difficulty of studying the phosphorylation states of tau in human brains. Most previous studies have used autopsy human brain samples with long post-mortem interval (PMI), during which tau undergoes dephosphorylation. In fact, tau in mouse brain is shown to be dephosphorylated quickly at a number of sites after sacrifice (16). We recently examined in detail the phosphor-ylation of tau in human brains, both for normal control and AD patients with 3-4 h of shortest PMI (14) by Phos-tag phosphoaffinity electrophoresis (15). There was abundant nonphosphorylated tau in human brains. In contrast, tau in biopsy brain samples taken from live humans is phosphorylated more than tau in postmortem autopsy samples (17). However, the effect of anesthesia was not considered at that time. It has been recently shown that anesthesia increases tau phosphorylation (18). Therefore, the exact in vivo phosphorylation states of tau in human brains are not yet satisfactorily uncovered.
Rodents have alternatively been used for the analysis of in vivo tau phosphorylation (8,19,20). They might have provided useful information implicating the phosphorylation of tau in human brains, but there are arguments that rodents do not necessarily serve as a proper model of tauopathies, which develop in aged individuals. The life span of rodents is much shorter than that of humans. The isoform expression of mouse tau is different from that of humans. Whereas 3R and 4R tau are expressed in adult human brains, only 4R tau is expressed in adult mouse brains (21)(22)(23)(24). Further, the hyperphosphorylation and aggregation of tau do not occur in mouse brains. The desirable model is that of the nonhuman primate (25), which captures the pathological features of tau observed in humans. The common marmoset (Callithrix jacchus), a New World monkey, is poised as a nonhuman primate model for age-dependent neurodegenerative diseases (26,27). Compared with other nonhuman primates, such as macaque and rhesus monkeys, marmosets are easy to house and handle due to their small size of ϳ300 g and have a relatively short life span with a maximum age of 16 -18 years (25). Moreover, it is possible to manipulate the genes of marmosets (28). In fact, marmosets have already been used as the model of 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP)-induced Parkinson disease and age-dependent accumulation of ␣-synuclein, for example (29,30). We also used marmosets in a study on the propagation of ␣-synuclein aggregates (31). Before using animals as a model, however, it is useful to identify the properties of the target protein (e.g. tau in the case of tauopathies). To our surprise, however, no biochemical work has been conducted on tau proteins in marmoset brains.
In this study, we isolated tau cDNA from marmoset brains and investigated the expression of tau isoforms and phosphorylation, two important properties of tau related to AD pathology. Unexpectedly, we found that marmoset tau resembled mouse tau more than human tau in isoform expression.

Marmoset tau is phylogenically distinct from the tau of human and Old World monkeys
The predicted amino acid sequence of marmoset tau can be obtained from the NCBI database (XP_017827929.1). The number of amino acids of this tau is 851, which may be constructed from all exons in the tau gene and is larger than highmolecular weight (or big) tau expressed in the peripheral nervous system (32,33). Tau protein species expressed in the central nervous system are much smaller due to the omission of exons 4a, 6, and 8 (Fig. S1A). Thus, there have been no reports describing the amino acid sequence of tau isoforms expressed in mar-moset brains. Therefore, we cloned tau cDNA from cDNA libraries of marmoset brains on postnatal day 0 (P0) and of adults (23 months) using 5Ј and 3Ј primers based on the predicted DNA sequence by PCR (see "Experimental procedures"). 0N3R tau cDNA with no N-terminal insertion and three MTB repeats (3R) was isolated from the cDNA library of a newborn marmoset brain, and 0N4R (no N-terminal insertion and four MTB repeats (4R)) and 2N4R (two N-terminal insertions and 4R) were obtained from that of an adult marmoset brain. The sequence of cloned tau cDNA matched the mRNA sequence deduced from the genome sequence data (XM_017972440.1). The full amino acid sequence of 2N4R marmoset tau is compared with that of human and mouse tau in Fig. 1A Figure 1. Comparison of amino acid sequences of marmoset tau with human and mouse tau. A, amino acid alignment of marmoset tau with human and mouse tau. Amino acids unique to marmoset tau are indicated by italic boldface type, FTLD-tau mutation sites are indicated by red, and major phosphorylation sites are indicated by blue. B, homology of amino acid sequences of marmoset tau with human or mouse tau. The total number of amino acids found in marmoset 2N4R tau is 431, which is 10 amino acids shorter than that found in human tau and one amino acid longer than that found in mouse tau. The amino acid identity of marmoset tau and human tau is 94.3%, and that for mouse tau is 88.2%. The three species share 388 of the same amino acids. The number of amino acids that are the same between each pair of species as well as those unique among the three species are denoted in blue, and the number of amino acids common to all three species is denoted in red, respectively. C, immunoblots of marmoset and human tau with Tau5 recognizing common mammalian tau (top) and TauE1 specific to human tau (bottom).

The isoforms and phosphorylation of marmoset tau
amino acids were found, which is 10 amino acids shorter than the 441 amino acids found in human 2N4R tau. Marmoset 2N4R tau exhibited 94.3 and 88.2% identity to the corresponding isoforms of human and mouse tau, respectively, with 12 amino acids unique to marmoset tau and with 388 amino acids common to human and mouse tau (Fig. 1B). Twelve amino acids unique to marmoset tau were found in the N-terminal region (shown in italic boldface type in Fig. 1A). N-terminal exon 2/3 insertions are indicated as well. The C-terminal 200 amino acids, including the MTB repeats, are exactly the same as those of human tau. The MAPT mutation sites, which are found in frontotemporal lobar degeneration (FTLD) with tau-immunoreactive inclusions (FTLD-tau) (34), previously referred to as FTDP-17 (frontotemporal dementia and parkinsonism linked to chromosome 17) (2-4), were conserved in marmoset tau (shown in red in Fig. 1A). Major physiological and pathological phosphorylation sites were also conserved (shown in blue in Fig. 1A). Ten fewer amino acids were found in the N-terminal region of marmoset tau, which corresponds to amino acids 23-32 of human 2N4R tau. Interestingly, the missing region overlapped with the 11 amino acids absent in mouse tau, corresponding to amino acids 22-32 of human tau (Fig.  1A). The absence of the sequence in marmoset tau was confirmed by immunoblotting with the human-specific tau antibody generated against amino acids 20 -38 of human tau ( Fig.  1C; see "Experimental procedures"). Whereas adult marmoset tau appeared as two separated bands on SDS-PAGE (Fig. 1C, Tau5), neither band was recognized by the human-specific TauE1 antibody (Fig. 1C).
The amino acid alignment and phylogenic tree of mammalian tau have been recently reported (35). In that report, the N-terminal 11-amino acid insertion was defined as the "primate-unique motif" because the sequence was found in chimpanzees and macaques but not in bovines, goats, cats, or rodents (35). However, the sequence was not present in tau of marmosets, although the marmoset is a primate. To determine the evolutional positioning of marmoset tau, we performed an amino acid sequence alignment of mammalian tau by adding the number of mammalian species to the list given by Stefanoska et al. (35). Those added are the gorilla, orangutan, gibbon, and baboon as Old World monkeys; the squirrel monkey, night monkey, and capuchin as New World monkeys; the sifaka and lemur as monkeys of prosimians; the cat, panda, and goat as other mammals; and the Tasmanian devil as a marsupial animal. The alignment was performed with N-terminal 120 amino acids of human tau, and the N-terminal 50-amino acid region is shown in Fig. 2A. Considerable variation in the region corresponding to amino acids 19 -32 of human tau was found, including in the 11-amino acid stretch referred to as the primate-unique motif ( Fig. 2A, highlight). The "primate-unique motif" was found in humans, all Old World monkeys, and Madagascar monkeys, but not in New World monkeys. Among the other mammals, the corresponding sequence was found in dogs, cats, and pandas, whereas mice, rats, goats, and cows did not have the sequence. Cat tau registered in the Ensemble Genome Browser (MAPT-201) encodes the primate-unique motif, different from that in the previous list of Stefanoska et al. (35). A phylogenic tree of mammalian tau is shown in Fig. 2B.
Although levels of bootstrap support were generally low, the phylogenetic relationship was largely consistent with that inferred from larger datasets (36,37). This finding suggests that New World monkeys, including marmosets, lost that sequence of tau after their separation from Old World monkeys and that mouse/rat and cow/goat lineages also independently lost the corresponding sequence after their separation from other animal groups.

Isoforms of tau expressed in marmoset brains
The expression of tau isoforms changes with brain development. 3R tau is expressed in embryonic and early postnatal days in both humans and mice (21)(22)(23)(24)38). In contrast, whereas both 3R and 4R tau are expressed in adult brains of humans (21,23), only 4R isoforms are expressed in the brains of adult mice (20,24,38). To reveal the isoforms of tau expressed in the marmoset brains, mRNA transcripts of tau were examined by PCR using newborn and adult marmoset brain cDNA libraries as templates with primers encompassing the N-terminal insertion (exons 2/3) or the second repeat of MTB (exon 10) (Fig. S2). With respect to N-terminal insertions, the 0N isoform was amplified predominantly from the newborn cDNA library with faint bands at the sizes expected from the 1N and 2N isoforms (Fig. 3A, P0). When the adult cDNA library was used as a template, PCR products corresponding to 0N and 2N tau were amplified (Fig. 3A, 23 months (23M)). We confirmed them to be tau DNA fragments by DNA sequencing, and a weak band above 2N (asterisk in Fig. 3A) was an artifact including unknown sequence. For the splicing of exon 10, a 3R product was detected as a stronger band in newborn marmosets, whereas a weak band corresponding to 4R tau was also detected (Fig. 3B, P0). In contrast, 4R tau was exclusively detected in the adult marmoset cDNA library (Fig. 3B, 23M). A faint band above 4R at P0 (asterisk in Fig. 3B) was again a PCR artifact. The similar results were obtained from another set of primers (Fig.  S3). These results indicate that the isoforms of 0N3R and, to a lesser extent, 0N4R tau are expressed in the brains of newborn marmosets, whereas those of 0N4R and 2N4R are expressed in adult marmoset brains at the mRNA level.
Next, we examined the expression of tau isoforms at the protein level by immunoblotting lysates of the cerebral cortex with Tau5, a phosphorylation-independent antibody for total tau (Fig. 4A). Adult human and mouse brains at P5 were blotted on the same membrane for comparison. Marmoset tau at 23 months was composed of two groups of bands: major bands at 55 kDa and weaker bands at ϳ69 kDa. The tau of newborn brains at P0 appeared as a single band in the same position as the major bands of adult marmoset tau, and they displayed the same mobility as those of human and mouse tau, whereas the upper minor band of adult marmoset tau was larger than the larger, faint band of adult human tau (compare the first and third lanes in Fig. 4A).
To identify the isoforms of two band groups in adult marmoset tau, their electrophoretic mobilities were compared with those of recombinant 0N4R and 2N4R marmoset tau expressed in Neuro-2a cells (Fig. 4B). Lysates of both the cerebral cortex and hippocampus showed two groups of bands. The upper bands exhibited the same migration patterns as 2N4R tau, and The isoforms and phosphorylation of marmoset tau the lower bands followed the patterns of 0N4R tau, suggesting that the upper and lower bands of adult marmoset tau are composed of 0N4R and 2N4R isoforms, respectively. Interestingly, the ratio of the upper and lower bands differed among brain regions. The upper band was stronger than the lower band in the cerebellum (Fig. 4B, left), whereas the ratio of the lower band was much larger in the olfactory bulb than in the hippocampus (Fig. 4B, right).
As is shown in Fig. 4A, 3R and 4R isoforms exhibited the same mobility in SDS-PAGE as those observed for mouse brain tau (20) despite their differing molecular sizes, rendering the defi-nite identification of isoforms difficult, particularly when both isoforms were coexpressed. To determine whether 3R, 4R, or both are expressed at P0 or in adult marmosets, their brain lysates were immunoblotted with 3R (RD3)-and 4R (RD4)specific antibodies (Fig. 4C). Mouse tau at P5 and P60 was used as a reference for 3R and 4R tau, respectively. The blots indicate that both the upper and lower bands in adult marmoset brains were composed of 4R tau, whereas the 3R isoform was predominantly found in newborn marmoset brains in the same position as the lower bands of adult tau by SDS-PAGE ( Fig. 4C and original blots shown in Fig. S4). ****** ** *** **** ******* *** ** ****** **** ** * *** **** * ******* ***** ** ** ****** *** * **** ** **** ****** * The upper lineage bloc of species includes primates; humans, apes (the chimpanzee, gorilla, orangutan, and gibbon), and Old World monkeys (the macaque and olive baboon) belong to Catarrhini; New World monkeys (the marmoset, squirrel monkey, capuchin, and night monkey) belong to Platyrrhini; and the sifaka and mouse lemur belong to Strepsirrhini. The second block denotes mice, rats, and other mammals as Rodentia; goats and cows as Ruminantia; and cats, dogs, and pandas as Carnivora. Tasmanian devil tau is shown at the bottom. The sequence defined as the "primate-unique motif" is highlighted. Asterisks indicate that the identical amino acids are found in each bloc. B, phylogenic tree based on N-terminal 120 amino acids of mammalian tau sequences listed in A. Bootstrap values are shown next to the branches. The scale bar indicates units of the number of amino acid substitutions per site. The presence (ϩ) or absence (Ϫ) of the "primate-unique motif" is indicated on the right.

The isoforms and phosphorylation of marmoset tau
The isoforms of tau with different numbers of N-terminal insertions can be distinguished by their electrophoretic mobility on SDS-PAGE. However, phosphorylation affects tau mobility and renders their definite identification difficult. Therefore, we dephosphorylated brain tau by alkaline phosphatase and then subjected it to SDS-PAGE and immunoblotting. To draw comparisons with nonphosphorylated tau, recombi-nant human (rHum, six isoforms) and mouse (rMou, three isoforms of 0N3R, 0N4R, and 1N4R) tau, which were synthesized in Escherichia coli, were used (Fig. 4D). The dephosphorylated marmoset tau in adult brains included two sharp bands that were slightly smaller than the human 0N4R and 2N4R isoforms. In contrast, those of the newborn marmoset brain lysate presented one major band close to mouse 0N3R tau and one minor

The isoforms and phosphorylation of marmoset tau
band close to the human 1N3R isoform (Fig. 4D). To validate the 3R or 4R isoforms, dephosphorylated tau was reacted with RD3 or RD4 (Fig. 4D). The results clearly demonstrate that adult marmoset tau was almost exclusively composed of 4R tau (Fig. 4D, bottom) and that P0 tau was mainly composed of 3R (Fig. 4D, middle). We observed weak reactions from RD4 in the P0 lysate, suggesting that a small amount of 4R tau of 0N, 1N, or 2N is expressed in marmoset brains at P0. These results show that 0N3R tau is mainly expressed in newborn marmoset brains and that 0N4R and 2N4R are predominantly expressed in adult marmoset brains.

Marmoset tau can be phosphorylated nearly the same as human or mouse tau when expressed in cultured cells
We have previously shown that Phos-tag SDS-PAGE is powerful for the analysis of the detailed and total phosphorylation of tau by expressing it in culture cell lines (39). The phosphorylation of marmoset tau was compared with that of human or mouse tau exogenously expressed in Neuro-2a cells by immunoblotting with Tau5 after Phos-tag SDS-PAGE (Fig. 5). Marmoset, human, and mouse 0N4R tau appeared as broad bands at a molecular mass of ϳ55 kDa in Laemmli's SDS-PAGE (Fig. 5,  Laem). In contrast, marmoset, human, and mouse tau were separated into approximately nine bands with a similar pattern of bands by Phos-tag SDS-PAGE (Fig. 5, Phos). Each band represents a distinct phosphoisotype of 0N4R tau with different quantities and sites of phosphorylation (see Fig. 3 in Kimura et al. (39)). Densitometric scanning results confirmed that their banding patterns are almost identical, although some minor differences were found, such as the stronger intensity of band 5 in mouse tau (Fig. 5). Thus, marmoset tau can be phosphorylated in Neuro-2a cells at the same sites and to a similar degree as in human or mouse tau, indicating that marmoset tau can be used for the analysis of phosphorylation in place of human tau.

Phosphorylation of tau at AD pathological sites in marmoset brains
We examined the phosphorylation of tau at several AD pathological sites in marmoset brains. We used the anti-phosphotau antibodies AT8, AT180, and PHF-1, which are typically used for the diagnosis of AD by the immunostaining of NFTs in post-mortem brain sections (12,13). The top panel of Fig. 6A is an immunoblot of total tau found in the brain lysates of three marmosets at P0 and in adults. Because mouse tau at P5 is highly phosphorylated, we used it as a positive control of phosphorylation in an immunoreaction (Fig. 6A, left lane). A normal adult human brain lysate was also run to draw comparisons (Fig. 6A, right lane). Total tau is shown in the top panel with Tau5. Newborn marmoset tau was recognized by the AT8, AT180, and PHF-1 antibodies, although the reactivity was weaker than that observed for neonatal mouse tau. In contrast, adult marmoset and adult human tau did not react with AT8 and AT180. A very faint reaction with PHF-1 was detected from adult marmoset tau but not for adult human tau.
The above listed anti-phospho-tau AD epitopes are generated by double phosphorylation at Ser-202 and Thr-205 for AT8, at Thr-231 and Ser-235 for AT180, and at Ser-396 and Ser-404 for PHF-1 (40). We then examined whether only one or both of the sites are not phosphorylated in adult marmoset brains. In this experiment, the reactivity of adult marmoset tau was compared with that of two adult human brains, whose PMI was 3.2 and 4.36 h, respectively (Fig. 6B, Hum). For the AT8 site, Ser-202 was phosphorylated in adult marmoset and human tau to almost the same extent, whereas Thr-205 was not phosphorylated. For the AT180 site, Thr-231 and Ser-235 were phosphorylated in marmoset brains, although to a much lesser extent than P5 mouse brain tau, but only a trace reaction was detected from pSer-235 in human brains. At the PHF-1 site, Ser-404 was phosphorylated in both marmoset and human brains with stronger reactions found in human tau, a lesser degree of phosphorylation at Ser-396 was found in marmoset brains, and no phosphorylation was observed in human brains. For quantification, we measured the intensity of the area covering total tau detected with Tau5 (a bracket in the top panel of Fig. 6B), as banding patterns observed in the marmosets and humans differed while also differing slightly among the antibodies used. Ser-202 and Ser-404 were highly phosphorylated at ϳ60 -70% and 45-55% of mouse tau at P5, respectively. Whereas marmoset tau was phosphorylated more than human tau at most sites, Ser-404 alone was more phosphorylated in the human brains than in the marmoset brains. These results describe, for the first time, the biochemical phosphorylation states of tau in marmoset brains.

Discussion
The marmoset is expected to be used as a nonhuman primate model of human neurodegenerative diseases (20,27). Tauopathies are neurodegenerative diseases found in the brain where hyperphosphorylated tau accumulates (2)(3)(4). To use marmosets as a model of tauopathies, it is crucial to determine the nature of marmoset tau, such as its isoform expression and phosphorylation. However, there has thus far been no bio-

The isoforms and phosphorylation of marmoset tau
chemical analysis of the tau protein found in marmoset brains. Here, we cloned marmoset tau cDNA and investigated the expression of tau isoforms and its phosphorylation. Unexpectedly, we found that marmoset tau did not contain the so-called "primate-unique motif" in the N-terminal region, and the isoform expressed in adult brains was the 4R isoform as found in mouse brains. Adult marmoset tau did not react or reacted very weakly with several anti-pathological phospho-tau antibodies, although brain tissues were removed shortly after sacrifice. Thus, we found that marmoset tau resembles mouse tau rather than human tau in several biochemical aspects.
It was recently proposed that the N-terminal 11 amino acids, which are found in human tau but not rodent tau, reveal a motif unique to primates (35). However, marmoset tau did not contain 10 of the primate-unique 11 amino acids, although the marmoset is a primate. In comparing the N-terminal amino acids of a larger group of primates, including both Old World and New World monkeys and several prosimians, we found that all New World monkeys do not include 10 of the 11 amino acids of the "primate-unique motif." In contrast, prosimians (sifaka and lemur), which separated from anthropoids before the separation of Old and New World monkeys, did contain the motif, suggesting that the deletion took place in New World monkeys after they separated from Old World monkeys. Further, Stefanoska et al. (35) note that dog tau is an exception, as it includes this N-terminal motif. However, we found that cat and panda tau also have this sequence, although the motif extends to 14 amino acids in length with an additional 3 amino acids at the N-terminal upstream. Interestingly, tau of the Tasmanian devil, a marsupial animal, contains an additional 15 amino acids with an extra 8 amino acids in the upstream region ( Fig. 2A). From the phylogenic tree of mammalian tau (Fig. 2B), it may be concluded that ancestral mammalian tau contains a certain portion of the amino acid sequence at the N-terminal region, including the origin of the primate-unique motif. Some groups of mammals lost the sequence after their divergence into their respective groups (Fig. 2B), such that New World monkeys, rodents, and ungulates might have independently lost the 10, 11, and 12 amino acids. The variability observed in the sequence suggests that this N-terminal region is highly

The isoforms and phosphorylation of marmoset tau
changeable in the tau molecule. A question that thus arises concerns whether such variability represents species-specific functions of the N-terminal region in the tau molecule or not. The isoform expression pattern of tau in marmosets was found to be similar to that of mice rather than humans. It is well-documented that both 3R and 4R isoforms are expressed in adult human brains (21,22). Other primates, such as chimpanzees and gibbons, also express both 3R and 4R in the adult stage (41). In contrast, 4R tau is exclusively expressed in adult mouse and rat brains (20,24,38). Although the marmoset is a primate, only 4R tau is expressed in its adult brain. 3R or 4R tau is generated through the alternative splicing of exon 10. The splicing of exon 10 is regulated by the nucleotide sequence at the 5Ј splicing site of exon 10 and in a nucleotide at Ex10ϩ13 in the stem-loop (42,43). Because marmoset tau includes the same nucleotide sequence in its stem-loop as that found in humans (Fig. 7) (42, 43), the present result was surprising. However, we identified a difference in the nucleotide sequence of the intron-splicing modulator (ISM) region (Fig. 7, blue in a box). The ISM counteracts the intron-splicing silencer (ISS)-mediated inhibition of the 5Ј splicing site (43,44). The ISM of the marmoset may lose its inhibitory activity, counteracting the inhibitory activity of the ISS. It may be interesting to examine the effects of splicing factor FUS or SFPQ on the splicing of exon 10 of marmoset tau (45). The regulatory mechanism of the splicing of exon 10 is important, as indicated by the fact that the ratio of 4R to 3R increases in AD, whereas the forced decrease in the ratio of 4R to 3R by antisense oligonucleotide ameliorates the occurrence of epilepsy (46).
Marmoset tau differed from that of humans and mice in terms of the N-terminal isoform expression. Splicing in or out of exons 2 and 3 generates 0N, 1N, and 2N isoforms (21,22). The 1N isoforms of both 3R and 4R tau are most abundantly expressed in adult human brains (9,47), whereas the 0N isoform is highly expressed in adult mouse brains with a lesser and equal expression of 1N and 2N (20). In contrast, the 0N and, to a lesser extent, 2N isoforms are expressed in adult marmoset brains, whereas the 1N isoform is rarely detected. Although the functions of N-terminal insertions are yet not known, differences in the localization and interacting proteins of tau isoforms have been reported (48,49), and the 2N isoform interacts with proteins related to neurodegenerative diseases (49). It may be interesting to observe 2N isoform expression in marmoset brain regions to know their possible function or relationship to the diseases.
Phosphorylation is another important property of tau to be considered. Immunohistochemistry has been employed to identify pathological tau phosphorylation in marmoset brains. Two reports on this issue have been published; one report shows that tau is not phosphorylated even in old age using the PHF-1 antibody (50), whereas another recent paper claims that the pathological phosphorylation of tau at Thr-231 commences in the adolescent stage and progressively increases with age (51). Thus, the study of marmoset tau phosphorylation is only at the beginning stage. In this study, we used both antibodies and several other anti-phosphoantibodies in immunoblots to detect phosphorylation of tau. In particular, using the immunoblotting method, we could identify the isoforms of tau expressed in marmoset brains. Further, we used the perinatal mouse tau as a positive control of the phosphorylation to make the analysis more quantitative. We found that both PHF1 and Thr-231 sites were weakly phosphorylated in adult marmoset brains. The results suggest that the previous difference in the immunocytochemical staining could be caused by difference in the sensitivity of the detection.
We first tested whether marmoset tau can be phosphorylated similarly to human tau under the same cellular conditions by expressing these tau proteins in Neuro-2a cells. We used Phostag phosphoaffinity SDS-PAGE to observe the overall phosphorylation profile of tau (39). Marmoset tau presented almost the same banding pattern as that of human and mouse tau, indicating that marmoset and human tau are phosphorylated at the same site and to the same extent in cultured cells. This result constitutes important basic information when we consider a particular animal as a model of human tauopathies, but in a strict sense, such information has not been available previously. The present data obtained by Phos-tag indicate that marmoset tau can be substituted for human tau in the analysis of phosphorylation.
There are at least two critical issues to overcome in the study of the in vivo phosphorylation of tau in primates. One is dephosphorylation during the PMI. Wang et al. (16) reported that the dephosphorylation of tau proceeds quickly during the PMI. As it takes more time to dissect the brains of larger animals after their sacrifice, phosphorylation studies of primate brains are challenging. We perfused marmosets with PBS immediately after sacrifice, followed by the dissection of brains after ϳ10 min of perfusion. It took ϳ10 -15 min to completely remove various brain regions. However, this time period was considerably shorter than the 3-4-h PMI required for the human brains used to draw comparisons, suggesting that the present data more closely reflect the phosphorylation of tau in primate brains than those reported. Nevertheless, the phosphorylation of adult marmoset tau was low. Interesting sites are Ser-202 and . . .

Marmoset Human
Exon 10 Stem-loop-stem Ex10+13 ISM ISS Figure 7. Nucleotide sequence comparison of human and marmoset tau genes around the exon 10 -splicing site. The regions of the stem-loop-stem, ISS (intron-splicing silencer), and ISM (intron-splicing modulator) according to Qian and Liu (43) are indicated. Ex10ϩ13 (red) is a site that regulates Exon 10 splicing, which is replaced with guanine in rodent tau. The nucleotide sequence (blue) indicated by the box differs between marmoset and human tau genes.

The isoforms and phosphorylation of marmoset tau
Ser-404 that were still strongly phosphorylated, although the levels were slightly less than those observed in P5 mouse brains, which were prepared within 1 min following cervical spinal cord dislocation (20). The results suggest resistance of these sites to dephosphorylation during the PMI. This is consistent with our previous observations of slow dephosphorylation at these sites (52,53) and with other findings that Ser-404 was phosphorylated in 50 -60% of tau molecules in adult human brains (14,54). Tau is physiologically phosphorylated at Ser-202 and Ser-404. Additional phosphorylation at Thr-205 and Ser-396 creates the pathological sites detected by AT8 (Ser-202 and Thr-205) and PHF-1 (Ser-396 and Ser-404) observed in AD. It may be important to clarify the molecular mechanism that induces Thr-205 and/or Ser-396 phosphorylation to understand the mechanisms of abnormal phosphorylation.
Another is the effect of anesthesia, which is an essential treatment of animals prior to the sacrifice. The anesthesia treatment is demonstrated to increase the phosphorylation of tau through hypothermia of animals (18). We used ketamine and pentobarbital to anesthetize marmosets, but the phosphorylation states of marmoset tau thus prepared were relatively low and rather similar to human tau in autopsy samples with 3-4-h PMI. There is a report that investigated tau phosphorylation using human biopsy samples (17), in which tau is highly phosphorylated. Although these data have been thought to represent the phosphorylation states of tau close to that in living brains, there is a possibility that tau in those biopsy samples was phosphorylated highly during the anesthesia at the time of brain surgery. We think our marmoset brain preparation is similar to biopsy of human brains, although marmosets were sacrificed shortly before the brain dissection. Nonetheless, the phosphorylation states were different between them; marmoset tau was low, and biopsy human tau was high. Marmoset tau, which was once highly phosphorylated during anesthesia, might have been dephosphorylated during the dissection process. By minimizing the effects of PMI and anesthesia in the preparation of brains from marmosets, however, we would be able to approach the in vivo phosphorylation states of tau in primate brains.
We showed here for the first time that the expression of tau isoforms in marmoset brains differs from that in humans brains but rather resembles that in mouse brains. These results may raise a question about the use of the marmoset as a model of human tauopathies. However, we still think that marmosets are a better model than mouse. There are a number of differences between mouse and marmoset brains. For example, the overall structure of the brain differs between these two species; whereas the striatum of marmoset is separated into caudate nucleus and putamen, these structures are undistinguishable in the mouse (25). The expression patterns of the genes that regulate brain development are different between the mouse and the marmoset, especially in the areas of the brain that have connections to the prefrontal cortex that are presumably involved in higher cognitive functions (55). Further, the long life span of the marmoset, about 16 -18 years, compared with ϳ2 years in rodents, is a big advantage in the study of age-dependent diseases. On the other hand, there would be some limitations, particularly upon using marmosets as a model of tauopathies where both 3R and 4R tau aggregates are involved. Thus, our present results provide useful information on tau proteins in marmosets, which will be used as a model of tauopathy neurodegeneration in the future.

Brain samples
Common marmoset (C. jacchus) maintenance and the tissue sample collection were performed at Keio University and the RIKEN Center for Brain Science. Marmoset experiments conducted for this project were approved by the Institutional Animal Care and Use Committees of the RIKEN (approval number H28-2-219(5)), the Keio University Institutional Animal Care and Use Committee (approval number 11006-(3)), and the Research Ethics Committee of Tokyo Metropolitan University (approval number A28-11, A29-21). Animal handling and experimental procedures were performed in accordance with RIKEN's institutional guidelines for the use of laboratory animals, which correspond with Guidelines for Proper Conduct of Animal Experiments of the Science Council of Japan (2006). Animal care was conducted in accordance with recommendations of the Guide for the Care and Use of Laboratory Animals (Institute of Laboratory Animal Resources, 1996). In this study, six common marmosets were used: three newborn marmosets at P0, two adult females of 1 year and 11 months and of 5 years and 7 months, and one adult male of 4 years and 8 months.
For brain tissue preparation, marmosets were anesthetized with ketamine (10 mg/kg) and pentobarbital (80 mg/kg) by intramuscular injection at room temperature around 25°C. Within 5 min, all marmosets were sacrificed by the incision of the diaphragm, and then after 30 s, they were perfused with PBS at room temperature. After 10 min of perfusion, various brain regions were removed in 25-35 min, immediately frozen in liquid nitrogen, and stored at Ϫ80°C until use. Human brain tissues were obtained from the brain bank of the Tokyo Metropolitan Institute of Gerontology, and the study 5 N. Sahara, manuscript in preparation.

The isoforms and phosphorylation of marmoset tau
was approved by the ethics committees of the Tokyo Metropolitan Institute of Gerontology (approval number Q-No.4 (5/27/ 2016)) and of Tokyo Metropolitan University (approval number, H28-26, H29-66). We declare that the present studies abide by the Declaration of Helsinki principles. Two individuals were used; a woman who died at the age of 84 years and a man who died at the age 80 years at PMIs of 4.36 and 3.2 h, respectively. The mouse experiments were approved by the Research Ethics Committee of Tokyo Metropolitan University (approval number A28-10, A29-20). C57BL/6J mice were obtained through Sankyo Lab Service (Tokyo, Japan). Brains were removed immediately or within 1 min after sacrifice by cervical dislocation.

Isolation of tau cDNA from marmoset brains
Total RNA was isolated from marmoset brain cortexes on P0 or after 23 months using ISOGEN (Nippon Gene, catalog no. 315-02504) according to the manufacturer's instructions. cDNA libraries were generated by the reverse transcription of mRNA using the Superscript III First-Strand synthesis system (Invitrogen, catalog no. 18080-051) with oligo(dT) as a primer. Marmoset tau cDNA was amplified by PCR while using the cDNA library as a template. Primers included 5Ј and 3Ј ends of marmoset tau DNA sequences tagged with a restriction enzyme site, an EcoRI site at the 5Ј end of the forward primer, and an XbaI site at the 3Ј end of the reverse primer; 5Ј-GAGCACAG-AATTCATGGCTGAGCCCCGC-3Ј and 5Ј-CGCTTATCTA-GATCACAAACCCTGCTTGGCC-3Ј were used as forward and reverse primers, respectively.
PCR was performed using PfuUltra High-Fidelity DNA polymerase (Agilent Technologies, Santa Clara, CA). PCR products were inserted into pcDNA3.1A-Myc-His (ϩ) after digestion at EcoRI and XbaI sites. DNA sequences were analyzed using an ABI 3130 sequencer (Applied Biosystems, Foster City, CA).
RT-PCR was performed using GoTaq Green Master mix (Promega, Madison, WI). The expression of exons 2 and 3 or exon 10 was determined by the size of PCR products on electrophoresis using cloned marmoset cDNA as markers.

SDS-PAGE and immunoblotting
SDS-PAGE was performed with 12.5% polyacrylamide gels. Phos-tag SDS-PAGE was performed with 10% polyacrylamide gels containing 50 M Phos-tag and 100 M MnCl 2 as described previously (39). Immunoblotting was performed as described previously (20).

Bioinformatics
Amino acid sequences of tau were deduced from the protein-coding nucleotide sequence of the mammalian species in the Ensemble genome browser registered as MAPT-210 for humans; MAPT-201s for the chimpanzee, gorilla, orangutan, squirrel monkey, capuchin, night monkey, sifaka, mouse lemur, dog, cat, panda, mouse, and Tasmanian devil; ENSMMUT00000005853 for the macaque; ENSPANT-00000056619 for the olive baboon; MAPT-202s for the gibbon and goat; 204 for the rat; and 205 for the cow. The alignment of the N-terminal 50 amino acid sequences of tau was performed by MEGA7 (57) with ClustalW using the following parameters: pairwise alignment, gap opening penalty ϭ 5 and gap extension penalty ϭ 0.1; multiple alignment, gap opening penalty ϭ 5 and gap extension penalty ϭ 0.1 (57). A phylogenetic tree was constructed in MEGA7 using the neighbor-joining method for the N-terminal 120-amino acid sequence of tau (58). The evolutionary distances were computed using the Poisson correction method (59) and are shown as the number of amino acid substitutions per site. The Poisson correction distance assumes the equality of substitution rates among sites and equal amino acid frequencies while correcting for multiple substitutions at the same site. The rate of variation among sites was modeled with a ␥ distribution (shape parameter ϭ 1). The optimal tree with the branch length sum ϭ 1.13109959 is shown. The percentage of replicate trees in which the associated taxa clustered together during the bootstrap test (1000 replicates) is shown next to the branches (60).