Development of the brain involves a fine balance between excitatory and inhibitory signaling at synapses, ensuring plasticity while avoiding excitotoxicity. Kalirin, originally reported as a protein that interacts with Huntington associated protein (HAP), is known to be involved in active remodeling of synapses (Colomer et al., 1997). Kalirin has been implicated in a variety of neuropsychiatric and neurodegenerative diseases like schizophrenia, depression and Alzheimer’s disease (Hill et al., 2006; Mandela and Ma, 2012; Remmers et al., 2014; Makrythanasis et al., 2016). At the molecular level, this RhoGEF kinase is expressed highly at excitatory synapses and reduced at inhibitory synapses. Its binding to receptors leads to RhoGTPase mediated signaling and regulation of actin cytoskeletal organization. This intra-cellular signaling pathway leads to synaptic remodeling through altered dendrite numbers and dendritic morphology (Penzes et al., 2001). The mammalian Kalirin (Kalrn) gene locus gives rise to several isoforms which have diverse and sometimes antagonistic effects on dendritogenesis. The Kalrn gene can code for a protein with a lipid binding motif, several tandem spectrin homology domains and additional protein-protein interaction domains like SH3 domain commonly found in members of signaling pathways (McPherson et al., 2002, 2004; Vishwanatha et al., 2012; Miller et al., 2017). The functional differences between isoforms are thought to arise from the combinations of protein domains retained in the splice isoforms. The effects of alternative isoforms of Kalrn on dendritic morphology have been studied extensively (Penzes et al., 2001; McPherson et al., 2002, 2004; Vishwanatha et al., 2012). However, the regulation of Kalrn transcription during development has not been explored extensively.

The human Kalrn gene is located on chromosome 3 and spans a huge 0.63 Mb region with a complex gene structure encompassing 17 alternatively spliced transcripts. The largest transcript of 19.9 kb contains 60 exons (McPherson et al., 2002). The shorter isoforms of human Kalrn fall into two broad categories, the ones containing combinations of the 30 exons from the 5′ end or 20 exons from the 3′ end. Johnson et al. (2000) and McPherson et al. (2002) reported the domain architecture of Kalirin protein isoforms and their relation to the alternative spliced transcript isoforms. Kalrn7, the major brain-specific isoform, and the closely related Kalrn8 isoform consist of the core Sec14p-like putative lipid-binding motif, nine spectrin-like repeats and a Dbl homology/pleckstrin homology (DH-PH) domain but differs at the amino terminus and have a distinct PDZ binding motif or an SH3 motif at the -COOH end respectively. Kalrn9 and Kalrn12 have additional C-terminal domains and show distinct sub-cellular localization in cortical neurons while Kalrn7 shows characteristic punctate localization in the neuronal processes. The transcript isoforms also have distinct spatio-temporal expression patterns with Kalrn7 being expressed in the adult brain, while Kalrn9 and Kalrn12 are expressed in the developing brain and yet another isoform Trio is expressed in many tissues (Hansel et al., 2001).

Zebrafish is emerged as a convenient model to explore axonal growth (Chen et al., 2013), signaling by neurotropic factors during development and in human diseases. Thus, we studied the kalrna locus in detail on chromosome 9 of the zebrafish genome (Zv10). The 2.9 kb predicted transcript from the longest isoform (ENSDART00000164543.1) consists of the expected core domains (spectrin repeats, DH-PH domains). The kalrna locus has been extensively re-annotated during the transition from zebrafish genome version 9 (Zv9) to version 10 (Zv10). A notable change is the extension of the 5′ end to include an exon positioned 95 kb upstream to the previously reported kalrna gene. We noticed that in Zv9, this region was inverted and hence the link with the kalrna gene was not evident. Large scale studies, most notably, the ENCODE project, have shown that extended 5′ terminal exons can play an important role in the regulation of the gene. On closer inspection, we found that the region also corresponds to an lncRNA reported by Kaushik et al. (2013) from an RNA-seq based developmental profiling of non-coding RNAs.

Here, we show that this lncRNA is transcribed in anti-sense to the largest kalrna transcript in zebrafish. The multi-domain Kalirin was named after the multi-armed goddess Kali from Indian mythology (Alam et al., 1997), and hence, we propose to name this lncRNA that is anti-sense to kalrna, durga- the mythological alter-ego of Kali. The lncRNA is expressed in the oocyte and as embryo is developing, it gets gradually restricted to the brain. The presence of this lncRNA seems to be critical for appropriate temporal expression of kalrna. The durga transcripts present in the cytoplasm of the fertilized embryo, subsequently increases and localize to the developing zebrafish brain. The rise in durga expression is accompanied by an increase in the level of kalrna mRNA and coincides with the developmental time when the zebrafish brain is being formed and neurites are growing as rapidly as 20 microns/h (Chen et al., 2013). Ectopic injection of the durga RNA results in further increase in the expression of kalrna. Primary zebrafish neurons with elevated durga and kalrna showed striking morphological changes in dendrites. Using a transgenic fish with neurons marked with kaede, we found that the ectopically injected lncRNA triggered a profound loss of dendrites, perhaps corresponding to the increased kalrna expression. Thus, transcription from the first exon of the kalrna gene, located about 95 kb upstream to the next exon in zebrafish, seems to be regulated by lncRNA durga.

The advent of next generation sequencing technologies has led to the rapid, high-throughput identification of thousands of putative non-coding RNAs. The zebrafish genome (latest version Zv10) has been recently re-annotated to incorporate modifications at many loci. We set out to study a putative lncRNA locus that was re-annotated in the recent version of the zebrafish genome.

Kalirin acts a pivotal point in the regulation of cytoskeletal organization and thus neuronal morphology because of integration of diverse, excitatory as well as inhibitory signals. Human Kalrn gene is a large gene that shows extensive alternative splicing leading to multi-domain protein isoforms whose functionality is determined by the combination of exons included in each transcript type. The alternative transcripts are of two broad categories, the ones that have predominantly 5′ exons, for instance, Kalrn-7, 9, 12 and others that have predominantly 3′ exons like Duo and Trio. These transcript isoforms also differs in the spatio-temporal expression pattern as seen in Kalrn7 found in the adult brain and Kalrn9 and 12 predominantly expressed during development. Kalirin protein isoforms seem to respond to a variety of signals including Nerve Growth Factor signaling through the TrkB receptor, BDNF signaling through Rac1 (Chakrabarti et al., 2005; Yan et al., 2016). It also interacts with many PDZ domain containing proteins such as PSD95 (Penzes et al., 2001). Kalirin proteins can integrate these signals to modify actin cytoskeletal dynamics eventually modifying neurite morphology and axonal growth of neurons.

We noted a putative, brain enriched lncRNA that directly overlaps with the extended exon of the kalrna gene in the Zv10 annotation of the zebrafish genome. Using a combination of semi-quantitative RT-PCR with directional primers and in situ hybridization, we established that the locus gives rise to an lncRNA durga and the transcript of kalrna gene. Both durga and kalrna showed increasing expression during development, finally localizing to the brain region by 24 hpf. Further, over-expression of the lncRNA resulted in an increase in the expression of kalrna and decrease in the number of dendrites in zebrafish neurons. Currently, no detailed information on the transcript isoforms of kalrn in zebrafish is known. However in future, it will be interesting to explore the isoform-specific effects of the durga in neurodevelopment and neurodegenerative disorders. Besides durga, additional factors are likely to be involved in the regulation of zebrafish kalrna.

Depletion of the endogenous durga lncRNA is technically challenging due to the overlap with the kalrna gene. Standard tools like CRISPR or morpholinos are not readily applicable. CRISPR based editing would disrupt the kalrna gene on the opposite strand. Lack of information on exon-intron structure in durga gene prevented use of morpholinos to knock-down the expression of the lncRNA. In future, it will be interesting to study the mechanism of kalrna regulation by durga.

The Kalrn over-expression, especially in the form of the transcript isoform Kalrn7 is usually associated with increased dendritic length while Kalrn9 and 12 isoforms seem to have age dependent, contrasting effects on dendritic length. In early development, Kalrn9 and 12 isoforms induce dendritic elongation while in mature neurons Kalrn9 over-expression reduces dendritic length (Grubisha et al., 2016). The dual role of Kalrn9 and 12 isoforms to induce or retract dendrites comes from their two GEF domains. The first GEF domain activates RhoG/Rac1, which promotes the formation and elongation of dendrites. The second GEF domain interacts with RhoA, which induces spine elimination (Tolias et al., 2011; Yan et al., 2015). The Kalrn7 isoform does not have a second GEF domain and thus over expression of Kalrn9 and 12 isoforms but not Kalrn7 reduces dendritic length (Deo et al., 2012; Grubisha et al., 2016). Interestingly Kalrn9 expression has been observed to be increased during normal human aging in the cortex and also in schizophrenia subjects (Deo et al., 2012; Grubisha et al., 2016).

Reduction in dendritic length, pruning of synaptic connections and decrease in brain size are common pathological symptoms of neurodevelopmental, neuro-degenerative and neuro-psychiatric disorders like autism, schizophrenia, Alzheimer’s (Remmers et al., 2014). On the other hand, controlled synaptic pruning and maintenance of dendritic arborization are also essential to maintain structural plasticity of brain throughout life. lncRNAs have been reported to be involved in almost every aspect of brain development with diverse mechanisms of action. For example, lncRNA POU3F2, Gomafu, HAR1F are involved in neural stem cell proliferation while lncRNA Pnky, NEAT1, TUG1, EVF2 regulates neuronal differentiation (Iyengar et al., 2014). Besides, the role of lncRNAs in regulation of neuronal architecture and synaptic plasticity is beginning to be understood. Here we showed that lncRNA durga shortens the dendrites and reduces dendritic arborization in zebrafish, which may have higher level implications during neurodevelopment, neuropathologies and aging.

lncRNAs are poorly conserved at nucleotide levels, but criteria of sequence level conservation is too narrow for lncRNAs. Large number of lncRNAs shows structural, functional conservation and expression from syntenic loci despite lack of sequence homology (Diederichs, 2014; Johnsson et al., 2014). We found that the 5′ end of human and mouse Kalrn locus gives rise to non-coding RNA. Although the sequence of durga, human and mouse lncRNA is not conserved, the gene arrangement is similar (Figure 5A). Putative human counterpart of durga has three exons and its first exon has overlap with first exon of Kalrn. Further expression of human Kalrn non-coding RNA was checked by PCR using cDNA prepared from RNA of SH-SY5Y cell line. We found both spliced (239 bp) as well as unspliced (500 bp) form of human Kalrn non-coding RNA (Figure 5B).

In both the mammalian genomes, the lncRNA is annotated in the same orientation as the Kalrn gene, while in zebrafish it is anti-sense and overlapping. Further functional studies and evaluation of durga in mouse model of neurodegenerative diseases may pave a way for better understanding of Kalrn regulation during development and in various neuropathologies.

Kalirin regulates higher level brain functions by modulating dendritic spine density and morphology. We find an antisense, promoter-proximal lncRNA co-expressed with Kalirin that affects dendrite length and density in zebrafish.

BP conceived the idea. BP and MAS designed the experiments. DC and MKM collected zebrafish samples. MKM and SR maintained zebrafish stocks. MAS and DC performed PCR, qPCR, primary cell culture experiments and data analysis. DC did Northern blotting experiments. AB and DC performed the in situ hybridization experiments. MK captured the confocal images. BP, MAS and DC co-wrote the article.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.