Human TrkB gene: novel alternative transcripts, protein isoforms and expression pattern in the prefrontal cerebral cortex during postnatal development

Authors

  • Kristi Luberg,

    1. Department of Gene Technology, Tallinn University of Technology, Tallinn, Estonia
    2. Competence Center for Cancer Research, Tallinn, Estonia
    Search for more papers by this author
  • Jenny Wong,

    1. Schizophrenia Research Institute, Sydney, Australia
    2. Schizophrenia Research Laboratory, Prince of Wales Medical Research Institute, Randwick, New South Wales, Australia
    3. School of Medical Sciences, Faculty of Medicine, University of New South Wales, Sydney, New South Wales, Australia
    Search for more papers by this author
  • Cynthia Shannon Weickert,

    1. Schizophrenia Research Institute, Sydney, Australia
    2. Schizophrenia Research Laboratory, Prince of Wales Medical Research Institute, Randwick, New South Wales, Australia
    3. School of Psychiatry, Faculty of Medicine, University of New South Wales, Sydney, New South Wales, Australia
    Search for more papers by this author
  • Tõnis Timmusk

    1. Department of Gene Technology, Tallinn University of Technology, Tallinn, Estonia
    2. Competence Center for Cancer Research, Tallinn, Estonia
    Search for more papers by this author

Address correspondence and reprint requests to Tõnis Timmusk, Department of Gene Technology, Tallinn University of Technology, Akadeemia tee 15, 12618 Tallinn, Estonia. E-mail: tonis.timmusk@ttu.ee

Abstract

J. Neurochem. (2010) 113, 952–964.

Abstract

Brain-derived neurotrophic factor and neurotrophin-4 high-affinity receptor tropomyosine related kinase (Trk) B is required for the differentiation and maintenance of specific neuron populations. Misregulation of TrkB has been reported in many human diseases, including cancer, obesity and neurological and psychiatric disorders. Alternative splicing that generates receptor isoforms with different functional properties also regulates TrkB function. Here, we describe numerous novel isoforms of TrkB proteins, including isoforms generated by alternative splicing of cassette exons in the regions encoding both the extracellular and intracellular domain and also N-terminally truncated isoforms encoded by novel 5′ exon-containing transcripts. We also characterize the intracellular localization and phosphorylation potential of novel TrkB isoforms and find that these proteins have unique properties. In addition, we describe the expression profiles of all the known human TrkB transcripts in adult tissues and also during postnatal development in the human prefrontal cortex. We show that transcripts encoding the full-length TrkB receptor and the C-terminally truncated TrkB-T1 have different expression profiles as compared to the proteins they encode. Identification of 36 potential TrkB protein isoforms suggests high complexity in the synthesis, regulation and function of this important neurotrophin receptor emphasizing the need for further study of these novel TrkB variants.

Abbreviations used:
DLPFC

dorsolateral prefrontal cortex

EST

expressed sequence tag

IG-like

immunoglobulin-like

IRES

internal ribosome entry site

NT

neurotrophin

PLC-γ

phospholipase C-γ

Trk

tropomyosin related kinase

UTR

untranslated region

Tropomyosin related kinase B (TrkB; official name – NTRK2) is a receptor for neurotrophins brain-derived neurotrophic factor, neurotrophin (NT) 4 and, in a lower affinity, NT-3 (Squinto et al. 1991; Klein et al. 1992). TrkB is closely related to TrkA and TrkC, which are receptors for neurotrophins nerve-growth factor and NT-3, respectively (Lewin and Barde 1996). TrkB is an essential modulator of neural differentiation and cell survival. Dysregulation of TrkB has been associated with various neurological diseases, diverse type of cancers, obesity and eating disorders (Desmet and Peeper 2006; Farooqi and O’Rahilly 2006; Altar et al. 2009).

Interaction with two TrkB receptor molecules by one neurotrophin homodimer triggers the intrinsic tyrosine kinase activity of the TrkB intracellular portion leading to autophosphorylation at multiple tyrosine residues in the cytoplasmic domain of the receptor. These phosphotyrosines serve as docking sites for proteins such as Shc and phospholipase C-γ (PLC-γ), which are activated by tyrosine phosphorylation and link TrkB to downstream signaling pathways (Reichardt 2006). The signals transmitted by activated TrkB promote survival, neurogenesis and synaptogenesis in neurons (Binder and Scharfman 2004), and cellular proliferation, invasivity and resistance to anoikis and chemotherapy in cancer cells accompanied by poor prognosis for cancer patients (Nakagawara et al. 1994; Eggert et al. 2001; Douma et al. 2004; Yu et al. 2008; Au et al. 2009; Li et al. 2009b). Brain-derived neurotrophic factor-activated TrkB is a mediator of many other functions, including activity-dependent synaptic plasticity (Kang et al. 1997; Yamada and Nabeshima 2003) and angiogenesis (Kermani et al. 2005; Kermani and Hempstead 2007). In addition, changes in TrkB signaling accompany and can lead to various nervous system disorders, including mood and anxiety disorders and neurodegenerative diseases, which make TrkB an important target for drug development (Allen et al. 1999; Pillai 2008; Rantamaki and Castren 2008; Altar et al. 2009).

The TrkB gene is relatively large (Fig. 1), spanning more than 350 kbp, and is located on chromosome 9 (Nakagawara et al. 1995). A thorough examination of the TrkB gene and its transcripts conducted by Stoilov and coworkers revealed the existence of 24 exons in the TrkB gene with the first five exons serving as alternative transcription start-sites and displaying intricate patterns of splicing (Stoilov et al. 2002). It has been shown that these five exons form an internal ribosome entry site (IRES) guiding the ribosome to the translational start site which is located in exon 5 (Dobson et al. 2005). Exons 5–14 encode the extracellular portion of the TrkB receptor which contains a signal sequence for membrane localization, post-translationally glycosylated cysteine and leucine rich regions, and two immunoglobulin-like (IG-like) domains (Schneider and Schweiger 1991; Shelton et al. 1995). Exon 12 encodes the second IG-like domain that has been postulated to be the region responsible for binding neurotrophins (Urfer et al. 1995). The transmembrane domain of TrkB is encoded by exon 15 and the intracellular tyrosine kinase domain is encoded by exons 20–24 (Middlemas et al. 1991).

Figure 1.

 Structure of the human TrkB gene and predicted TrkB protein isoforms. Exons are shown as boxes and introns are shown as lines. Exons and introns are drawn to scale (in a and b). (a) Schematic representation of all TrkB exons (numbers are shown above exons) and introns. (b) Alternative 5′ exons of TrkB transcripts; (c) TrkB transcripts grouped by the type of protein isoforms they encode. Names of alternative protein isoforms are shown on the left. 5′ and 3′ UTR regions are shown as empty boxes, sequences encoding protein are shown as filled boxes and numbered. Shown are locations of tyrosine residues important for Shc-binding (Shc) and PLC-γ binding (PLC-γ). ATG, translation initiation codon; *Translation stop codon. (d) Schematic representation of TrkB protein domains. C, cysteine rich region; Leu rich, leucine rich region; IG like, immunoglobulin like-domain. (e) Amino acid sequences of the C-termini of TrkB, TrkB-T-TK and TrkB-Δ22 protein isoforms. Isoform-specific unique amino acids are shown in bold.

Additional to the full-length TrkB receptor, the human TrkB gene is known to encode C-terminal truncated receptors TrkB-T1 and TrkB-T-Shc, which are generated by the usage of exon 16 or exon 19, respectively. Exons 16 and 19 contain alternative polyadenylation signals and translational stop-codons (Klein et al. 1990; Stoilov et al. 2002). Both TrkB-T1 and TrkB-T-Shc may act as dominant negative inhibitors of the full-length receptor by preventing ligand-induced phosphorylation (Brodeur et al. 2009). In addition, it has been suggested that TrkB-T1 may have signaling properties that are different from the signal pathways activated by the full-length TrkB, such as evoking calcium signaling and mediating Rho GDP dissociation inhibitor 1 functions (Rose et al. 2003; Ohira et al. 2005). It has been found that TrkB-T1 can induce liver metastasis of pancreatic cancer cells by promoting RhoA activation (Li et al. 2009a). Interestingly, a recent study showed a decrease in the expression level of TrkB-T1 in the frontal region of the brain in 10 of 28 suicide completers (Ernst et al. 2009).

The intracellular juxtamembrane region-encoding exon 17 of TrkB has been shown to be a cassette exon (Stoilov et al. 2002). An isoform of TrkB which lacks the extracellular juxtamembrane region encoded by exon 13 has been shown to exist in the human retinal pigmented epithelial cells (Hackett et al. 1998). Additional splice variants have been described in mouse (C-terminal truncated isoforms and isoforms lacking leucine-rich regions), rat (C-terminal truncated isoforms) and chicken (isoforms with deletions in the extracellular and intracellular juxtamembrane regions; Middlemas et al. 1991; Garner et al. 1996; Strohmaier et al. 1996; Ninkina et al. 1997; Kumanogoh et al. 2008).

Temporal fluctuations in the expression level of mRNAs encoding the full-length TrkB and truncated TrkB-T1 receptors have been described in the dorsolateral prefrontal cortex (DLPFC), temporal cortex and hippocampus using in situ hybridization (Romanczyk et al. 2002; Webster et al. 2006). The DLPFC is important for higher level cognitive processing and working memory in humans. The maturation process of the DLPFC is protracted extending up to the first two decades of life making the DLPFC a good system for characterizing developmentally-regulated events (Bourgeois 1997). In the DLPFC, the expression levels of the full-length TrkB receptor-encoding mRNAs were highest in the young adult age group and lowest in the aged, whereas the level of the mRNAs encoding the TrkB-T1 isoform showed only minor increases over the postnatal life span.

As the TrkB gene has many functions in different tissues during development and in adulthood, both in health and disease, detailed knowledge of its structure, alternatively spliced mRNAs and protein isoforms is very important. In this study we have re-examined the human TrkB gene structure, identified novel isoforms of human TrkB transcripts that encode novel proteins and characterized tissue-specific expression patterns of alternative TrkB mRNAs. In addition, we have characterized age-related expression patterns of alternative TrkB mRNAs in the human DLPFC.

Materials and methods

Computer analysis and RT-PCR were performed as described previously (Koppel et al. 2009). Primers used for the analysis of TrkB mRNA expression and for cloning of TrkB riboprobes and full-length TrkB transcripts for protein expression are shown in Table S1. Quantitative PCR and western blotting were performed as described in Wong et al. (2009); and RNAse protection assay as detailed in Timmusk et al. (1993).

HEK293 cells were grown in Dulbecco’s Modified Eagle’s Medium (Invitrogen, Carlsbad, CA, USA) containing 10% fetal bovine serum and transfected with PEI reagent (InBio, Tallinn, Estonia). For immunofluorescence, the cells were grown on cover slips, fixed with 4% paraformaldehyde (Scharlau, Barcelona, Spain), blocked with 2% bovine serum albumine (Sigma, St Louis, MO, USA) and stained with Alexa Fluor 488-conjugated concanavalin A (Invitrogen), primary anti-V5 antibody (Sigma, V8137) and secondary Alexa Fluor 568-conjugated goat anti-rabbit antibody (Invitrogen). Cover slips were mounted using ProLong Gold antifade reagent with 4′,6-diamidino-2-phenylindole (DAPI; Invitrogen). Labeled cells were analyzed using the LSM 510 (Zeiss, Oberkochen, Germany) confocal microscopy system.

For immunoprecipitation, cells were lysed with radioimmunoprecipitation buffer. 0.5 mg of a 1 mg/mL total protein extract was immunoprecipitated with 1.8 μg mouse anti-V5 antibody (Invitrogen) or anti-TrkB antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA, SC-8316). Precipitated proteins were analysed with western blotting using mouse anti-phosphotyrosine (Millipore Corporation, Bedford, MA, USA, 4G10) or mouse anti-V5 antibody.

Detailed information of materials and methods can be found in Appendix S1.

Results

Structure and alternative transcripts of the human TrkB gene

The human TrkB gene structure and expression profiles of its alternative transcripts were analyzed by bioinformatics and RT-PCR. For this purpose, we first used the Blat application to align human TrkB mRNAs and expressed sequence tags (ESTs) from GenBank to the genome in order to determine the exon/intron sites of transcripts that have been entered into the public databases. Expression of alternative splice variants in different tissues was verified by RT-PCR and sequencing of PCR products. The gene structure arising from this analysis is depicted on Fig. 1. In most part, this study uses the nomenclature described by Stoilov et al. (2002). In addition, we identified the existence of a novel exon, exon 5c, which is located approximately 1200 bp downstream of exon 5. Exon 5c comprises a new transcription start site. Finally, we identified novel splice variants of the human TrkB transcripts that are described in detail below.

Exons 1, 2, 3, 4 and the majority of exon 5 constitute the conventional 5′ untranslated region (UTR) of the human TrkB gene. All of these exons can serve as transcription start sites (Stoilov et al. 2002). In GenBank, there are more than 200 ESTs covering the region corresponding to the TrkB gene. A majority of these were identified in a study characterizing alternative promoters of human genes (Kimura et al. 2006). According to Stoilov and coworkers and the alignment of the TrkB ESTs, the alternative splicing pattern of TrkB exons 1–5 is rather complex, arising from exon skipping (exons 2 and 3) and the usage of alternative 5′ and 3′ exon splice sites for exons 2, 3 and 4 (Fig. 1b). It has been shown that the G/C rich exons 1–5 exhibit internal initiation entry site (IRES) activity that functions to guide and enhance the translation initiation from exon 5 (Dobson et al. 2005). The IRES was localized to the exon 5 and at least six sub-regions were shown to either promote or inhibit the IRES activity. Therefore, we hypothesised that the intricate splicing pattern of the TrkB 5′ UTR might be tissue-specific with variable translation initiation regulation capabilities in different tissues. However, RT-PCR experiments did not support this hypothesis, as all the tissues studied showed highly similar patterns of expression and splicing of exons 1–5 (Fig. S1a). Thus, the functional meaning of this intricate splicing still remains to be determined.

To date, all TrkB isoforms have been considered to include exon 5 which also contains the start codon for protein translation. Here, we describe a novel exon that we named 5c, which is located approximately 1200 bp downstream of exon 5 (Fig. 1a–c). None of the ESTs or the mRNA in GenBank containing exon 5c includes exons 1–5. Therefore, we hypothesized that exon 5c serves as an alternative transcription start-site for TrkB mRNAs. This assumption was supported by RT-PCR which did not render any products with primers specific for TrkB exons 1 and 5c, 2 and 5c, 3 and 5c, 4 and 5c or 5 and 5c (data not shown). RT-PCR analysis of TrkB transcripts containing exon 5c showed that these mRNAs were expressed mainly in the nervous system with the highest expression in the cerebellum (Fig. 2). The novel exon 5c is not specific to humans as we detected TrkB transcripts containing exon 5c in the mouse brain (Fig. S1b and c). In silico analysis showed that protein translation of human TrkB transcripts containing exon 5c are likely to start from exon 9 (Fig. 1c). This creates N-terminally truncated TrkB proteins that lack 156 N-terminal amino acid residues encoding the signal sequence, leucine-rich repeats and most of the cysteine-rich repeats compared to the full-length protein. We have named the novel protein isoforms TrkB-N, TrkB-N-T-TK, TrkB-N-T-Shc and TrkB-N-T1 according to the C-termini of these proteins that are encoded by exons 24, 22b, 19 and 16, respectively. Because of the lack of N-terminal signal sequence, it is highly unlikely that these protein isoforms could be transported to the cell membrane, although they contain the region that functions as a transmembrane domain in the full-length TrkB receptor.

Figure 2.

 Semiquantitative analysis of expression levels of alternative TrkB transcripts in different human tissues. Numbers of amplified TrkB exons are shown on the right. Hypoxanthine guanine phosphoribosyltransferase (HPRT) was used as the housekeeping control.

We observed exon skipping in the case of three exons outside the 5′ UTR: for exons 12, 17 and 22 (Figs 1c and 2). Skipping of exon 17 has been described before (Stoilov et al. 2002), however, the possible tissue-specificity and the function of the proteins (named as TrkB-Δ17, TrkB-T-TK-Δ17, TrkB-T-Shc-Δ17, TrkB-N-Δ17, TrkB-N-T-TK-Δ17 and TrkB-N-T-Shc-Δ17) encoded by these transcripts has not been studied so far. Interestingly, the splice donor site of exon 17 is present in the genome sequence of all mammals analyzed by us but is lost in mouse (Fig. S1d). Alternative splicing of human TrkB exon 12 and 22 is first described in this study. Skipping of exons 12 or 17 does not lead to a frame-shift in the encoded protein. Exon 12 encodes for 101 amino acids in the extracellular domain and exon 17 for 16 amino acids in the intracellular juxtamembrane domain of the TrkB protein. Skipping of exon 12 leads to production of protein isoforms without the neurotrophin-binding domain (TrkB-Δ12, TrkB-T-TK-Δ12, TrkB-T-Shc-Δ12, TrkB-T1-Δ12, TrkB-N-Δ12, TrkB-N-T-TK-Δ12, TrkB-N-T-Shc-Δ12 and TrkB-N-T1-Δ12; Fig. 1c), suggesting that these protein isoforms may be phosphorylated ligand-independently. Although skipping of exon 22 causes a frame-shift, the first translational stop-codon remains in exon 24, the most 3′ exon, and thus, these mRNAs will most probably not be subject to non-sense-mediated decay. The resulting protein (named TrkB-Δ22) is 139 amino acids shorter than the full length TrkB protein, lacks the PLC-γ binding site and part of the tyrosine kinase domain, and has 53 different amino acids in its C-terminus as compared to the full length TrkB protein (Fig. 1c–e). Sequence analysis showed that the unique C-terminus of mouse putative TrkB isoform TrkB-Δ22 has the same length as human TrkB-Δ22 and that the regions covering the 53 C-terminal amino acids of mouse and human TrkB-Δ22 are 76% homologous (data not shown). According to our RT-PCR studies, skipping of exon 17 was a relatively frequent event in all tissues studied, with approximately 50% of transcripts lacking exon 17. In contrast, skipping of exons 12 and 22 was a rare event. Skipping of exon 12 was observed in the heart and skeletal muscle, and at lower levels also in the brain, and skipping of exon 22 was noticed at extremely low levels in the frontal cerebral cortex and left cerebellum.

Additional diversity among TrkB protein isoforms is achieved by the use of alternative 3′ exons in the TrkB mRNAs. Our results showed that TrkB transcripts containing exon 16 were expressed in all tissues studied, and transcripts with exon 19 or 24 were expressed mainly in neural tissues (Fig. 2), as described earlier (Shelton et al. 1995; Stoilov et al. 2002). In addition, we identified a new transcript of the human TrkB gene that encodes a novel truncated isoform of the TrkB protein. This transcript includes an extended version of exon 22 as the 3′ exon, that we named 22b, which contains a translational stop-codon. Proteins encoded by these mRNAs (named TrkB-T-TK) lack the 114 C-terminal amino acids of the full-length TrkB encoding part of the tyrosine kinase and the PLC-γ binding site and have unique 27 amino acids in their C-termini (Fig. 1c and e). We found that TrkB mRNAs containing exon 22b were predominantly expressed in the nervous system (Fig. 2). The region covering the 27 C-terminal amino acids of the human TrkB-T-TK is not conserved among other mammals, as the mouse putative TrkB-T-TK isoform has only six unique C-terminal amino acids, with the first four amino acids being identical with the human homologue (data not shown).

Next, the relative quantities of alternative TrkB transcripts were studied. RT-PCR analysis showed that the frequency of exon 17 skipping was approximately 50%, whereas the skipping of exons 12 or 22 was a very rare event (less than 1%). For the determination of relative quantities of alternative 5′ and 3′ exons we used the RNAse protection assay.

Quantification of mRNAs containing exon 5c compared to mRNAs with the conventional TrkB 5′ UTR (containing exon 5; Fig. 3) showed that the levels of transcripts with exon 5c were much lower than transcripts with exon 5. In the cerebellum, 5.7% of all TrkB mRNAs started with exon 5c and 94.3% lacked exon 5c and thus started with the conventional 5′ UTR. In the frontal cerebral cortex, 0.2% TrkB transcripts included exon 5c and 99.8% used exon 5.

Figure 3.

 Relative quantification of the alternative 5′ and 3′ termini of human TrkB transcripts. RNA extracted from the human left cerebellum and frontal cerebral cortex was used for RNAse protection assay. (a) Exons contained in protected RNA fragments are shown on the right. (b) Relative quantities of TrkB transcripts with alternative 5′ and 3′ termini.

Comparison of transcripts with alternative 3′ exons revealed that TrkB mRNAs containing exons 24 and 16 were the most abundant. Approximately 60% of all the TrkB transcripts in both cerebellum and frontal cerebral cortex incorporated exon 24. In the frontal cerebral cortex, 32% of TrkB transcripts contained exon 16, and 8.2% contained exon 19. Extremely low levels of transcripts with the extended form of the exon 22 (22b) were detected using this method in the cerebral cortex. However, in the cerebellum, 11.7% of TrkB transcripts contained exon 22b. Of all the TrkB mRNAs in the cerebellum only 16.4% contained exon 16 and only 6.6% contained exon 19.

Taken together, in silico translation of novel TrkB mRNA alternative transcripts revealed the potential existence of many TrkB protein isoforms (Fig. 1c) that have not been described to date: protein isoforms with truncated N-terminal domain lacking sequences encoded by exons 5–8; protein isoforms lacking sequences encoded by exon 12, protein isoforms lacking sequences encoded by exon 22 and having unique C-termini encoded by exons 23 and 24 in an unconventional reading frame; protein isoforms with truncated C-terminal domain encoded by exon 22b. We were able to confirm the existence of all the alternative full-length transcripts of TrkB in the human brain, including the novel transcripts described in this study, by using RT-PCR analysis with primers targeting alternative translation initiation and stop codons (Fig. S1e). However, we did not detect previously described TrkB mRNA isoforms without exon 13 (Hackett et al. 1998), possibly because they are in low abundance or because tissues where these isoforms are expressed were not included in our study. Together, our present findings in accordance with reports previously published suggest that at least 36 different protein isoforms are encoded by the human TrkB gene (Fig. 1c).

Regulation of TrkB mRNA and protein expression in the human prefrontal cerebral cortex during postnatal development

Next, we determined the expression profiles of the major known and novel TrkB transcripts identified in this study during postnatal development of the human prefrontal cortex. We investigated the expression profiles of TrkB transcripts containing alternative 5′ exons 5 and 5c, alternative 3′ exons 16, 19, 22b and 24, as well as transcripts with and without exon 17. For this purpose, we isolated RNA from a set of 52 postmortem prefrontal cerebral cortices and performed RT-qPCR studies using primers specific for alternative TrkB transcripts. The subjects were divided into seven pre-defined groups based on their age. The data were normalized to the geometric mean of the expression levels of four control (housekeeping) genes including cyclophillin A, glucuronidase beta, ubiquitin C and porphobilinogen deaminase.

To determine the most significant changes occurring in postnatal brain development, we statistically analyzed age groups with the highest and lowest change in mRNA expression for all the alternative TrkB transcripts. Expression levels of TrkB mRNAs with exon 24, encoding the full length TrkB receptor containing the tyrosine kinase domain, did not change from neonates to infants (Fig. 4a). Interestingly, expression levels peaked in the toddler stage, decreased slightly during school age, remained relatively unchanged up to the young adult stage and decreased significantly in adulthood as compared to neonates (t = 2.60, df = 12, p = 0.02). In contrast, expression levels of TrkB mRNAs containing exon 16, encoding the TrkB-T1 receptor lacking the tyrosine kinase domain, were relatively similar between neonates and infants, but decreased in toddlers and in the school age group where it reached significance compared to neonates (t = 3.76, df = 9, p = 0.004; Fig. 4b). Expression levels of TrkB-T1 mRNA then increased to the level of neonates in the young adult and adult age groups. While TrkB mRNAs containing exon 19 encoding the TrkB-T-Shc protein isoform were low in the neonatal cortex, transcript levels increased by 5-fold during infancy but this did not reach statistical significance (t = −1.47, df = 14, p = 0.16). This is likely because of the variability in the infant group. Expression then decreased gradually during later postnatal development with a 40% reduction in adults (compared to infants; Fig. 4c).

Figure 4.

 Regulation of TrkB mRNA expression in the human prefrontal cerebral cortex during postnatal development. Expression of TrkB transcripts encoding: (a) full length TrkB (exon 24), (b) truncated TrkB-T1 (exon 16), (c) TrkB-T-Shc (exon 19), (d) juxtamebrane region [with and without exon 17 (Δ17)], (e) N-terminal region (exon 5 and exon 5c) and (f) C-terminal region or TrkB-T-TK (exon 22b). Data are expressed relative to the adult age group as ΔΔCT and presented as mean + SEM. *Significance p < 0.05; **p < 0.005; #Trends toward significance.

Next, we examined the expression levels of TrkB transcripts with and without exon 17. Interestingly, alternative splicing of exon 17 was developmentally regulated: expression of transcripts with and without exon 17 was lowest in neonates but differed thereafter (Fig. 4d). Expression of transcripts containing exon 17 increased gradually from neonates up to school age where the levels were significantly higher than in neonates (t = −2.27, df = 9, p = 0.049). Expression then decreased gradually during later postnatal development. In contrast, the levels of transcripts lacking exon 17 were not significantly changed across postnatal development. Although expression levels during infancy was increased 2-fold from neonates, this remained relatively stable into adulthood (Fig. 4d).

We next examined the changes in mRNA expression of TrkB transcripts containing exons 5 and 5c. Expression levels of TrkB mRNAs containing exon 5 were not significantly changed across postnatal development (Fig. 4e). Expression of TrkB mRNAs with exon 5c increased during postnatal development reaching their highest level in young adults as compared to neonates, although this did not reach statistical significance (t = −1.52, df = 10, p = 0.17). No further change was observed in adulthood (Fig. 4e).

We next determined the changes in TrkB transcripts containing exon 22b across postnatal development. The expression levels of transcripts containing exon 22b were variable. Transcript expression was lower in infants as compared to neonates, increased in toddlers and decreased again in school age (Fig. 4f). Thereafter, expression levels remained unchanged and decreased again in adulthood (neonate to adults: t = 1.89, df = 10, p = 0.09). However, changes in expression were not statistically significant.

To determine whether TrkB mRNA levels coincide with the levels of corresponding TrkB protein isoforms, lysates from the human prefrontal cerebral cortex were analysed using western blotting. Considering that the number of potential TrkB protein isoforms is as high as 36 (Fig. 1c) and that many TrkB protein isoforms have similar sizes, we chose the anti-TrkB antibody that does not recognize the minor N-terminally truncated TrkB-N isoforms to make the interpretation of results easier. The results (Fig. 5a) indicated that there are many different isoforms of TrkB proteins expressed, most of which cannot be identified using this method. The two most highly expressed protein isoforms are probably TrkB-T1 (molecular weight ∼ 90 kDa) and the full-length TrkB protein (molecular weight ∼ 140 kDa). We detected significant differences in the expression level of these isoforms in the human prefrontal cortex throughout the human lifespan (Fig. 5b and c). Namely, the expression level of full-length TrkB was significantly higher in infants as compared to neonates (t = −2.25, df = 18, p = 0.04) and decreasing in older age groups. The expression level of the truncated TrkB-T1 isoform, however, rose steadily from neonates to teenagers (t = −3.94, df = 13, p = 0.002) and decreased slightly in older age groups. We also quantified signals from an isoform of approximately 100 kDa corresponding most probably to TrkB-T-Shc. The expression level of this isoform did not show significant changes (Fig. S1f). Thus, our data suggest that the expression patterns of the full-length TrkB and TrkB-T1 protein isoforms differ from the expression patterns of the transcripts encoding these proteins during the postnatal development of human cerebral cortex.

Figure 5.

 Expression of TrkB protein isoforms in the human prefrontal cerebral cortex during postnatal development. (a) Representative western blot analysis of TrkB expression in the human prefrontal cerebral cortex during postnatal development. N, neonate; I, infant; T, toddler; SA, school age; YA, young adult; A, adult; (i) full-length TrkB; (ii) TrkB-T-Shc; (iii) TrkB-T1. (b and c) Quantification of the expression levels of the full-length TrkB (b) and TrkB-T1 (c) isoforms. Data are expressed relative to the adult age group as ΔΔCT and presented as mean + SEM. *Significance p < 0.05; **p < 0.005.

Intracellular localization and phosphorylation of novel protein isoforms of TrkB

Bioinformatic analysis did not predict a signal sequence for membrane localization in the novel N-terminal truncated isoforms of the TrkB receptor encoded by transcripts with exon 5c. Because of this, we were interested in the intracellular localization of these proteins. Considering that approximately 50% of TrkB transcripts in humans do not contain exon 17 (Fig. 2), and because exon 17 is not used in mouse TrkB transcripts because of the loss of the splice donor site of exon 17 (Fig. S1d), TrkB-N-Δ17 was chosen for the analysis as a representative of N-terminal truncated isoforms. In addition, the novel C-terminal truncated TrkB-T-TK-Δ17 isoform (encoded by the exon 22b) and the full-length TrkB and TrkB-Δ17 receptor isoforms with intact tyrosine kinase domains, were studied for their intracellular localization pattern. DNA sequences encoding these protein isoforms were cloned into expression vector encoding C-terminal V5-His tagged TrkB fusion proteins, transfected into HEK293 cells and intracellular localization of tagged proteins was studied. Our results (Fig. 6) showed that all tested TrkB isoforms localized to the cell membrane and to the cytoplasm, with the exception of TrkB-N-Δ17 which lacks a membrane-localization signal sequence and was found only in the cytoplasm.

Figure 6.

 Intracellular localization of different TrkB protein isoforms. Over-expression of TrkB isoforms in HEK293 cells. Red – TrkB; green – concanavalin A; blue – DAPI.

The TrkB-T-TK protein isoforms encoded by TrkB mRNAs with exon 22b lack a part of the intracellular tyrosine kinase (Fig. 1c). To study if these protein isoforms contain intrinsic tyrosine kinase activity, we expressed a representative of this type of human TrkB protein isoforms, TrkB-T-TK-Δ17, in HEK293 cells and compared the autophosphorylation levels of this isoform with the levels of autophosphorylation of TrkB, TrkB-Δ17 and TrkB-N-Δ17. Our results (Fig. 7a) showed that in contrast to all other TrkB protein isoforms tested, the TrkB-T-TK-Δ17 protein was not autophosphorylated.

Figure 7.

 Auto-phosphorylation and in trans phosphorylation potential of different TrkB protein isoforms. (a) HEK293 cells were transfected with expression constructs encoding V5-His tagged TrkB, TrkB-Δ17, TrkB-N-Δ17 and TrkB-T-TK-Δ17 proteins. The cells were lysed and subjected to immunoprecipitation with anti-V5 antibody 48 h post-transfection. Precipitated proteins were analysed with anti-phospho-tyrosine (anti-pY) and anti-V5 antibodies using SDS–PAGE and western blotting techniques. (b) HEK293 cells were co-transfected with V5-His tagged expression plasmids encoding TrkB, TrkB-Δ17, TrkB-N-Δ17 and E2 tagged TrkB. The cells were lysed and subjected to immunoprecipitation with anti-V5 or anti-TrkB antibodies 48 h post-transfection. Precipitated proteins were analysed with anti-phospho-tyrosine (anti-pY) and anti-V5 antibodies using SDS–PAGE and western blotting techniques.

We next studied whether the TrkB-T-TK-Δ17 isoform could be phosphorylated in trans by other TrkB isoforms that are enzymatically active. This aspect is especially interesting, as the TrkB-T-TK-Δ17 isoform contains the tyrosine residue used by the full-length receptor for Shc protein binding. Therefore, we co-expressed E2 tagged TrkB and V5-His tagged TrkB fusion proteins in HEK293 cells to determine if the E2-tagged full-length TrkB receptor is capable of phosphorylating the V5-His tagged TrkB-T-TK-Δ17 protein. We detected two different sub-populations of TrkB isoforms, probably because of different levels of post-translational modifications of the over-expressed proteins. TrkB-T-TK-Δ17-V5-His proteins had molecular weights of approximately 120 kDa and 100 kDa, whereas TrkB-E2 proteins had molecular weights of approximately 140 and 120 kDa. Although the TrkB-E2 protein co-immunoprecipitated with the V5-His-tagged TrkB isoforms and despite the fact that signals from TrkB-E2 partially overlapped with signals from TrkB-T-TK-Δ17-V5-His, we could detect a TrkB-T-TK-Δ17-V5-His-specific signal corresponding to the 100 kDa isoform of TrkB-T-TK-Δ17-V5-His proteins using the anti-phosphotyrosine antibody (Fig. 7b). This result indicates that similarly to other tested isoforms of TrkB, TrkB-T-TK-Δ17 can be phosphorylated in trans by the full-length kinase domain-containing TrkB-E2 isoform. This would suggest that the novel C-terminal truncated isoforms encoded by TrkB transcripts with exon 22b could act as functional signaling receptors that can be activated by the full-length TrkB which can then transfer the signal to Shc.

Discussion

In this study, we have re-examined the structure, alternative splicing pattern and spatio-temporal expression profile of the human TrkB gene. The 5′ UTR of the gene is known to be complex, however, we have shown an even more diversified structure of the region formed by the use of alternative promoters and alternative splicing generating at least 16 different 5′ UTRs. The 5′ UTR of the TrkB gene has previously been shown to possess an IRES that contains sub-regions of translation activation or inhibition (Dobson et al. 2005). Thus, the 5′ UTR is likely to regulate the expression of TrkB proteins at the level of translation. This would explain the rationale behind the existence of so many alternative 5′ UTRs. Because of this, we were interested to know whether there was a tissue-specific regulation of transcription initiation and splicing of the TrkB 5′ UTR. As our results indicated, all the tissues studied showed highly similar patterns of expression of TrkB exons 1–5, suggesting that the tissue-specific expression of TrkB receptors is not affected by the use of alternative 5′ UTRs. It is still possible, however, that tissue-specific factors such as 5′ UTR binding proteins could differentially affect the spatial regulation of translation efficiency in different tissues.

Interestingly, we have described a novel 5′ exon named 5c, which gives rise to novel N-terminal truncated TrkB proteins. All other TrkB 5′ UTRs are linked to transcripts encoding proteins with identical N-termini. The new exon 5c was expressed mainly in the nervous system with the highest levels observed in the cerebellum, where approximately 6% of all TrkB transcripts start with exon 5c. There are many different protein isoforms that are encoded by transcripts starting with exon 5c, depending on the 3′ exons and alternative splicing, but all of them lack the signal sequence for intra-membrane localization, the whole leucine-rich domain and one cysteine-rich domain. We studied in detail the isoform TrkB-N-Δ17 that encodes a tyrosine kinase domain, and showed that, as predicted, this isoform is not localized to the cell plasma membrane, but is instead cytosolic. In addition, TrkB-N-Δ17 becomes phosphorylated if over-expressed in HEK293 cells, which suggests a unique function for these proteins – activated TrkB-N-Δ17 isoforms can potentially phosphorylate other, yet unidentified proteins in the cytosol and hence, take part in signaling cascades. Because of spatial restrictions, TrkB-N-Δ17 isoforms are unlikely to be activated by neurotrophins – from translation to secretion, neurotrophins are kept stored in membraneous compartments, such as the endoplasmic reticulum (ER), Golgi complex and extracellular space, and thus, cannot come into contact with the cytosolic TrkB-N-Δ17 isoform. It is possible that the novel TrkB isoforms can be activated independently of neurotrophins by other cytosolic proteins. For example, it has been shown that TrkB receptor can be transactivated in the cytosol independently of neurotrophins by G-protein-coupled receptors, including adenosine 2A receptor, pituitary adenylate cyclase-activating polypeptide receptors and dopamine D1 receptor (Lee and Chao 2001; Lee et al. 2002; Rajagopal et al. 2004; Iwakura et al. 2008), or Src family kinases activated by zinc ions (Huang et al. 2008). The same proteins could be responsible for activating the novel N-terminally truncated TrkB isoforms. In addition, under certain conditions, the expression level of TrkB-N isoforms can rise and at higher intracellular concentrations the proteins can be autophosphorylated, as we have demonstrated by TrkB-N-Δ17 over-expression studies.

In this study, we have also characterized a novel 3′ exon of TrkB transcripts – an extended form of exon 22, named 22b. Similar to transcripts with exon 5c, exon 19 or exon 24, TrkB mRNAs with exon 22b present a highly specific spatial expression pattern – they are mainly expressed in neural tissues. Remarkably, transcripts with exon 22b account for approximately 10% of all TrkB-encoding transcripts in the cerebellum – an expression level that is higher than that of previously described TrkB-T-Shc isoform-encoding transcripts. No statistically significant age-related differences in the expression level of TrkB mRNAs with exon 22b could be detected in the DLPFC, however it is possible that these transcripts are temporally regulated in some other brain region, for example, in the cerebellum, where these mRNAs are expressed at a higher level, and thus, seem to have a more important function. A representative of the proteins encoded by transcripts with exon 22b, named TrkB-T-TK-Δ17, was localized to the cell plasma membrane when over-expressed and although it cannot get autophosphorylated because of a disruption in the tyrosine kinase domain, it can be phosphorylated by the full-length TrkB receptor. Phosphorylated TrkB-T-TK-Δ17 protein could function as an important signaling molecule – for example, it contains a docking site for Shc protein binding and can, thus, actuate signaling cascades leading to survival and differentiation of the cell. On the other hand, the TrkB-T-TK-Δ17 protein does not contain a docking site for PLC-γ and therefore, cannot activate the signaling cascades involving this signaling molecule. Hence, it can be concluded that TrkB-T-TK-Δ17 exerts different effects compared to the full length TrkB receptor. It is also possible, that the unique C-terminal sequence of 27 amino acids of TrkB-T-TK-Δ17 contains additional binding sites for yet unidentified proteins.

It is of interest to note that unlike the TrkB-T-TK-Δ17 isoform, another C-terminal truncated isoform of TrkB, named TrkB-T-Shc, which also contains a binding site for Shc protein and is expressed mainly in the nervous system, cannot be phosphorylated by the full-length receptor (Stoilov et al. 2002). Therefore, it can be concluded that these structurally similar truncated isoforms of TrkB receptor most probably have very distinct functions.

Exon 17 of the TrkB gene has been described to be a cassette exon (Stoilov et al. 2002), but thus far, no functional relevance has been connected to the alternative splicing of this exon. We have shown here that proteins with or without sequences encoded by exon 17 showed identical localization patterns and phosphorylation capabilities. However, the mRNAs encoding them displayed different expression patterns in the DLPFC – no statistically significant fluctuation was seen in the expression level of transcripts without exon 17. This is in contrast to transcripts with exon 17, which showed a relatively low expression level in neonates, peaking in school age, and after which the expression level started to descend gradually. This finding indicates that the presence of exon 17 is likely to have functional significance. It is possible, for example, that the presence or absence of protein sequences encoded by exon 17 is influencing the receptor’s ability to transfer signals to downstream signaling molecules.

In this study, we confirmed the spatial expression patterns of full-length TrkB and C-terminal truncated TrkB-T1-encoding transcripts described previously (Shelton et al. 1995; Stoilov et al. 2002) – nervous system specific versus ubiquitous expression pattern, respectively. In contrast, the temporal regulation of expression of these isoforms did not compare with previous data (Romanczyk et al. 2002). As was described previously using in situ hybridization in the DLPFC, the expression level of mRNAs encoding the full-length TrkB peaked in young adults while the levels of TrkB-T1 isoform-encoding transcripts did not show remarkable change throughout the lifespan (Romanczyk et al. 2002). On the other hand, using qPCR, we showed in this study that full-length TrkB-encoding transcripts are expressed at their highest level in the toddler age group but fall remarkably in adults. For TrkB-T1-encoding mRNAs, we found a relatively high expression level in the neonate, infant, young adult and adult age groups, and a lower expression level in toddlers and school aged children. This difference could be the result of either different quantification methods used or the result of analysing different cohorts. Interestingly, the expression patterns of the full-length TrkB and TrkB-T1 proteins differed from the expression patterns of the transcripts encoding these proteins during the postnatal development of human cerebral cortex. We have observed a similar phenomenon for TrkC transcripts and protein isoforms (Beltaifa et al. 2005). These results suggest that in addition to transcriptional control, post-transcriptional mechanisms are involved in the regulation of the expression of neurotrophin receptors.

We also detected minor quantities of TrkB mRNAs with the exclusion of exon 12 in heart and skeletal muscle. Proteins encoded by these mRNAs (TrkB-Δ12) lack the second IG-like domain, which is important for binding neurotrophins, and are similar to a variant of TrkA receptor (ΔTrkA) that has been found in acute myeloid leukemia cells (Reuther et al. 2000). ΔTrkA lacks part of the second IG-like domain and was shown to be constitutively active in promoting cell growth and resistance to apoptosis (Reuther et al. 2000). It would be of interest to determine whether the functional properties of TrkB-Δ12 and ΔTrkA are similar or whether the TrkB-Δ12 protein is activated by some ligand molecules.

Taken together, we have performed a detailed analysis of the human TrkB gene structure, expression profile of alternatively spliced transcripts, and protein isoforms encoded by alternatively spliced transcripts. Our findings emphasize the structural and functional variability of alternative TrkB receptor isoforms and suggest a diversified role of the receptor in the functioning of the human nervous system.

Acknowledgements

We thank Epp Väli for technical assistance, Heiti Paves for help in confocal microscopy and Enn Jõeste from North Estonian Regional Hospital, Tallinn, for collaboration. KL and TT are supported by Estonian Ministry of Education and Research (Grant 0140143), Estonian Enterprise and Baltic Technology Development Ltd. CSW is supported by the Schizophrenia Research Institute utilising infrastructure funding from NSW Health, the University of New South Wales School of Psychiatry, and the Prince of Wales Medical Research Institute. JW is supported by the National Health and Medical Research Council Postdoctoral Training Fellowship (568884).

Ancillary