Are Protocadherin Gamma Cluster transcripts considered separate genes?

Are Protocadherin Gamma Cluster transcripts considered separate genes?

We are searching data for your request:

Forums and discussions:
Manuals and reference books:
Data from registers:
Wait the end of the search in all databases.
Upon completion, a link will appear to access the found materials.

The [email protected] cluster of genes share exons and are therefore isoforms of each other. However, they are considered separate genes, by both HUGO and Ensembl.

A "gene" is defined as "a locus of co-transcribed exons" (Venter 2001), and this locus clearly fits that definition with no ambiguity. So why are these transcripts considered separate genes?

I would say they are considered different genes because each isoform is under the control of its own promoter.


Marine mammals are a classic example of convergent evolution in terms of adaptation of terrestrial mammals to the marine environment. During secondary adaptation to the marine environment, marine mammals experienced similar environmental challenges, which have resulted in shared morphological or physiological features across distant taxa. For instance, they have experienced similar changes in skin and limbs, and subsequently became streamlined 1,2 . Adaptive traits related to hypoxia are shared features of marine mammals 2,3 .

Marine mammals include three orders: cetaceans (whales, dolphins, and porpoises), pinnipeds (seals, sea lions, and walruses), and sirenians (manatees and dugongs) 4 . They have evolved to inhabit the ocean in multiple lineages. Cetaceans and sirenians emerged around 40–50 million years ago (mya) from Cetartiodactyla and Afrotheria, respectively 5 . Pinnipeds emerged within the Carnivora approximately 20 million years later 5 . This implies that different molecular changes occurred across separate lineages, possibly resulting in divergent phenotypic changes. However, most studies related to marine mammals have focused on convergent evolution, although some of the adaptations of marine mammals to an aquatic lifestyle vary among species 5 .

Pinnipeds, which consist of three families (Phocidae, Otariidae, and Odobenidae) are distinguishable from other marine mammals 6 . Most pinnipeds are semi-aquatic, unlike other marine mammals that spend their entire lives in the water 4 , and have modified limbs as flippers that propel them both in the water and on land 7 . In addition, with the exception of the walrus, which is the only extant species of the family Odobenidae, all pinnipeds have fur coats 8 . These distinct characteristics have not been sufficiently characterized at the molecular level. Although a draft fur seal genome has recently been assembled 9 , the evolutionary and biological aspects of pinnipeds have not been investigated. Indeed, the genome of the Weddell seal (family Phocidae) has not been completed ( In addition, most phylogenetic studies of pinnipeds have used limited marker sequences, such as that of the mitochondrial genome 10,11,12 .

Comparative genomics enables investigation of the convergent evolution of distant species. For example, convergent amino acid changes for vocal learning were identified by sequencing 48 avian genomes 13 . Similarly, Parker et al. 14 reported nearly 200 convergent loci in the genomes of echolocating mammals. Although there are more studies to demonstrate to phenotypic convergence-linked sequence convergence, molecular convergence toward phenotypic convergence, at least in marine mammals, seems to be uncommon. By analyzing 22 mammalian genomes, including those of three marine mammals, Foote et al. 15 suggested that different molecular pathways could be used to reach the same phenotype. In a study of the Hox gene family in mammals, only a fraction of sites had positive selection signatures shared by three independent marine mammal lineages 16 . Rather than sequence-level, gene-level convergence was presented as widespread signatures when evolutionary rates were used 2 . Therefore, there is convergence at the functional level or higher in separate mammalian lineages, and different marine mammal lineages have used different molecular pathways to achieve phenotypic convergence.

Here, we constructed draft genomes of three species of two pinniped families: Phoca largha (Phocidae) and Callorhinus ursinus and Eumetopias jubatus (Otariidae) (Fig. S1 and Supplementary Note S1). We identified genes with a positive selection signature that were common to the three pinnipeds but absent from other mammals, which are likely related to the unique traits of pinnipeds. In addition, divergent molecular changes likely to occur only in the pinniped lineage during phenotypic convergence of marine mammals were investigated.

Protocadherin-1: epithelial barrier dysfunction in asthma and eczema

Asthma and eczema result from the interaction between genetic susceptibility and environmental exposures. Over the last two decades, genetic studies have led to the identification of multiple susceptibility genes, increasing our understanding of the biological pathways leading to these diseases.

Many asthma and eczema susceptibility genes are expressed in epithelial cells of airway mucosa and skin [1]. The epithelium constitutes the first line of defence against allergens and invading microbes. Recently, loss of epithelial integrity has been implicated as an important mechanism leading to both asthma [2] and eczema [3].

In this issue of the European Respiratory Journal, M ortensen et al. [4] report the association of genetic variants in the gene encoding protocadherin-1 (PCDH1) with asthma, wheeze and eczema in (early) childhood in the Danish birth cohort study COPSAC (Copenhagen Studies on Asthma in Childhood). Their studies also suggest a gene–environment interaction of PCDH1 with passive smoke exposure in early childhood in asthma development. These studies expand on previous data reporting a role for PCDH1 polymorphisms in the susceptibility to both asthma [4–7] and eczema [4, 6, 8]. Given the replicated association of PCDH1 with both diseases, this gene might contribute to a biological pathway that constitutes a shared cause for both diseases. Here, we will summarise the cumulative data on PCDH1 polymorphisms in asthma and eczema, and make an attempt to interpret these genetic data in light of PCDH1 expression and function as a contributor to epithelial integrity.

In 1993, PCDH1 was originally identified by S ano et al. [9] as protocadherin-42 and was reported to induce cell adhesion upon ectopic membrane expression in a mouse fibroblast L-cell assay. PCDH1 is a member of the δ1 subfamily of the nonclustered protocadherin genes. PCDH1 is characterised by seven extracellular cadherin (EC) repeats, a transmembrane domain and three evolutionarily conserved motifs (CM1–3) in the intracellular tail that have been suggested to participate in intracellular signalling processes [10]. The PCDH1 gene spans 25 kb on chromosome 5q31–33 and encodes two main transcripts that contain either three or five exons through alternative splicing. The three-exon isoform does not encode the cytoplasmic domains CM1–3 that are present in the five-exon isoform. Recently, a sixth exon and alternative expression of exon 1 were identified in bronchial epithelial cells [11], whereas skin epithelial expression of PCDH1 has not been extensively characterised to date.

In 2009, PCDH1 was identified as a susceptibility gene for bronchial hyperresponsiveness (BHR) and asthma in children and adults. Subsequently, PCDH1 was shown to be expressed in epithelial cells of the airway [5, 9] and the skin [12]. In four genetic association studies on PCDH1 published to date, PCDH1 gene variants were associated with BHR, asthma, eczema, nonallergic childhood asthma and transient early wheeze (table 1) [4–6, 8]. However, in large-scale genome-wide association studies of asthma, such as the GABRIEL consortium study [7], PCDH1 was not reported to be associated, at a genome-wide significance level, with asthma defined as a self-reported “asthma ever” diagnosis. In our opinion, these contradictory results may be explained by asthma heterogeneity.

Asthma is a chronic heterogeneous disease of the airways with different disease subtypes, characterised by specific onset of respiratory symptoms, such as cough and wheeze. M ortensen et al. [4] report, in a longitudinal analysis of specific disease subtypes, that PCDH1 gene variants were associated with specific temporal episodes of wheezing and coughing symptoms, and with the onset of asthma. Moreover, characteristic respiratory symptom patterns in early childhood, such as transient early respiratory symptoms and recurrent troublesome symptoms, were associated with different PCDH1 polymorphisms. These specific associations with subphenotypes of the disease may well have been missed in an association analysis with “Have you ever had asthma?” as an outcome parameter.

In summary, while different association analyses identify a variety of genetic signals in the PCDH1 gene, the general picture that starts to emerge indicates that PCDH1 gene function is associated with asthma, BHR and eczema in early life. Importantly, gene by passive smoking interactions were found to be relevant for the association of PCDH1 with asthma in two populations [4, 5]. These data suggest that the contribution of PCDH1 gene variants to asthma may become evident when studied in the right environmental context. Well-powered, collaborative meta-analyses of cohorts with specific asthma subphenotypes in early life are needed to provide more robust evidence of these associations (table 1). Finally, the lack of association with specific or total IgE levels or with allergic asthma [6] suggests that PCDH1 is relevant to non-IgE-mediated mechanisms of disease susceptibility. Consequently, the contribution of PCDH1 to asthma susceptibility appears to be IgE independent (table 1).

The cumulative data gathered to date indicate that PCDH1 harbours a complexity of genetic signals associated with asthma and eczema that we do not yet fully understand, without a clear dominant polymorphism that is associated with asthma or eczema in most studies. In order to understand the interrelation of these associations observed for the different PCDH1 single-nucleotide polymorphisms (SNPs), we considered the possibility of linkage disequilibrium (LD) between these SNPs. LD is defined as the preferential occurrence of specific combinations of SNPs in the population, due to their co-inheritance as a result of their close physical proximity. In a recent study in a German population, T oncheva et al. [6] investigated the possibility that associations of PCDH1 with asthma were caused by LD with the nearby cytokine cluster on chromosome 5q31–33, but this was not the case. Our analysis of the LD pattern in Caucasian subjects did not reveal strong evidence for LD between the various SNPs (fig. 1), indicating that these SNPs do not represent a single genetic signal but apparently make individual, independent contributions to disease susceptibility.

Protocadherin-1 gene, single-nucleotide polymorphism (SNP) positions and linkage disequilibrium (LD) pattern: position on reference contig NT_ AC094107.3 (Homo sapiens) based on genomic data. LD plots are displayed in r 2 values. The stronger the LD, the darker the squares r 2 =0 is white and r 2 =1 is represented in black. LD plots were constructed with Haploview (version 4.2) software using SNP data from the HapMap Project ( for the CEU population. Bold type represents SNPs that also appear in table 1.

Another explanation for the identification of multiple PCDH1 SNPs to be independently associated with asthma and eczema in different studies is the presence of multiple SNPs within the PCDH1 gene that have diverse functional effects. Firstly, very specific SNPs might interact with environmental factors such as passive smoke exposure. Secondly, other SNPs may have more generic effects, for instance, by regulating PCDH1 gene expression. It is important to realise that the SNPs tested in the original study by K oppelman et al. [5] were only situated in coding regions of the gene. Over the past years, it has become increasingly clear that noncoding SNPs can have important regulatory effects on gene expression (by acting as expression quantitative trait loci (eQTLs)). The current publication by M ortensen et al. [4] adds the association of further noncoding PCDH1 polymorphisms to asthma and eczema (table 1), whereas none of the initial associations were replicated in this study. Thus, eQTL studies on these PCDH1 SNPs in lung and skin tissue are therefore urgently needed to assess their functional relevance. Thirdly, coding SNPs may have generic effects by affecting PCDH1 protein function. The coding SNP Ala514Thr maps to an EC domain in the PCDH1 protein, which is compatible with a potential effect on cell–cell adhesion that needs further study. So far, no data are available on the functional effects of any of the asthma- and eczema-associated SNPs. Finally, adding to the complexity of PCDH1, its expression is also probably regulated by epigenetic mechanisms that include gene methylation, given the identification of exon 1A as a CpG island that is potentially methylated [11].

So, how does PCDH1 contribute to both asthma and eczema? We hypothesise that PCDH1 dysfunction contributes to the defect in epithelial barrier function observed in both asthma and eczema by regulating the process of epithelial cell differentiation and/or repair. PCDH1 mRNA and protein expression is strikingly upregulated during primary bronchial epithelial cell differentiation, suggesting a role for PCDH1 in achieving or maintaining terminal differentiation of the airway epithelium [11]. The regulation of PCDH1 expression in the skin during epidermal differentiation has not been characterised to date, although PCDH1 expression is markedly induced in primary keratinocytes late during the repair process after scratch wounding [12]. A similar role in epidermal differentiation has been previously identified for multiple eczema genes in genetic studies. All in all, we propose a role for PCDH1 in epithelial differentiation and repair in both asthma and eczema.

PCDH1 is capable of intracellular signalling. Downstream pathways or PCDH1 interaction partners relevant to asthma or eczema pathogenesis have not been reported to date. Moreover, data on the differential expression of PCDH1 in asthma or eczema versus healthy controls are still lacking. Future studies are needed that can validate the functional role of PCDH1 in asthma and eczema by controlled experimental approaches, such as small interfering RNA or genetically engineered mouse models, that are currently being developed. These studies also need to address the possibility of environmental exposures affecting PCDH1 expression, with cigarette smoke exposure being of particular interest. Clearly, more studies are needed to resolve the functional effects of PCDH1 SNPs and their relevance for airway and skin epithelial cell biology. These studies will ultimately validate the initial genetic findings and perhaps provide clues for future novel therapeutic interventions in asthma and eczema.


Astrocytes express multiple γ-Pcdhs

We focused on the developing spinal cord, which matures earlier than the brain and in which the consequences of γ-Pcdh disruption have best been characterized (Wang et al., 2002b Weiner et al., 2005 Prasad et al., 2008). The 6 extracellular cadherin repeats, transmembrane domain, and proximal cytoplasmic domain of each γ-Pcdh protein is encoded by one of 22 large “variable” (V) exons, each of which is spliced to three “constant” (C) exons that encode a shared 125-amino acid C-terminal domain (Wu and Maniatis, 1999 Tasic et al., 2002 Wang et al., 2002a). Using an in situ hybridization (ISH) probe against the C exons, we found that, in addition to ubiquitous expression in the gray matter, Pcdh-γ transcripts are expressed by cells in the white matter of the neonatal spinal cord (Fig. 1 B). Double-label fluorescent ISH (Fig. 1 B, inset) and immunofluorescence in both neonatal and adult spinal cord (Fig. 1 C,D) showed that many of these cells were GFAP-positive astrocytes, which extend radial processes that contain γ-Pcdhs. Coexpression of Pcdh-γ genes with the PLP gene, a marker for oligodendrocytes, was not observed in double-label fluorescent ISH to neonatal sections (data not shown). GFAP-positive astrocytes in the gray matter also appeared to contain γ-Pcdh proteins, but the ubiquitous localization of neuronal γ-Pcdhs throughout the neuropil made it difficult to ascertain their subcellular localization (see below, and Fig. 2, for subsequent analyses addressing this). Astrocytes cultured from the neonatal spinal cord (Fig. 1 G) and cortex (data not shown) expressed multiple Pcdh-γ genes, with 20 of the 22 expected spliced transcripts detectable by RT-PCR of astrocytic RNA.

Astrocytes express multiple γ-Pcdhs. In situ hybridization (ISH) using a Pcdh-γ C exon riboprobe (diagrammed in A) revealed expression in cells of the white matter (arrowheads) as well as the gray matter in neonatal spinal cord ( B). Double-labeling with a GFAP riboprobe indicated that many of these cells were astrocytes ( B, inset shows higher-magnification view). Immunofluorescence confirmed γ-Pcdh protein expression by GFAP-positive astrocytes in neonatal ( C) and adult ( D) spinal cords. Immunostaining of cultured astrocytes ( E) and spinal cord sections ( F) demonstrated that γ-Pcdh proteins colocalize with ezrin and Glt-1a, both markers of perisynaptic astrocyte processes. Reverse-transcriptase (RT)-PCR analyses detected expression of 20 of the 22 possible Pcdh-γ spliced transcripts in cultured spinal cord astrocytes ( G). Scale bars: B, 100 μm (inset, 10 μm) C, D, 25 μm E, 3 μm F, 4 μm.

γ-Pcdh proteins localize to perisynaptic astrocyte processes. A, Schematic diagram of the Pcdhfcon3 conditional mutant allele, in which a GFP-fused C exon 3 is flanked by loxP sites. The A–C subfamilies of V exons are in shades of blue, and the 3 C exons are in red. B, In Cre-ER Pcdhfcon3/+ mice injected with tamoxifen ( B, right), Cre excision resulted in loss of the ubiquitous γ-Pcdh-GFP immunofluorescence observed in uninjected or Cre-ER-negative mice ( B, left). GFP-tagged γ-Pcdhs remained detectable, however, in a small number of GFAP-positive gray matter protoplasmic astrocytes ( B). C, A higher-magnification image of a separate double-immunostained cryosection is shown. D, E, Confocal analysis of immunostained Cre-ER Pcdhfcon3/+ spinal cord sections demonstrated that astrocytic γ-Pcdh puncta (stained for both GFP and γ-Pcdh closed arrowheads yellow in merged images) can localize directly adjacent to neuronal γ-Pcdh at synaptic puncta (stained for both bassoon and γ-Pcdh open arrowheads, magenta in merged images). F, G, Immunostaining for GFP and presynaptic and postsynaptic markers confirmed that astrocytic γ-Pcdhs were frequently apposed to and/or wrapped around synaptic terminals. Large images are of single planes from z-stacks perpendicular cross-sections through the entire z-stacks at locations indicated by white lines are shown at right and below. H, Tracings of several such perisynaptic contacts for the indicated boxes in F and G are presented. Scale bars: B, 100 μm C, 6 μm D, E, 1 μm F, G, 2 μm.

Γ-Pcdhs localize to perisynaptic astrocyte processes

In astrocyte cultures, γ-Pcdh proteins were concentrated at cell–cell junctions and in small, filopodia-like processes that also contained ezrin (Fig. 1 E), a marker of perisynaptic astrocyte processes in vivo (Derouiche and Frotscher, 2001). γ-Pcdh proteins colocalized extensively with another marker of these processes, the glutamate transporter Glt-1a (Rothstein et al., 1994 Cholet et al., 2002), in the spinal cord in vivo (Fig. 1 F). To more clearly determine the subcellular localization of astrocytic γ-Pcdhs, we developed a genetic method using a conditional mutant allele, termed Pcdhfcon3 , in which a C exon 3-GFP fusion is flanked by loxP sites (Fig. 2 A). Ubiquitous homozygous excision of the Pcdhfcon3 allele results in a phenocopy of Pcdhdel/del -null mutants (Prasad et al., 2008) because of the GFP fusion, heterozygous Pcdhfcon3/+ mice can also be used to report on Cre activity in phenotypically normal tissues.

We crossed Pcdhfcon3 mice with a transgenic line expressing a Cre-estrogen receptor (Cre-ER) fusion protein that can localize to the nucleus only on binding of the synthetic ligand tamoxifen (Guo et al., 2002). After tamoxifen injection of Cre-ER Pcdhfcon3/+ heterozygous mice, we found that the GFP-tagged exon was removed from almost all cells in the spinal cord. A small number of isolated cells in the gray matter, however, escaped excision, and thus retained GFP-tagged γ-Pcdh proteins (Fig. 2 B). These cells had the morphology of protoplasmic astrocytes, and expressed GFAP (Fig. 2 C) but not the neuronal marker NeuN (see supplemental Fig. S2, available at as supplemental material). By using confocal microscopy to examine sections from these heterozygous mice stained with antibodies against GFP and excitatory and inhibitory synaptic markers, we thus could visualize astrocytic γ-Pcdhs in isolation and unambiguously determine their relationship to synapses in phenotypically normal tissue.

These analyses confirmed that astrocytic γ-Pcdhs were often concentrated in puncta directly adjacent to synaptic contacts in the spinal cord (Fig. 2 F,G), as well as in the hippocampus, where a subset of astrocytes also escaped Cre excision (supplemental Fig. S2, available at as supplemental material). These perisynaptic puncta were closely apposed to both excitatory (86% of 198 apposed synaptic puncta) and inhibitory (87% of 247 apposed synaptic puncta) synapses and often appeared to wrap around or interdigitate between them (Fig. 2 F–H). Similar analyses in which Cre-ER Pcdhfcon3/+ sections were triple-immunostained with antibodies against GFP (to detect astrocytic γ-Pcdhs), the constant domain (to detect all γ-Pcdhs) (see supplemental Fig. S1, available at as supplemental material), and the presynaptic marker bassoon identified γ-Pcdh-positive synaptic puncta adjacent to astrocytic GFP/γ-Pcdh-double-positive puncta (Fig. 2 D,E). Together, these results indicate that the γ-Pcdhs are appropriately situated to mediate adhesive contacts between neuronal synapses and perisynaptic astrocyte processes.

Astrocytic γ-Pcdhs are critical for synaptogenesis in developing neurons in vitro

To determine whether astrocytic γ-Pcdhs play a role in synaptogenesis, we began by modifying a coculture system used in our previous work, in which embryonic spinal interneurons grow atop a lawn of astrocytes, differentiating and forming fully mature synapses over a period of ∼9 d (Weiner et al., 2005). Instead of mixed cultures prepared from the same animals, we separately established a monolayer of neonatal spinal cord astrocytes, on which interneurons from E13 spinal cords were subsequently plated: this allowed us to independently control the genotype of each cell type. Neurons do not survive the astrocyte preparation (data not shown), obviating any concern that interneurons of the undesired genotype might be carried over in cocultures. The converse complication—carrying over astrocytes when plating the interneuron preparation—is, however, possible. Therefore, to ensure that no astrocytes of the wrong genotype were present, we established test cocultures of WT astrocytes with neurons from Pcdhfusg/+ mice, which harbor the same C exon 3-GFP fusion as the Pcdhfcon3 allele (Wang et al., 2002b) (Fig. 3 A). In these cultures, only TuJ1-positive neurons were GFP-positive, confirming that our method allowed for complete separability of astrocyte and neuron genotype. We also confirmed that γ-Pcdhs can localize to perisynaptic processes in these cultures as they do in vivo, by conversely coculturing wild-type neurons with astrocytes prepared from Pcdhfusg/+ mice. Confocal microscopy of these cocultures demonstrated the presence of astrocytic γ-Pcdh puncta adjacent to many excitatory and inhibitory synapses (Fig. 3 B–D).

A–C, Cocultures of spinal cord interneurons and astrocytes of distinct genotypes. Cocultures of neurons from Pcdhfusg mice with wild type astrocytes were immunostained with antibodies against the indicated proteins ( A). Only TuJ1-positive neurons expressed GFP-tagged γ-Pcdhs, indicating that the coculture system allowed for effective separability of neuron and astrocyte genotypes with no carryover. Cocultures of wild type neurons with astrocytes from Pcdhfusg mice stained for GFP and inhibitory ( B) or excitatory ( C) synaptic markers demonstrated that astrocytic γ-Pcdhs localized to many sites adjacent to synaptic terminals in vitro, as observed in vivo. Large images are of single planes from z-stacks perpendicular cross-sections through the entire z-stacks at locations indicated by white lines are shown at right and below. D, Tracings of several such perisynaptic contacts for the indicated boxes in B and C are presented. Scale bars: A, 20 μm B, C, 2 μm.

We then proceeded to establish cultures in which either astrocytes or neurons were prepared from null mutant Pcdhdel/del mice (Wang et al., 2002b Weiner et al., 2005 Prasad et al., 2008) [“astro-null” (astrocyte-null) and “neuron-null,” respectively], and compared the progression of synaptogenesis in these cultures to that in cocultures in which both cell types were normal (“control”). We observed no apparent differences in survival, growth rate, morphology, or GFAP expression between control and mutant astrocyte monolayers. Neuronal differentiation and neurite outgrowth, assessed here by immunostaining for the markers MAP2 (Fig. 4 O) and TuJ1 (data not shown) were also indistinguishable across all types of cultures, consistent with our extensive previous demonstration of normal neurogenesis, differentiation, and axon outgrowth in Pcdh-γ-null mutant mice (Wang et al., 2002b Prasad et al., 2008).

Astrocytic γ-Pcdhs are critical for synaptogenesis in developing neuronal cultures. Cocultures of embryonic spinal interneurons growing directly atop a confluent astrocyte monolayer were prepared such that either astrocytes (astro-null), neurons (neuron-null), or neither (control) were Pcdh-γ-null, and synaptogenesis monitored by quantifying the density of appositions of presynaptic and postsynaptic markers (i.e., the number of such synaptic contacts per a defined area) as neurons matured between 6 and 9 d in vitro (DIV). When astrocytes were mutant, the numbers of both excitatory and inhibitory synapses were significantly reduced at 6 DIV ( B, H, M, N) compared with control cultures ( A, G, M, N), although some recovery occurred by 9 DIV ( E, K, M, N). When neurons were mutant, synapse density was drastically reduced at both 6 and 9 DIV ( C, F, I, L–N). Neuronal differentiation and survival, monitored by quantifying the area of MAP2 immunostaining, did not differ across the three culture conditions ( O). Data in M-O are graphed as percentage of control values. Graphs in M–O represent means ± SEM from 22 to 44 fields from 6 cultures per time point. *p < 0.05 ***p < 0.001. Scale bar (in L): A–L, 2 μm.

In control cultures, synapses (defined by apposition and partial overlap of presynaptic and postsynaptic puncta) were already numerous by 6 d in vitro (DIV) and increased in number by 9 DIV (Fig. 4 A,D,G,J,M,N). In astro-null cultures, however, few synapses had formed by 6 DIV: excitatory synapses were reduced by 77% (n = 22 fields from 6 cultures p < 0.001) and inhibitory synapses by 39% (n = 33 fields from 6 cultures p < 0.001) (Fig. 4 B,H,M,N). By 9 DIV, synapse density in astro-null cultures recovered toward control levels, although inhibitory synapses were still significantly reduced in number (Fig. 4 E,K,M,N). In neuron-null cultures, formation of both excitatory and inhibitory synapses was drastically reduced at both 6 DIV (by 67% combined total of n = 44 fields from 12 cultures p < 0.001) and 9 DIV (by 80% combined total of n = 44 fields from 12 cultures p < 0.001) (Fig. 4 C,F,I,L–N). These results indicate that astrocytic γ-Pcdhs are critical for synapse formation or stabilization in developing cultures, but that synaptogenesis can eventually proceed, provided that neurons themselves express the γ-Pcdhs.

Astrocytic γ-Pcdh control of synaptogenesis is contact dependent

Although the γ-Pcdhs can mediate homophilic adhesion in non-neural cell lines (Obata et al., 1995 Frank et al., 2005) it remains unclear whether this is their only mechanism of action in the nervous system. To ask whether loss of the γ-Pcdhs in astrocytes might have a secondary effect on their release of synaptogenic secreted factors, we collected tissue culture media that had been conditioned by either control (WT) or Pcdh-γ-null [knock-out (KO)] astrocytes. Astrocyte-conditioned media (ACM) from KO cultures contained levels of the known synaptogenic molecule TSP-2 (Christopherson et al., 2005) indistinguishable from that of WT ACM (Fig. 5 A). Consistent with this, addition of concentrated WT or KO ACM produced a quantitatively similar promotion of synaptogenesis in control cultures (Fig. 5 B), indicating that Pcdh-γ mutation does not disrupt astrocyte secretion of synaptogenic factors. Additionally, concentrated WT ACM was unable to completely rescue the reduced synapse density observed in astro-null cocultures (Fig. 5 C), suggesting that Pcdh-γ mutation disrupts other, presumably contact-dependent, mechanisms.

Astrocytic γ-Pcdhs regulate synaptogenesis through a contact-dependent mechanism. Astrocyte-conditioned medium (ACM) was collected from astrocyte cultures derived from WT or Pcdhdel/del mice (KO). Western blot analysis of equal amounts of ACM showed no detectible difference between genotypes in the levels of thrombospondin-2 (TSP-2) secreted ( A). Addition of concentrated WT or KO ACM to control cocultures over 6 DIV resulted in equally significant increases in the density of synapses (i.e., the number of synaptic contacts per a defined area) compared with control media ( B) thus, WT and KO ACM are equivalent in their synaptogenic potency. Addition of concentrated WT ACM was unable to completely rescue the decrease in synapse density observed in astro-null cultures ( C). Furthermore, the difference in synapse density between control and astro-null cocultures was maintained when the neurons were plated onto astrocytes that had been fixed with paraformaldehyde ( D). Data are graphed as percentage of control values. Graphs represent means ± SEM from 12 fields from 3 cultures per time point.

To directly assess whether astrocytic γ-Pcdhs promote synaptogenesis in a contact-dependent manner, we prepared cocultures in which the astrocytic monolayer had been fixed with paraformaldehyde before addition of interneurons. We reasoned, as others have (Yamagata et al., 1995 Barker et al., 2008), that this manipulation should remove any differences in cell signaling or in release of secreted factors between WT and KO astrocytes. At 6 DIV in these cocultures, both excitatory and inhibitory synapses remained significantly reduced when astrocytes were Pcdh-γ-null (Fig. 5 D), further supporting a role for astrocytic γ-Pcdhs in providing contact-dependent cues for neurons that promote synapse formation or stabilization.

Astrocyte-restricted loss of the γ-Pcdhs results in delayed synaptogenesis in vivo

We next asked whether astrocytic γ-Pcdhs are important for synaptogenesis in vivo, by using the conditional Pcdhfcon3 floxed allele and a transgenic line expressing the Cre recombinase under the control of the human GFAP promoter (Zhuo et al., 2001). Because many neurons are derived from radial glial precursors that may transiently express GFAP, this Cre line results in the excision of floxed alleles in some neurons as well as in astrocytes, and thus cannot be used for astrocyte-specific mutation in much of the CNS (data not shown see Zhuo et al., 2001). There is evidence that neurons of the spinal cord, however, are not derived from GFAP+ radial glial precursors (Barry and McDermott, 2005), which indicated that astrocyte-restricted excision might be attainable there.

We ascertained this by crossing GFAP-Cre mice with Z/EG reporter mice, in which ubiquitous expression of a β-galactosidase/neo fusion protein is replaced by that of GFP following Cre excision (Novak et al., 2000) (Fig. 6 A–C). In neonatal GFAP-Cre Z/EG double-transgenic spinal cords, immunostaining demonstrated that all GFP-positive cells were GFAP-positive and NeuN-negative, whereas β-galactosidase inclusions were detected only in NeuN-positive neurons (Fig. 6 B,C). To confirm that the floxed allele was being efficiently excised in the astrocytes of GFAP-Cre Pcdhfcon3 mice, we immunostained double-heterozygous mice for the GFP-tagged constant exon and observed that nearly all colocalization of γ-Pcdh-GFP with the perisynaptic astrocyte marker Glt1a was lost (Fig. 6 D,E). Having ascertained that GFAP-Cre effects astrocyte-specific excision in the developing spinal cord, we quantified synapses in sections from GFAP-Cre Pcdhfcon3/fcon3 mice and Pcdhfcon3/fcon3 control littermates at E15, E17, and P0, during the first major wave of synaptogenesis (May and Biscoe, 1975 Vaughn, 1989), the onset of which coincides with the appearance of GFAP+ astrocytes (Lee et al., 2003 Stolt et al., 2003 Barry and McDermott, 2005) (data not shown).

Astrocyte-restricted Pcdh-γ mutation in vivo. In GFAP-Cre Z/EG double transgenics ubiquitous expression of a β-galactosidase/neo fusion protein is replaced by that of GFP following Cre excision. GFAP-positive, but not NeuN-positive, cells express GFP, whereas NeuN-positive, but not GFAP-positive, cells express β-gal ( A, higher-magnification images of gray matter in B, C), indicating that Cre is not expressed in neurons or their progenitors in the spinal cord. In the gray matter of GFAP-Cre Pcdhfcon3/+ spinal cords, there is reduced colocalization of γ-Pcdh-GFP with Glt1a, confirming that Cre recombinase is able to efficiently excise the floxed allele in astrocytes in these mice ( D, E). Scale bars: A, 100 μm B, C, 25 μm D, E, 10 μm.

Because individual astrocytes are known to occupy nonoverlapping domains within which they contact multiple neuronal processes and synaptic terminals (Bushong et al., 2002 Halassa et al., 2007), we quantified synapses in circular fields 50 μm in diameter and centered on GFAP-positive, Pcdh-γ mutant astrocytes (Fig. 7 A–D). Both excitatory and inhibitory synapses were significantly reduced in fields from GFAP-Cre Pcdhfcon3/fcon3 mice compared with those from controls at E15 and E17 (Fig. 7 E). Importantly, no difference in the number of cleaved caspase-3-positive cells was observed between astroctye-restricted mutants and controls, indicating that the reduction in synapse density was not caused by increased levels of neuronal apoptosis (supplemental Fig. S3, available at as supplemental material). Similar to the recovery of synapse density observed in mature astro-null cocultures (Fig. 4 M,N), we found that at P0, the number of synapses in fields from GFAP-Cre Pcdhfcon3/fcon3 mice had reached control levels (Fig. 7 E supplemental Fig. S4B,D, available at as supplemental material). This astrocyte-restricted mutant result contrasts with our previous experiments using Pcdhdel/del Bax −/− mice (Weiner et al., 2005 Prasad et al., 2008), in which synapse density remains low at P0. Consistent with this, and with our coculture experiments (Fig. 4), we also found a significant reduction in synapse density persisted at P0 in Actin-Cre Pcdhfcon3/fcon3 mice, in which both neurons and astrocytes are mutant (Fig. 7 E supplemental Fig. S4C,D, available at as supplemental material). The remarkable congruence between these in vivo data and our coculture experiments provides strong support for developmentally regulated roles of both astrocytic and neuronal γ-Pcdhs in the control of synaptogenesis.

Astrocytic γ-Pcdhs control synaptogenesis in the embryonic spinal cord in vivo. Quantification of immunostained synapses was performed on 50 μm circular fields imaged in spinal cord sections from GFAP-Cre Pcdhfcon3/fcon3 astrocyte-restricted mutants and controls at E15, E17 ( A–D), and P0 (see supplemental Fig. S4, available at as supplemental material). In A′–D′, the yellow overlap between the red and green channels has been extracted and converted in Adobe Photoshop to black for clarity. E, Synapse density (i.e., the number of such synaptic contacts per 50 μm circle) was significantly reduced in GFAP-Cre Pcdhfcon3/fcon3 mutants (gray bars) compared with controls (white bars) at embryonic ages, but had recovered by P0, in contrast to Actin-Cre Pcdhfcon3/fcon3 mutants (black bars). Graphs represent means ± SEM from 18 fields from 3 animals per genotype per time point. *p < 0.05 **p < 0.01 ***p < 0.001.


Serotonergic neurons, like the other monoaminergic neurons, project their axons from the brain stem, including the raphe nuclei, to most regions of the CNS. However, the mechanism that controls the global projections of serotonergic neurons to their targets is largely unknown. Here we found that Pcdha transcripts were strongly expressed in the raphe nuclei from embryonic stages through adulthood, and therefore we examined the distribution of serotonergic projections from these nuclei in several mouse lines encoding mutant Pcdha proteins.

The deletion of the Pcdha-CR exons (Pcdha ΔCR/ΔCR mice) led to abnormally distributed serotonergic fibers in many brain regions by P21. These abnormalities were also detected in a full knock-out mutant of the Pcdha, in which the Pcdha gene cluster is completely deleted (Okayama A, Hasegawa S, Hirabayashi T, Katori S, Yagi T, unpublished data). Early in postnatal life (before the formation of terminal arbors in the target regions), the distribution pattern of serotonergic fibers in the Pcdha ΔCR/ΔCR mice was similar to that in WT mice. However, at P21, the distribution of serotonergic axons in Pcdha ΔCR/ΔCR mice was substantially altered. That is, the axons approached their target regions normally, but they did not form the extensive axonal arbors seen in adult WT mice at the same age, when the serotonergic projections are almost complete.

Lidov and Molliver (1982) proposed that the ontogeny of the serotonergic axonal projections is divided into three periods: first, axon elongation (E13–E16) second, development of selective pathways (E15–E19) and third, terminal field development (E19–P21). Our results indicate that the Pcdha proteins may function during the final period, during terminal field development. The Pcdha ΔCR/ΔCR mice showed low densities of innervating fibers in the central part of several target regions, such as the globus pallidus and substantia nigra, which could be explained by insufficient terminal arborization of the serotonergic fibers, which did penetrate the targets' periphery. Unlike the fiber termini, serotonergic axon bundles, such as the medial forebrain bundle, appeared normal in Pcdha ΔCR/ΔCR mice (Fig. 4C,D). During the terminal field development, the formation of serotonergic terminal arborization is dependent on the maturation of other elements in the local neuronal circuitry (Lidov and Molliver, 1982), suggesting that the Pcdha proteins might be involved in the interactions that promote terminal field development.

Factors that regulate serotonergic axon outgrowth and terminal arborization have been suggested by studies on the ontogeny of serotonergic innervation. S-100β, a Ca 2+ -binding protein secreted from astroglia, functions as a growth factor in vitro by inducing serotonergic sprouting (Azmitia et al., 1990 Haring et al., 1993), but S-100β-null mice show normal serotonergic projections (Nishiyama et al., 2002). Brain-derived neurotrophic factor (BDNF) can induce serotonergic axonal growth in the adult cortex (Mamounas et al., 1995, 2000), and serotonergic projections are normal in younger BDNF +/− mice, but a progressive loss of serotonergic fibers is observed when they are 12–18 months old (Lyons et al., 1999). Thus, the secreted S-100β and BDNF proteins regulate serotonergic projections, and the loss of either may be compensated for by other extracellular proteins, at least early in development. The cytoplasmic protein GAP-43, a growth-associated protein, is strongly expressed in monoaminergic neurons (Bendotti et al., 1991 Wotherspoon et al., 1997), and is a regulator for serotonergic projections. GAP-43-null mice have abnormal serotonergic projections, but the abnormalities are different from those of our Pcdha ΔCR/ΔCR mice. At P0, few serotonergic fibers in GAP-43-null mice have arrived at the hippocampus (Donovan et al., 2002), whereas in Pcdha mutants the fibers appear to arrive at the target almost normally. At P21, in the Pcdha mutants the amount of serotonergic axons arriving at the hippocampal region was normal, although their distribution within the hippocampal subregions was abnormal (Fig. 5F). In contrast, in GAP-43-null mice, the density of serotonergic axons arriving at the hippocampal region was lower, compared with WT mice (Donovan et al., 2002). Thus, in the regulation of serotonergic projections, GAP-43's function may be required in early developmental stages, and the roles of the Pcdha proteins may be significant only later in development.

We recently demonstrated that the OSNs of Pcdha ΔCR/ΔCR mice project abnormally to the olfactory bulb (Hasegawa et al., 2008). In the olfactory bulb of Pcdha ΔCR/ΔCR mice, axons expressing a single olfactory receptor (M71 or MOR23) do not coalesce into one glomerulus on each side of the olfactory bulb, but instead form multiple small glomeruli that persist even at P30 and P60. In WT mice, abnormal axonal projections to small multiple glomeruli are observed in the developing olfactory bulb however, these supernumerary glomeruli are gradually eliminated during postnatal maturation. Therefore, the abnormality of the OSN projections in the Pcdha ΔCR/ΔCR mice is consistent with the idea that Pcdha proteins function in the final maturation stage of axonal projections.

In the present study, we could not propose suitable models to explain the phenotype of serotonergic fibers in the Pcdha ΔCR/ΔCR mice because the mechanism for how Pcdha proteins modify serotonergic axon outgrowth and terminal arborization is still unclear. This ignorance is considerably due to the lack of the information on the adhesion properties and subcellular localization of Pcdha proteins. Although it has been suggested that Pcdha proteins interact with β1-integrin in a trans-heterophilic manner (Mutoh et al., 2004), trans-homophilic adhesion of Pcdha proteins could not be detected (Morishita et al., 2006). Furthermore, the subcellular localization of Pcdha proteins in serotonergic neurons and their targets could not be determined. During the formation of serotonergic terminal arbors, most of the neurons in the brain express detectable Pcdha transcripts. Our immunohistochemical examination with the Pcdha antibodies demonstrate intense immunoreactivity in the serotonergic somata but fail to determine the subcellular localization of the Pcdha proteins in the target regions of the serotonergic axons Pcdha immunoreactivity was rather homogeneously distributed in the neuropil (data not shown, Hasegawa et al., 2008). To clarify the subcellular localization of Pcdha proteins, further studies including immunoelectron microscopy will be necessary.

Here we demonstrated that the A-type cytoplasmic tail of Pcdha proteins is important for the normal distribution of serotonergic projections. The A-type cytoplasmic tail of Pcdhas is reported to associate with neurofilament M (Triana-Baltzer and Blank, 2006). Furthermore, Pcdha proteins are subject to matrix metalloprotease cleavage followed by presenilin-dependent intramembrane proteolysis, and the cleaved portion of the A-type cytoplasmic region translocates into the nucleus from the plasma membrane (Bonn et al., 2007). Therefore, the A-type cytoplasmic region may play a crucial role in signal transduction.

Although the signals transduced by the Pcdha proteins are not known, some possibilities have been suggested by recent findings. For example, a similar phenotype of abnormal, clumped axons was reported for the retinal axons of Dscam (Down syndrome cell adhesion molecule) mutant mice (Fuerst et al., 2008). Mammalian DSCAM, an Ig superfamily member, shows homophilic binding (Agarwala et al., 2000) it promotes self-avoidance (isoneuronal) for arborizing processes and inhibits the fasciculation of processes from different neurons (heteroneuronal) (Fuerst et al., 2008). This similarity raises the possibility that the Pcdha protein might function in the self-avoidance of serotonergic neurons. A-type cytoplasmic tail may transduce repulsive signals in the serotonergic neurons. The defect of self-avoidance may directly influence serotonergic axon outgrowth and terminal arborization. Alternatively, the insufficient self-avoidance may affect the configuration of recurrent collaterals of serotonergic neurons which regulate the activity of these neurons, and then indirectly change the terminal arbors by activity-dependent mechanisms.

Here we demonstrated that Pcdha molecules play significant roles in the establishment of normal serotonergic projections in every brain region we studied. We also found that in Pcdha ΔCR/ΔCR and Pcdha ΔA/ΔA mice, the serotonin levels of the hippocampus were significantly increased, compared with WT. Previously, we reported that hypomorphic Pcdha A-type mutant mice (Pcdha ΔBneo/ΔBneo ), in which Pcdha B-type transcripts are not detected and the expression level of Pcdha A-type transcripts is reduced, show an increased amount of serotonin in the hippocampus, and enhanced contextual fear conditioning and abnormal spatial learning (Fukuda et al., 2008). Together, these data are consistent with the idea that abnormal levels of serotonin in the hippocampus result from mutations of the Pcdha cytoplasmic tail, and that the altered serotonin levels could affect behavior.

The human protocadherin gene clusters (Pcdha, Pcdhb, and Pcdhg) are located on chromosome 5q31 (Wu and Maniatis, 1999). This locus is associated with BD (Lewis et al., 2003 Hong et al., 2004 Herzberg et al., 2006 Kerner et al., 2007). A recent study demonstrated a significant increase in homozygosity of the minor allele, which contains a single-nucleotide polymorphism in a putative enhancer for the Pcdha gene, in patients with BD (Pedrosa et al., 2008).

Whether or not the association with BD holds up, abnormal serotonergic innervation is implicated in a number of neuropsychiatric disorders, including depression, anxiety, schizophrenia, and autism (Baumgarten and Grozdanovic, 1995 Hen, 1996 Mann, 1998). Here, our observation that Pcdhas contributed to appropriate serotonergic projections in much if not all of the brain, suggests that Pcdhas could also be involved with the pathogenic mechanisms of a number of neuropsychiatric disorders, via their regulation of serotonergic projections.

Materials and Methods

Identification and Annotation of Cadherins in B. floridae

A dual approach was used to identify the cadherin repertoire in the cephalochordate B. floridae (amphioxus or lancelet). The assembly release version 2 of its genome (May, 2008), downloaded from DOE Joint Genome Institute (JGI:, contains 398 scaffolds ( Putnam et al. 2008). First, the six-frame translation of the amphioxus genome was searched using the five existing profile hidden Markov models (HMM) in the Pfam database ( Finn et al. 2010) and one newly built profile HMM based on an alignment of the first cadherin repeat (EC1), which we described previously ( Hulpiau and van Roy 2009) ( supplementary table S1 , Supplementary Material online). All 1,050 significant domain hits, sorted by scaffold, are shown in supplementary table S2 ( Supplementary Material online) and summarized in supplementary table S3 ( Supplementary Material online). Second, 292 cadherin sequences from a wide variety of metazoan taxa were aligned to the amphioxus genome by using tBLASTn ( Johnson et al. 2008) to identify potential orthologs ( supplementary table S4 and summarized in supplementary table S5 , Supplementary Material online). Finally, the results of both analyses were merged to yield the full list of cadherin superfamily members in amphioxus ( supplementary fig. S1 , supplementary table S6 , and supplementary note S1 , Supplementary Material online). In some cases, additional gene predictions using GenScan ( Burge and Karlin 1998) were performed in and around the genomic region in which the putative genes are located, and these predictions were then compared with the available RefSeq data and JGI gene models. Finally, the domains in all candidate genes were annotated based on CD search ( Marchler-Bauer et al. 2009) and Phobius ( Kall et al. 2007). The start and end of every cadherin repeat were corrected manually. The start was taken at the beginning of the adhesion arm featuring a conserved Glu residue as position 11 the end was the DxNDxxPxF motif, which is, together with Glu11, important for calcium binding.

Identification and Annotation of Cadherins in N. vectensis

The approach described above for identification of cadherins in amphioxus was also used on the genome of the nonbilaterian sea anemone N. vectensis. The N. vectensis genome assembly 1.0 contains 10,804 genome scaffolds ( Putnam et al. 2007). Profile HMM analysis yielded 1,016 hits ( supplementary table S7 , Supplementary Material online), which were grouped by scaffold and summarized in supplementary table S8 ( Supplementary Material online). The results of the tBLASTn analysis of 322 bilaterian cadherin sequences (the 292 sequences listed in supplementary table S4 , Supplementary Material online, plus 30 from B. floridae) versus the N. vectensis genome are listed in Supplementary Data ( Supplementary Material online) and summarized in supplementary table S10 ( Supplementary Material online). All the results were merged and annotated into the N. vectensis cadherin repertoire, as described for amphioxus ( supplementary fig. S2 , supplementary table S11 , and supplementary note S2 , Supplementary Material online).

Identification and Annotation of Cadherins in T. adhaerens

Again, both the profile HMM and tBLASTn methods were used to identify putative placozoan cadherins encoded by the T. adhaerens genome. The draft release v1.0 of T. adhaerens Grell-BS-1999 is assembled into 1,415 scaffolds ( Srivastava et al. 2008). Only 196 profile HMM hits, shown in supplementary table S12 ( Supplementary Material online), were found using the six HMMs listed in supplementary table S1 ( Supplementary Material online). This HMM analysis was combined with a tBLASTn analysis of 338 metazoan cadherins versus the T. adhaerens genome (see list in supplementary table S13 , Supplementary Material online). The metazoan cadherins analyzed were the 322 sequences listed in supplementary table S9 ( Supplementary Material online) plus 16 from N. vectensis. The results were merged and annotated into the T. adhaerens cadherin repertoire as described above ( supplementary fig. S3 , supplementary table S14 , and supplementary note S3 , Supplementary Material online).

Comparative and Phylogenetic Study of the Metazoan Cadherin Superfamily Members

For pairwise homology analysis of protein domains, we used “basic local alignment search tool two sequences” (bl2seq) ( Johnson et al. 2008). The analysis included sequence comparison of EC blocks, EC-by-EC analyses, and comparison of non-EC domains.

For multisequence homology analysis, sequences were aligned by ClustalX2 ( Larkin et al. 2007) using the PAM protein weight matrix, a gap open penalty of 5 and a gap extension penalty of 0.05 in both the pairwise and the multiple parameter settings. A neighbor joining (NJ) tree was constructed with 1,000 bootstrap replicates, and a Bayesian inference (BI) consensus tree was built by using MrBayes 3 ( Ronquist and Huelsenbeck 2003) (100,000 generations, sample frequency 100, and burnin 25%). Both types of trees were drawn by using Dendroscope ( Huson et al. 2007). The NJ tree is represented as a radial cladogram with bootstrap values and the BI tree as a radial phylogram with Bayesian posterior probabilities.

We used VectorNTI ( Lu and Moriyama 2004) to generate Clustal W alignments of the 7EC and the 6EC ectodomains of Ciona intestinalis protocadherins and also of the human protocadherins. By using the AlignX module of VectorNTI, we added protein domain annotations below the relevant sequence alignment blocks. To compare individual EC domains from Ciona protocadherins with the corresponding genomic sequences, we performed BLAT analyses ( Kuhn et al. 2009) in the University of California Santa Cruz genome browser ( The full protein sequences of selected members in each family in the phylogenetic tree were aligned similarly by the Clustal W algorithm in VectorNTI.


Takeichi M. Dynamic contacts: rearranging adherens junctions to drive epithelial remodelling. Nat Rev Mol Cell Biol. 201415:397–410.

Suzuki ST. Protocadherins and diversity of the cadherin superfamily. J Cell Sci. 1996109(Pt 11):2609–11.

Sano K, Tanihara H, Heimark RL, Obata S, Davidson M, St JT, et al. Protocadherins: a large family of cadherin-related molecules in central nervous system. EMBO J. 199312:2249–56.

Frank M, Kemler R. Protocadherins. Curr Opin Cell Biol. 200214:557–62.

Hulpiau P, van Roy F. Molecular evolution of the cadherin superfamily. Int J Biochem Cell Biol. 200941:349–69.

Boggon TJ, Murray J, Chappuis-Flament S, Wong E, Gumbiner BM, Shapiro L. C-cadherin ectodomain structure and implications for cell adhesion mechanisms. Science. 2002296:1308–13.

Patel SD, Chen CP, Bahna F, Honig B, Shapiro L. Cadherin-mediated cell-cell adhesion: sticking together as a family. Curr Opin Struct Biol. 200313:690–8.

Morishita H, Umitsu M, Murata Y, Shibata N, Udaka K, Higuchi Y, et al. Structure of the cadherin-related neuronal receptor/protocadherin-alpha first extracellular cadherin domain reveals diversity across cadherin families. J Biol Chem. 2006281:33650–63.

Morishita H, Yagi T. Protocadherin family: diversity, structure, and function. Curr Opin Cell Biol. 200719:584–92.

Kim SY, Yasuda S, Tanaka H, Yamagata K, Kim H. Non-clustered protocadherin. Cell Adhes Migr. 20115:97–105.

Takagi J. Structural basis for ligand recognition by RGD (Arg-Gly-Asp)-dependent integrins. Biochem Soc Trans. 200432:403–6.

Mutoh T, Hamada S, Senzaki K, Murata Y, Yagi T. Cadherin-related neuronal receptor 1 (CNR1) has cell adhesion activity with beta1 integrin mediated through the RGD site of CNR1. Exp Cell Res. 2004294:494–508.

Yagi T. Molecular codes for neuronal individuality and cell assembly in the brain. Front Mol Neurosci. 20125:45.

Ribich S, Tasic B, Maniatis T. Identification of long-range regulatory elements in the protocadherin-alpha gene cluster. Proc Natl Acad Sci U S A. 2006103:19719–24.

Tasic B, Nabholz CE, Baldwin KK, Kim Y, Rueckert EH, Ribich SA, et al. Promoter choice determines splice site selection in protocadherin alpha and gamma pre-mRNA splicing. Mol Cell. 200210:21–33.

Kaneko R, Kato H, Kawamura Y, Esumi S, Hirayama T, Hirabayashi T, et al. Allelic gene regulation of Pcdh-alpha and Pcdh-gamma clusters involving both monoallelic and biallelic expression in single Purkinje cells. J Biol Chem. 2006281:30551–60.

Hirano K, Kaneko R, Izawa T, Kawaguchi M, Kitsukawa T, Yagi T. Single-neuron diversity generated by Protocadherin-beta cluster in mouse central and peripheral nervous systems. Front Mol Neurosci. 20125:90.

Esumi S, Kakazu N, Taguchi Y, Hirayama T, Sasaki A, Hirabayashi T, et al. Monoallelic yet combinatorial expression of variable exons of the protocadherin-alpha gene cluster in single neurons. Nat Genet. 200537:171–6.

Zipursky SL, Sanes JR. Chemoaffinity revisited: dscams, protocadherins, and neural circuit assembly. Cell. 2010143:343–53.

Lefebvre JL, Kostadinov D, Chen WV, Maniatis T, Sanes JR. Protocadherins mediate dendritic self-avoidance in the mammalian nervous system. Nature. 2012488:517–21.

Wu Q, Maniatis T. A striking organization of a large family of human neural cadherin-like cell adhesion genes. Cell. 199997:779–90.

Hayashi S, Takeichi M. Emerging roles of protocadherins: from self-avoidance to enhancement of motility. J Cell Sci 2015.

Rubinstein R, Thu CA, Goodman KM, Wolcott HN, Bahna F, Mannepalli S, et al. Molecular logic of neuronal self-recognition through protocadherin domain interactions. Cell. 2015163:629–42.

Thu CA, Chen WV, Rubinstein R, Chevee M, Wolcott HN, Felsovalyi KO, et al. Single-cell identity generated by combinatorial homophilic interactions between alpha, beta, and gamma protocadherins. Cell. 2014158:1045–59.

Hirayama T, Yagi T. Clustered protocadherins and neuronal diversity. Prog Mol Biol Transl Sci. 2013116:145–67.

Wu Q, Zhang T, Cheng JF, Kim Y, Grimwood J, Schmutz J, et al. Comparative DNA sequence analysis of mouse and human protocadherin gene clusters. Genome Res. 200111:389–404.

Golan-Mashiach M, Grunspan M, Emmanuel R, Gibbs-Bar L, Dikstein R, Shapiro E. Identification of CTCF as a master regulator of the clustered protocadherin genes. Nucleic Acids Res. 201240:3378–91.

Yokota S, Hirayama T, Hirano K, Kaneko R, Toyoda S, Kawamura Y, et al. Identification of the cluster control region for the protocadherin-beta genes located beyond the protocadherin-gamma cluster. J Biol Chem. 2011286:31885–95.

Phillips JE, Corces VG. CTCF: master weaver of the genome. Cell. 2009137:1194–211.

Hirayama T, Tarusawa E, Yoshimura Y, Galjart N, Yagi T. CTCF is required for neural development and stochastic expression of clustered Pcdh genes in neurons. Cell Rep. 20122:345–57.

Victoria-Acosta G, Vazquez-Santillan K, Jimenez-Hernandez L, Munoz-Galindo L, Maldonado V, Martinez-Ruiz GU, et al. Epigenetic silencing of the XAF1 gene is mediated by the loss of CTCF binding. Sci Rep. 20155:14838.

Walker CJ, Miranda MA, O’Hern MJ, McElroy JP, Coombes KR, Bundschuh R, et al. Patterns of CTCF and ZFHX3 mutation and associated outcomes in endometrial cancer. J Natl Cancer Inst. 2015107.

Katainen R, Dave K, Pitkanen E, Palin K, 0000000246216128 AO, Kivioja T, et al. CTCF/cohesin-binding sites are frequently mutated in cancer. Nat Genet. 201547:818–21.

Zou Y, Huang MZ, Liu FY, Yang BC, Wang LQ, Wang F, et al. Absence of and hotspot mutations in patients with various subtypes of ovarian carcinomas. Biomed Rep. 20153:33–7.

Remeseiro S, Cuadrado A, Gomez-Lopez G, Pisano DG, Losada A. A unique role of cohesin-SA1 in gene regulation and development. EMBO J. 201231:2090–102.

Monahan K, Rudnick ND, Kehayova PD, Pauli F, Newberry KM, Myers RM, et al. Role of CCCTC binding factor (CTCF) and cohesin in the generation of single-cell diversity of protocadherin-alpha gene expression. Proc Natl Acad Sci U S A. 2012109:9125–30.

Tan YP, Li S, Jiang XJ, Loh W, Foo YK, Loh CB, et al. Regulation of protocadherin gene expression by multiple neuron-restrictive silencer elements scattered in the gene cluster. Nucleic Acids Res. 201038:4985–97.

Kehayova P, Monahan K, Chen W, Maniatis T. Regulatory elements required for the activation and repression of the protocadherin-alpha gene cluster. Proc Natl Acad Sci U S A. 2011108:17195–200.

Majumder S. REST in good times and bad: roles in tumor suppressor and oncogenic activities. Cell Cycle. 20065:1929–35.

Wagoner MP, Gunsalus KT, Schoenike B, Richardson AL, Friedl A, Roopra A. The transcription factor REST is lost in aggressive breast cancer. PLoS Genet. 20106:e1000979.

Huang Z, Bao S. Ubiquitination and deubiquitination of REST and its roles in cancers. FEBS Lett. 2012586:1602–5.

Robertson KD, Wolffe AP. DNA methylation in health and disease. Nat Rev Genet. 20001:11–9.

Kawaguchi M, Toyama T, Kaneko R, Hirayama T, Kawamura Y, Yagi T. Relationship between DNA methylation states and transcription of individual isoforms encoded by the protocadherin-alpha gene cluster. J Biol Chem. 2008283:12064–75.

Toyoda S, Kawaguchi M, Kobayashi T, Tarusawa E, Toyama T, Okano M, et al. Developmental epigenetic modification regulates stochastic expression of clustered protocadherin genes, generating single neuron diversity. Neuron. 201482:94–108.

Novak P, Jensen T, Oshiro MM, Watts GS, Kim CJ, Futscher BW. Agglomerative epigenetic aberrations are a common event in human breast cancer. Cancer Res. 200868:8616–25.

Miyamoto K, Fukutomi T, Akashi-Tanaka S, Hasegawa T, Asahara T, Sugimura T, et al. Identification of 20 genes aberrantly methylated in human breast cancers. Int J Cancer. 2005116:407–14.

Jones S, Zhang X, Parsons DW, Lin JC, Leary RJ, Angenendt P, et al. Core signaling pathways in human pancreatic cancers revealed by global genomic analyses. Science. 2008321:1801–6.

Vincent A, Omura N, Hong SM, Jaffe A, Eshleman J, Goggins M. Genome-wide analysis of promoter methylation associated with gene expression profile in pancreatic adenocarcinoma. Clin Cancer Res. 201117:4341–54.

Waha A, Guntner S, Huang TH, Yan PS, Arslan B, Pietsch T, et al. Epigenetic silencing of the protocadherin family member PCDH-gamma-A11 in astrocytomas. Neoplasia. 20057:193–9.

Shi H, Guo J, Duff DJ, Rahmatpanah F, Chitima-Matsiga R, Al-Kuhlani M, et al. Discovery of novel epigenetic markers in non-Hodgkin’s lymphoma. Carcinogenesis. 200728:60–70.

Dallosso AR, Hancock AL, Szemes M, Moorwood K, Chilukamarri L, Tsai HH, et al. Frequent long-range epigenetic silencing of protocadherin gene clusters on chromosome 5q31 in Wilms’ tumor. PLoS Genet. 20095:e1000745.

Dallosso AR, Oster B, Greenhough A, Thorsen K, Curry TJ, Owen C, et al. Long-range epigenetic silencing of chromosome 5q31 protocadherins is involved in early and late stages of colorectal tumorigenesis through modulation of oncogenic pathways. Oncogene. 201231:4409–19.

Severson PL, Tokar EJ, Vrba L, Waalkes MP, Futscher BW. Agglomerates of aberrant DNA methylation are associated with toxicant-induced malignant transformation. Epilepsia. 20127:1238–48.

Yu JS, Koujak S, Nagase S, Li CM, Su T, Wang X, et al. PCDH8, the human homolog of PAPC, is a candidate tumor suppressor of breast cancer. Oncogene. 200827:4657–65.

Zhang D, Zhao W, Liao X, Bi T, Li H, Che X. Frequent silencing of protocadherin 8 by promoter methylation, a candidate tumor suppressor for human gastric cancer. Oncol Rep. 201228:1785–91.

Lin YL, Wang YL, Ma JG, Li WP. Clinical significance of protocadherin 8 (PCDH8) promoter methylation in non-muscle invasive bladder cancer. J Exp Clin Cancer Res. 201433:68.

Leshchenko VV, Kuo PY, Shaknovich R, Yang DT, Gellen T, Petrich A, et al. Genomewide DNA methylation analysis reveals novel targets for drug development in mantle cell lymphoma. Blood. 2010116:1025–34.

Yu B, Yang H, Zhang C, Wu Q, Shao Y, Zhang J, et al. High-resolution melting analysis of PCDH10 methylation levels in gastric, colorectal and pancreatic cancers. Neoplasma. 201057:247–52.

Yu J, Cheng YY, Tao Q, Cheung KF, Lam CN, Geng H, et al. Methylation of protocadherin 10, a novel tumor suppressor, is associated with poor prognosis in patients with gastric cancer. Gastroenterology. 2009136:640–51.e1.

Heitzer E, Artl M, Filipits M, Resel M, Graf R, Weissenbacher B, et al. Differential survival trends of stage II colorectal cancer patients relate to promoter methylation status of PCDH10, SPARC, and UCHL1. Mod Pathol. 201427:906–15.

Danese E, Minicozzi AM, Benati M, Montagnana M, Paviati E, Salvagno GL, et al. Epigenetic alteration: new insights moving from tissue to plasma—the example of PCDH10 promoter methylation in colorectal cancer. Br J Cancer. 2013109:807–13.

Deng J, Liang H, Ying G, Dong Q, Zhang L, Yu J, et al. Clinical significance of the methylated cytosine-phosphate-guanine sites of protocadherin-10 promoter for evaluating the prognosis of gastric cancer. J Am Coll Surg. 2014219:904–13.

Ying J, Li H, Seng TJ, Langford C, Srivastava G, Tsao SW, et al. Functional epigenetics identifies a protocadherin PCDH10 as a candidate tumor suppressor for nasopharyngeal, esophageal and multiple other carcinomas with frequent methylation. Oncogene. 200625:1070–80.

Narayan G, Scotto L, Neelakantan V, Kottoor SH, Wong AH, Loke SL, et al. Protocadherin PCDH10, involved in tumor progression, is a frequent and early target of promoter hypermethylation in cervical cancer. Genes Chromosom Cancer. 200948:983–92.

Harada H, Miyamoto K, Yamashita Y, Taniyama K, Mihara K, Nishimura M, et al. Prognostic signature of protocadherin 10 methylation in curatively resected pathological stage I non-small-cell lung cancer. Cancer Med. 2015.

Cheung HH, Lee TL, Davis AJ, Taft DH, Rennert OM, Chan WY. Genome-wide DNA methylation profiling reveals novel epigenetically regulated genes and non-coding RNAs in human testicular cancer. Br J Cancer. 2010102:419–27.

Ying J, Gao Z, Li H, Srivastava G, Murray PG, Goh HK, et al. Frequent epigenetic silencing of protocadherin 10 by methylation in multiple haematologic malignancies. Br J Haematol. 2007136:829–32.

Haruki S, Imoto I, Kozaki K, Matsui T, Kawachi H, Komatsu S, et al. Frequent silencing of protocadherin 17, a candidate tumour suppressor for esophageal squamous cell carcinoma. Carcinogenesis. 201031:1027–36.

Hu X, Sui X, Li L, Huang X, Rong R, Su X, et al. Protocadherin 17 acts as a tumour suppressor inducing tumour cell apoptosis and autophagy, and is frequently methylated in gastric and colorectal cancers. J Pathol. 2013229:62–73.

Imoto I, Izumi H, Yokoi S, Hosoda H, Shibata T, Hosoda F, et al. Frequent silencing of the candidate tumor suppressor PCDH20 by epigenetic mechanism in non-small-cell lung cancers. Cancer Res. 200666:4617–26.

Gendrel AV, Tang YA, Suzuki M, Godwin J, Nesterova TB, Greally JM, et al. Epigenetic functions of smchd1 repress gene clusters on the inactive X chromosome and on autosomes. Mol Cell Biol. 201333:3150–65.

Chen K, Hu J, Moore DL, Liu R, Kessans SA, Breslin K, et al. Genome-wide binding and mechanistic analyses of Smchd1-mediated epigenetic regulation. Proc Natl Acad Sci U S A 2015.

Leong HS, Chen K, Hu Y, Lee S, Corbin J, Pakusch M, et al. Epigenetic regulator Smchd1 functions as a tumor suppressor. Cancer Res. 201373:1591–9.

Bamford S, Dawson E, Forbes S, Clements J, Pettett R, Dogan A, et al. The COSMIC (catalogue of somatic mutations in cancer) database and website. Br J Cancer. 200491:355–8.

Cerami E, Gao J, Dogrusoz U, Gross BE, Sumer SO, Aksoy BA, et al. The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data. Cancer Discov. 20122:401–4.

Marambaud P, Wen PH, Dutt A, Shioi J, Takashima A, Siman R, et al. A CBP binding transcriptional repressor produced by the PS1/epsilon-cleavage of N-cadherin is inhibited by PS1 FAD mutations. Cell. 2003114:635–45.

Maretzky T, Reiss K, Ludwig A, Buchholz J, Scholz F, Proksch E, et al. ADAM10 mediates E-cadherin shedding and regulates epithelial cell-cell adhesion, migration, and beta-catenin translocation. Proc Natl Acad Sci U S A. 2005102:9182–7.

Marambaud P, Robakis NK. Genetic and molecular aspects of Alzheimer’s disease shed light on new mechanisms of transcriptional regulation. Genes Brain Behav. 20054:134–46.

Bonn S, Seeburg PH, Schwarz MK. Combinatorial expression of alpha- and gamma-protocadherins alters their presenilin-dependent processing. Mol Cell Biol. 200727:4121–32.

Asad M, Wong MK, Tan TZ, Choolani M, Low J, Mori S, et al. FZD7 drives in vitro aggressiveness in Stem-A subtype of ovarian cancer via regulation of non-canonical Wnt/PCP pathway. Cell Death Dis. 20145:e1346.

Wang C, Tao B, Li S, Li B, Wang X, Hu G, et al. Characterizing the role of PCDH9 in the regulation of glioma cell apoptosis and invasion. J Mol Neurosci. 201452:250–60.

Zhu P, Lv J, Yang Z, Guo L, Zhang L, Li M, et al. Protocadherin 9 inhibits epithelial-mesenchymal transition and cell migration through activating GSK-3beta in hepatocellular carcinoma. Biochem Biophys Res Commun. 2014452:567–74.

Xu Y, Yang Z, Yuan H, Li Z, Li Y, Liu Q, et al. PCDH10 inhibits cell proliferation of multiple myeloma via the negative regulation of the Wnt/beta-catenin/BCL-9 signaling pathway. Oncol Rep. 201534:747–54.

Jao TM, Tsai MH, Lio HY, Weng WT, Chen CC, Tzeng ST, et al. Protocadherin 10 suppresses tumorigenesis and metastasis in colorectal cancer and its genetic loss predicts adverse prognosis. Int J Cancer. 2014135:2593–603.

Zhao Y, Yang Y, Trovik J, Sun K, Zhou L, Jiang P, et al. A novel wnt regulatory axis in endometrioid endometrial cancer. Cancer Res. 201474:5103–17.

Chen T, Long B, Ren G, Xiang T, Li L, Wang Z, et al. Protocadherin20 acts as a tumor suppressor gene: epigenetic inactivation in nasopharyngeal carcinoma. J Cell Biochem. 2015.

Wu, JC, Wang FZ, Tsai ML, Lo CY, Badmaev V, Ho CT, et al. Se-Allylselenocysteine induces autophagy by modulating the AMPK/mTOR signaling pathway and epigenetic regulation of PCDH17 in human colorectal adenocarcinoma cells. Mol Nutr Food Res 2015.

Dang Z, Shangguan J, Zhang C, Hu P, Ren Y, Lv Z, et al. Loss of protocadherin-17 (PCDH-17) promotes metastasis and invasion through hyperactivation of EGFR/MEK/ERK signaling pathway in hepatocellular carcinoma. Tumour Biol 2015.

Chen MW, Vacherot F, De La Taille A, Gil-Diez-De-Medina S, Shen R, Friedman RA, et al. The emergence of protocadherin-PC expression during the acquisition of apoptosis-resistance by prostate cancer cells. Oncogene. 200221:7861–71.

Terry S, Maille P, Baaddi H, Kheuang L, Soyeux P, Nicolaiew N, et al. Cross modulation between the androgen receptor axis and protocadherin-PC in mediating neuroendocrine transdifferentiation and therapeutic resistance of prostate cancer. Neoplasia. 201315:761–72.

Terry S, Queires L, Gil-Diez-de-Medina S, Chen MW, de la Taille A, Allory Y, et al. Protocadherin-PC promotes androgen-independent prostate cancer cell growth. Prostate. 200666:1100–13.

Yang X, Chen MW, Terry S, Vacherot F, Chopin DK, Bemis DL, et al. A human- and male-specific protocadherin that acts through the wnt signaling pathway to induce neuroendocrine transdifferentiation of prostate cancer cells. Cancer Res. 200565:5263–71.

Bos PD, Zhang XH, Nadal C, Shu W, Gomis RR, Nguyen DX, et al. Genes that mediate breast cancer metastasis to the brain. Nature. 2009459:1005–9.

Li AM, Tian AX, Zhang RX, Ge J, Sun X, Cao XC. Protocadherin-7 induces bone metastasis of breast cancer. Biochem Biophys Res Commun. 2013436:486–90.

Du W, Liu X, Fan G, Zhao X, Sun Y, Wang T, et al. From cell membrane to the nucleus: an emerging role of E-cadherin in gene transcriptional regulation. J Cell Mol Med. 201418:1712–9.

Han MH, Lin C, Meng S, Wang X. Proteomics analysis reveals overlapping functions of clustered protocadherins. Mol Cell Proteomics. 20109:71–83.

Wilker EW, van Vugt MA, Artim SA, Huang PH, Petersen CP, Reinhardt HC, et al. 14-3-3sigma controls mitotic translation to facilitate cytokinesis. Nature. 2007446:329–32.

Urano T, Saito T, Tsukui T, Fujita M, Hosoi T, Muramatsu M, et al. Efp targets 14-3-3 sigma for proteolysis and promotes breast tumour growth. Nature. 2002417:871–5.

Kohmura N, Senzaki K, Hamada S, Kai N, Yasuda R, Watanabe M, et al. Diversity revealed by a novel family of cadherins expressed in neurons at a synaptic complex. Neuron. 199820:1137–51.

Zhang B, Wang J, Wang X, Zhu J, Liu Q, Shi Z, et al. Proteogenomic characterization of human colon and rectal cancer. Nature. 2014513:382–7.

Vanhalst K, Kools P, Staes K, van Roy F, Redies C. Delta-Protocadherins: a gene family expressed differentially in the mouse brain. Cell Mol Life Sci. 200562:1247–59.

Heggem MA, Bradley RS. The cytoplasmic domain of Xenopus NF-protocadherin interacts with TAF1/set. Dev Cell. 20034:419–29.

Yasuda S, Tanaka H, Sugiura H, Okamura K, Sakaguchi T, Tran U, et al. Activity-induced protocadherin arcadlin regulates dendritic spine number by triggering N-cadherin endocytosis via TAO2beta and p38 MAP kinases. Neuron. 200756:456–71.

Nakao S, Platek A, Hirano S, Takeichi M. Contact-dependent promotion of cell migration by the OL-protocadherin-Nap1 interaction. J Cell Biol. 2008182:395–410.

Grove EA. Turning neurons into a nervous system. Development. 2008135:2203–6.

Homayouni R, Rice DS, Curran T. Disabled-1 interacts with a novel developmentally regulated protocadherin. Biochem Biophys Res Commun. 2001289:539–47.

Borrell V, Pujadas L, Simo S, Dura D, Sole M, Cooper JA, et al. Reelin and mDab1 regulate the development of hippocampal connections. Mol Cell Neurosci. 200736:158–73.

Kahr I, Vandepoele K, van Roy F. Delta-protocadherins in health and disease. Prog Mol Biol Transl Sci. 2013116:169–92.

Kietzmann A, Wang Y, Weber D, Steinbeisser H. Xenopus paraxial protocadherin inhibits Wnt/beta-catenin signalling via casein kinase 2beta. EMBO Rep. 201213:129–34.

Kai M, Ueno N, Kinoshita N. Phosphorylation-dependent ubiquitination of paraxial protocadherin (PAPC) controls gastrulation cell movements. PLoS One. 201510:e0115111.

Schalm SS, Ballif BA, Buchanan SM, Phillips GR, Maniatis T. Phosphorylation of protocadherin proteins by the receptor tyrosine kinase Ret. Proc Natl Acad Sci U S A. 2010107:13894–9.

Santoro M, Carlomagno F, Romano A, Bottaro DP, Dathan NA, Grieco M, et al. Activation of RET as a dominant transforming gene by germline mutations of MEN2A and MEN2B. Science. 1995267:381–3.

Fernandez-Sanchez ME, Barbier S, Whitehead J, Bealle G, Michel A, Latorre-Ossa H, et al. Mechanical induction of the tumorigenic beta-catenin pathway by tumour growth pressure. Nature. 2015523:92–5.

Faura TG, Vandepoele K, Brouwer U, Koning H, Elderman RM, Hackett TL, et al. Protocadherin-1 binds to SMAD3 and suppresses TGFbeta1-induced gene transcription. Am J Physiol Lung Cell Mol Physiol. 2015:ajplung.00346.2014.

Wu C, Niu L, Yan Z, Wang C, Liu N, Dai Y, et al. Pcdh11x negatively regulates dendritic branching. J Mol Neurosci. 201556:822–8.

Takeichi M. Morphogenetic roles of classic cadherins. Curr Opin Cell Biol. 19957:619–27.

Chen X, Gumbiner BM. Paraxial protocadherin mediates cell sorting and tissue morphogenesis by regulating C-cadherin adhesion activity. J Cell Biol. 2006174:301–13.

Kazmierczak P, Sakaguchi H, Tokita J, Wilson-Kubalek EM, Milligan RA, Muller U, et al. Cadherin 23 and protocadherin 15 interact to form tip-link filaments in sensory hair cells. Nature. 2007449:87–91.

Chen X, Koh E, Yoder M, Gumbiner BM. A protocadherin-cadherin-FLRT3 complex controls cell adhesion and morphogenesis. PLoS One. 20094:e8411.

Kraft B, Berger CD, Wallkamm V, Steinbeisser H, Wedlich D. Wnt-11 and Fz7 reduce cell adhesion in convergent extension by sequestration of PAPC and C-cadherin. J Cell Biol. 2012198:695–709.

Jung B, Kohler A, Schambony A, Wedlich D. PAPC and the Wnt5a/Ror2 pathway control the invagination of the otic placode in Xenopus. BMC Dev Biol. 201111:36.

Himmelreich N, Kaufmann, LT, Steinbeisser H, Korner C, Thiel C. Lack of phosphomannomutase 2 affects Xenopus laevis morphogenesis and the non-canonical Wnt5a/Ror2 signalling. J Inherit Metab Dis. 2015.

Kozu Y, Gon Y, Maruoka S, Kazumichi K, Sekiyama A, Kishi H, et al. Protocadherin-1 is a glucocorticoid-responsive critical regulator of airway epithelial barrier function. BMC Pulm Med. 201515:80.

Chen J, Lu Y, Meng S, Han MH, Lin C, Wang X. Alpha- and gamma-protocadherins negatively regulate PYK2. J Biol Chem. 2009284:2880–90.

Suo L, Lu H, Ying G, Capecchi MR, Wu Q. Protocadherin clusters and cell adhesion kinase regulate dendrite complexity through Rho GTPase. J Mol Cell Biol. 20124:362–76.

Zhang X, Chattopadhyay A, Ji QS, Owen JD, Ruest PJ, Carpenter G, et al. Focal adhesion kinase promotes phospholipase C-gamma1 activity. Proc Natl Acad Sci U S A. 199996:9021–6.

Tvorogov D, Wang XJ, Zent R, Carpenter G. Integrin-dependent PLC-gamma1 phosphorylation mediates fibronectin-dependent adhesion. J Cell Sci. 2005118:601–10.

Heemskerk FM, Chen HC, Huang FL. Protein kinase C phosphorylates Ser152, Ser156 and Ser163 but not Ser160 of MARCKS in rat brain. Biochem Biophys Res Commun. 1993190:236–41.

Li H, Chen G, Zhou B, Duan S. Actin filament assembly by myristoylated alanine-rich C kinase substrate-phosphatidylinositol-4,5-diphosphate signaling is critical for dendrite branching. Mol Biol Cell. 200819:4804–13.

Keeler AB, Schreiner D, Weiner JA, Protein Kinase C. Phosphorylation of a gamma-protocadherin C-terminal lipid binding domain regulates focal adhesion kinase inhibition and dendrite arborization. J Biol Chem. 2015290:20674–86.

Biswas S, Emond MR, Duy PQ, Hao lT, Beattie CE, Jontes JD. Protocadherin-18b interacts with Nap1 to control motor axon growth and arborization in zebrafish. Mol Biol Cell. 201425:633–42.

Takenawa T, Suetsugu S. The WASP-WAVE protein network: connecting the membrane to the cytoskeleton. Nat Rev Mol Cell Biol. 20078:37–48.

Kurisu S, Takenawa T. WASP and WAVE family proteins: friends or foes in cancer invasion. Cancer Sci. 2010101:2093–104.

Hayashi S, Inoue Y, Kiyonari H, Abe T, Misaki K, Moriguchi H, et al. Protocadherin-17 mediates collective axon extension by recruiting actin regulator complexes to interaxonal contacts. Dev Cell. 201430:673–87.

Hoshina N, Tanimura A, Yamasaki M, Inoue T, Fukabori R, Kuroda T, et al. Protocadherin 17 regulates presynaptic assembly in topographic corticobasal ganglia circuits. Neuron. 201378:839–54.

Brigidi GS, Bamji SX. Cadherin-catenin adhesion complexes at the synapse. Curr Opin Neurobiol. 201121:208–14.

Biswas S, Emond MR, Jontes JD. Protocadherin-19 and N-cadherin interact to control cell movements during anterior neurulation. J Cell Biol. 2010191:1029–41.

Morrow EM, Yoo SY, Flavell SW, Kim TK, Lin Y, Hill RS, et al. Identifying autism loci and genes by tracing recent shared ancestry. Science. 2008321:218–23.

Tsai NP, Wilkerson JR, Guo W, Maksimova MA, DeMartino GN, Cowan CW, et al. Multiple autism-linked genes mediate synapse elimination via proteasomal degradation of a synaptic scaffold PSD-95. Cell. 2012151:1581–94.

Bassermann F, Eichner R, Pagano M. The ubiquitin proteasome system - implications for cell cycle control and the targeted treatment of cancer. Biochim Biophys Acta. 20141843:150–62.

Schaefer A, Nethe M, Hordijk PL. Ubiquitin links to cytoskeletal dynamics, cell adhesion and migration. Biochem J. 2012442:13–25.

Haas IG, Frank M, Veron N, Kemler R. Presenilin-dependent processing and nuclear function of gamma-protocadherins. J Biol Chem. 2005280:9313–9.

Trimarchi T, Bilal E, Ntziachristos P, Fabbri G, Dalla-Favera R, Tsirigos A, et al. Genome-wide mapping and characterization of Notch-regulated long noncoding RNAs in acute leukemia. Cell. 2014158:593–606.

Keith B, Simon MC. Hypoxia-inducible factors, stem cells, and cancer. Cell. 2007129:465–72.

Racioppi L, Means AR. Calcium/calmodulin-dependent protein kinase kinase 2: roles in signaling and pathophysiology. J Biol Chem. 2012287:31658–65.

Colomer J, Means AR. Physiological roles of the Ca2+/CaM-dependent protein kinase cascade in health and disease. Subcell Biochem. 200745:169–214.

Gardner HP, Ha SI, Reynolds C, Chodosh LA. The caM kinase, Pnck, is spatially and temporally regulated during murine mammary gland development and may identify an epithelial cell subtype involved in breast cancer. Cancer Res. 200060:5571–7.

Takai N, Ueda T, Nasu K, Yamashita S, Toyofuku M, Narahara H. Targeting calcium/calmodulin-dependence kinase I and II as a potential anti-proliferation remedy for endometrial carcinomas. Cancer Lett. 2009277:235–43.

Onouchi T, Kishino-Kaneko Y, Kameshita I, Ishida A, Sueyoshi N. Regulation of Ca/calmodulin-dependent protein kinase phosphatase (CaMKP/PPM1F) by protocadherin-gammaC5 (Pcdh-gammaC5). Arch Biochem Biophys 2015.

ELife digest

Multicellular life depends on cells being able to stick together. The human body, for example, consists of trillions of cells grouped into tissues and organs. The brain alone contains some 87 billion neurons organized into complex networks. To stay together, cells use proteins on their surface called cell adhesion molecules (CAMs). There are four major families of CAMs, each with multiple members, and the CAMs on one cell recognize and interact with the CAMs on another.

But how does this process work? One possibility is that different combinations of CAMs allow different cells to stick together. Bisogni et al. tested this idea by studying a family of CAMs called the delta-protocadherins. This family has nine members, each with its own gene. Before cells can use a gene to produce a protein, they must first use the gene’s DNA as a template to build an RNA molecule. By counting the number of different types of RNA molecules inside individual cells, Bisogni et al. showed that sensory neurons in the mouse each produce up to seven different delta-protocadherins.

Further experiments revealed that cells fine-tune their interactions by varying the number, type and combination of delta-protocadherins on their surface. In addition, the delta-protocadherins also alter interactions between members of a related gene family, the clustered protocadherins. This further increases their ability to regulate how cells interact.

In contrast to previous studies that focused on single molecules, Bisogni et al. have shown how combinations of molecules work together to influence cell adhesion. Deciphering this combinatorial code is key to understanding how interactions between cells go awry in disease.

Mutations in the genes for CAMs often impair brain development. The reported findings may provide insights into how such mutations disrupt the CAM combinatorial code and alter cell to cell interactions.

Tracing the dynamics of gene transcripts after organismal death

In life, genetic and epigenetic networks precisely coordinate the expression of genes—but in death, it is not known if gene expression diminishes gradually or abruptly stops or if specific genes and pathways are involved. We studied this by identifying mRNA transcripts that apparently increase in relative abundance after death, assessing their functions, and comparing their abundance profiles through postmortem time in two species, mouse and zebrafish. We found mRNA transcript profiles of 1063 genes became significantly more abundant after death of healthy adult animals in a time series spanning up to 96 h postmortem. Ordination plots revealed non-random patterns in the profiles by time. While most of these transcript levels increased within 0.5 h postmortem, some increased only at 24 and 48 h postmortem. Functional characterization of the most abundant transcripts revealed the following categories: stress, immunity, inflammation, apoptosis, transport, development, epigenetic regulation and cancer. The data suggest a step-wise shutdown occurs in organismal death that is manifested by the apparent increase of certain transcripts with various abundance maxima and durations.

1. Introduction

A healthy adult vertebrate is a complex biological system capable of highly elaborate functions such as the ability to move, communicate and sense the environment—all at the same time. These functions are tightly regulated by genetic and epigenetic networks through multiple feedback loops that precisely coordinate the expression of thousands of genes at the right time, in the right place and in the right level [1]. Together, these networks maintain homeostasis and thus sustain ‘life’ of a biological system.

While much is known about gene expression circuits in life, there is a paucity of information about what happens to these circuits after organismal death. For example, it is not well known whether gene expression diminishes gradually or abruptly stops in death—nor whether specific gene transcripts increase in abundance in death. In organismal ‘death’, defined here as the cessation of the highly elaborate system functions in vertebrates, we conjecture that there is a gradual disengagement and loss of global regulatory networks as well as the activation of regulatory genes involved in survival and stress compensation. To test this, we examined the global postmortem abundances of mRNAs in two model organisms: the zebrafish, Danio rerio, and the house mouse, Mus musculus. The purpose of the research was to investigate the ‘unwinding of the clock’ by identifying mRNA transcripts that increase in abundance with postmortem time and assessing their functions based on the primary literature. The biological systems investigated in this study are different from those examined in other studies, such as individual dead and/or injured cells in live organisms, i.e. apoptosis and necrosis (reviewed in [2–5]). In contrast to previous studies, the abundances of mRNA transcripts from the entire D. rerio body, and the brains and livers of M. musculus were assessed through postmortem time. The mRNA transcripts were measured using the ‘Gene Meter’ approach that precisely reports transcript abundances based on a calibration curve for each microarray probe [6–9].

2. Material and methods

2.1. Induced death and postmortem incubation

2.1.1. Zebrafish

Forty-four female Danio rerio were transferred from several flow-through aquaria kept at 28°C to a glass beaker containing 1 l of aquarium water. Four individuals were immediately taken out, snap frozen in liquid nitrogen and stored in Falcon tubes at −80°C (two zebrafish per tube). These samples were designated as the first set of live controls. A second set of live controls was immersed in an open cylinder (described below). Two sets of live controls were used to determine whether putting the zebrafish back into their native environment had any effects on gene expression (we later discovered no significant effects).

The rest of the zebrafish were subjected to sudden death by immersion in a ‘kill’ chamber. The chamber consisted of an 8 l styrofoam container filled with chilled ice water. To synchronize the death of the rest of the zebrafish, they were transferred to an open cylinder with a mesh-covered bottom and the cylinder was immersed into the kill chamber. After 20–30 s of immersion, four zebrafish were retrieved from the chamber, snap frozen in liquid nitrogen and stored at −80°C (two zebrafish per Falcon tube). These samples were designated as the second set of live controls. The remaining zebrafish were kept in the kill chamber for 5 min and then the cylinder was transferred to a flow-through aquarium kept at 28°C so that they were returned to their native environment.

Postmortem sampling of the zebrafish occurred at: time 0, 15 min, 30 min, 1 h, 4 h, 8 h, 12 h, 24 h, 48 h and 96 h. For each sampling time, four expired zebrafish were retrieved from the cylinder, snap frozen in liquid nitrogen and stored at −80°C in Falcon tubes (two zebrafish to a tube). One zebrafish sample was lost, but extraction volumes were adjusted to one individual.

2.1.2. Mouse

The mouse strain C57BL/6JRj (Janvier SAS, France) was used for our experiments. The mice were 20-week old males of approximately the same weight. The mice were highly inbred and were expected to have a homogeneous genetic background. Prior to euthanasia, the mice were kept at room temperature and were given ad libitum access to food and water. Each mouse was euthanized by cervical dislocation and placed in an individual plastic bag with holes to allow air/gas exchange. The bagged carcasses were kept at room temperature in a large, open polystyrene container. Sampling of the deceased mice began at 0 h (postmortem time zero) and continued at 30 min, 1 h, 6 h, 12 h, 24 h and 48 h postmortem. At each sample time, three mice were sampled (except for 48 h when two mice were sampled) and the entire brain (plus stem) and two portions of the liver were extracted from each mouse. For liver samples, clippings were taken from the foremost and rightmost lobes of the liver. The brain and liver samples were snap frozen in liquid nitrogen and stored individually in Falcon tubes at −80°C.

2.2. RNA extraction, labelling, hybridization and DNA microarrays

The number of individuals was 43 for zebrafish and 20 for mice. Samples from two fish were pooled for analysis, resulting in two replicate measurements at each time point. The number of replicated measurements for mice was three at each of the first six time points and two at 48 h. Thus, the total number of samples analysed was 22 for zebrafish and 20 for mice. For the zebrafish, samples were mixed with 20 ml of Trizol and homogenized using a TissueLyzer (Qiagen). For the mice, 100 mg of brain or liver samples were mixed with 1 ml of Trizol and homogenized. One millilitre of the emulsion from each sample was put into a fresh 1.5 ml centrifuge tube for RNA extraction and the rest was frozen at −80°C.

RNA was extracted by adding 200 µl of chloroform, vortexing the sample and incubating it at 25°C for 3 min. After centrifugation (15 min at 12 000g at 4°C), the supernatant (approx. 350 µl) was transferred to a fresh 1.5 ml tube containing an equal volume of 70% ethanol. The tube was vortexed, centrifuged and purified following the procedures outlined in the PureLink RNA Mini Kit (Life Technologies, USA).

The isolated RNA, 400 ng per sample, was labelled, purified and hybridized according to the One-Color Microarray-based Gene Expression Analysis (Quick Amp Labeling) with Tecan HS Pro Hybridization kit (Agilent Technologies). For the zebrafish, the labelled RNA was hybridized to the Zebrafish (v2) Gene Expression Microarray (Design ID 019161). For the mouse, the labelled RNA was hybridized to the SurePrint G3 Mouse GE 8×60K Microarray Design ID 028005 (Agilent Technologies). The microarrays were loaded with 1.65 µg of labelled cRNA for each postmortem sample.

2.3. Microarray calibration

Oligonucleotide (60 nt) probes on the zebrafish and mouse microarrays were calibrated using pooled labelled cRNA of all zebrafish and all mouse postmortem samples, respectively. The dilution series for the Zebrafish array was created using the following concentrations of labelled cRNA: 0.41, 0.83, 1.66, 1.66, 1.66, 3.29, 6.60 and 8.26 µg. The dilution series for the Mouse arrays was created using the following concentrations of labelled cRNA: 0.17, 0.33, 0.66, 1.32, 2.64, 5.28, 7.92 and 10.40 µg. Calibration involved plotting the signal intensities of the probes against a dilution factor and determining the isotherm model (e.g. Freundlich and/or Langmuir) that best fit the relationship between signal intensities and gene abundances.

Consider zebrafish gene transcripts targeted by A_15_P110618 (which happens to be one of the transcriptional profiles of gene Hsp70.3 shown in figure 1a). External file FishProbesParameters.txt shows that a Freundlich model best fit the dilution curve with R 2 = 0.99. The equation for this probe is the following:

where SI is the observed average signal intensity for the dilution x. The transcript abundance G was calculated by inverting this equation. For each probe signal intensity at postmortem time, SIt, the gene abundance G= (SIt/exp(7.1081)) 1/0.67632 . Specifically, consider two biological replicates of 15 min postmortem zebrafish, the signal intensities of the probe A_15_P110618 are 770.5 and 576, which translates into the abundances 0.50 and 0.33 arbitrary units (arb. units), respectively. The target abundances were further converted to log10 and are shown in external file Fish_log10_AllProfiles.txt.

Figure 1. Transcriptional profiles of representative genes (arb. units), ordination plots based on transcript abundances by postmortem time (h) with corresponding transcript contributions (biplots), and averaged transcript abundances by group. (a–c) Transcriptional profiles of (a) the Hsp70.3 gene, (b) the Tox2 gene and (c) a non-annotated transcript ‘NULL’ (i.e. no annotation, probe number shown) gene as a function of postmortem time. (d,e) Ordination plots of the (d) zebrafish and (e) mouse were based on all gene transcript profiles found to have a significantly increased abundance. Gene transcripts in the biplots were arbitrarily assigned alphabetical groups based on their positions in the ordination. The average transcript abundances for each group are shown.

Further details of the calibration protocols used to calculate RNA transcript relative abundances are provided elsewhere [6,7].

2.4. Statistical analysis

Abundance levels were log-transformed for analysis to stabilize the variance. A one-sided Dunnett's T-statistic was applied to test for increase at one or more postmortem times compared to live control (fish) or time 0 (mouse). A bootstrap procedure with 10 9 simulations was used to determine the critical value for the Dunnett's statistics in order to accommodate departures from parametric assumptions and to account for multiplicity of testing. The transcript profile for each gene was centred by subtracting the mean values at each postmortem time point to create ‘null’ profiles. Bootstrap samples of the null profiles were generated to determine the 95th percentile of the maximum (over all genes) of the Dunnett's statistics. A transcript was considered to have a significantly increased abundance when one or more points had Dunnett's T-values larger than the 95th percentile. The corresponding genes were retained for further analyses.

Orthogonal transformation of the abundances to their principal components (PCs) was conducted, and the results were graphed on a two-dimensional ordination plot. The m ×n matrix of abundances (sampling times by number of gene transcripts), which is 10 × 548 for zebrafish and 7 × 515 for mouse, was used to produce an m × m matrix D of Euclidean distances between all pairs of sampling times. Principal component analysis (PCA) was performed on the matrix of distances, D. To investigate and visualize differences between the sampling times, a scatterplot of the first two principal components (PC1 and PC2) was created. To establish relative contributions of the gene transcripts, the projection of each sampling time onto the (PC1 and PC2) plane was calculated and those gene transcripts with high correlations (greater than or equal to 0.70) between abundances and either component (PC1 or PC2) were displayed as a biplot.

2.5. Gene annotation and functional categorization

Microarray probe sequences were individually annotated by performing a BLASTN search of the zebrafish and mouse NCBI databases (February 2015). The gene annotations were retained if the bit score was greater than or equal to 100 and the annotations were in the correct 5′–3′ orientation. Transcription factors, transcriptional regulators and cell signalling components (e.g. receptors, enzymes and messengers) were identified as global regulatory genes. The rest were considered response genes.

Functional categorizations were performed by querying the annotated gene transcripts in the primary literature and using UniProt ( Genes not functionally categorized to their native organism (zebrafish or mouse) were categorized to genes of phylogenetically related organisms (e.g. human). Cancer-related genes were identified using a previously constructed database (see Additional File 1: table S1 in [10]).

3. Results

Extracting the total mRNA, calibrating the microarray probes, and determining the transcript abundances at each postmortem sampling time produced a fine-grain series of transcriptome data for the zebrafish and the mouse. Approximately 84.3% (36 811 of 43 663) zebrafish probes and 67.1% (37 368 of 55 681) mouse probes were found to provide suitable dose–response curves for calibration (electronic supplementary material, files S1–S7

Figure 2 shows the sum of all transcript abundances calculated from the calibrated probes in dependence of postmortem time. In general, the sum of all abundances decreased with time, which means that less transcript targets hybridized to the microarray probes. In the zebrafish, mRNA decreased abruptly at 12 h postmortem (figure 2a), while for the mouse brain (figure 2b), mRNA increased in the first hour and then gradually decreased. For the mouse liver, mRNA gradually decreased with postmortem time. The fact that total mRNA shown in figure 2a,b mirrors the electrophoresis patterns shown in the electronic supplementary material, figures S1 and S2 (ignoring the 28S and 18S rRNA bands) indicates a general agreement of the Gene Meter approach to the gel-based approach (i.e. Agilent Bioanalyzer). Hence, mRNA abundances depend on the organism (zebrafish, mouse), organ (brain, liver) and postmortem time, which is aligned with previous studies [11–16].

Figure 2. Total mRNA abundance (arbitrary units, arb. units) by postmortem time determined using all calibrated microarray probes: (a) extracted from whole zebrafish, (b) extracted from the brain and liver tissues of whole mice. Each datum point represents the mRNA from two individuals in the zebrafish and a single individual in the mouse.

The abundance of a transcript is determined by its rate of synthesis and its rate of degradation [17]. We focused here on transcripts that significantly increased in abundance—relative to live controls—because these genes might be actively transcribed after organismal death despite an overall decrease in total mRNA with time. A transcript was defined as having a significantly increased abundance when at least one time point was statistically higher than that of the control (figure 1a–c). It is important to understand that the entire profiles, i.e. 22 data points for the zebrafish and 20 points for the mouse, were subjected to a statistical test to determine significance (see Material and methods). We found 548 zebrafish profiles and 515 mouse profiles had significantly increased transcript abundances.

Based on GenBank gene annotations, we found that, among the transcripts with significantly increased abundances, for the zebrafish 291 were protein-coding genes (53%) and 257 non-annotated mRNA (47%), and for the mouse 324 were known protein-coding genes (63%), 190 non-annotated mRNA (37%) and one an Agilent control sequence of unknown composition. Hence, in the zebrafish and mouse, about 58% of the total genes with significant transcript abundances are known and the rest (42%) are putatively non-annotated RNA.

Examples of genes yielding transcripts that significantly increased in abundance with postmortem time are: the Heat shock protein (Hsp70.3) gene, the Thymocyte selection-associated high mobility group box 2 (Tox2) gene, and an unknown (NULL) gene (figure 1a–c). While the Hsp70.3 transcript abundance increased after 1 h postmortem to reach a maximum at 12 h, the Tox2 transcript increased after 12 h postmortem to reach a maximum at 24 h, and the NULL transcript consistently increased with postmortem time. These figures provide typical examples of transcript profiles and depict the high reproducibility of the sample replicates as well as the quality of output obtained by the Gene Meter approach.

3.1. Non-random patterns in transcript profiles

Ordination plots of the transcript profiles that had significantly increased abundances revealed prominent differences with postmortem time (figure 1d,e), suggesting the increases in transcript abundances of genes followed a discernable (non-random) pattern in both organisms. The biplots showed that 203 zebrafish transcript profiles and 226 mouse profiles significantly contributed to the ordinations. To identify patterns in the transcript profiles, we assigned them to groups based on their position in the biplots. Six profile groups were assigned for the zebrafish (A to F) and five groups (G to K) were assigned for the mouse. Determination of the average gene transcript abundances by group revealed differences in the shapes of the averaged profiles, particularly the timing and magnitude of peak abundances, which accounted for the positioning of data points in the ordinations.

Genes coding for global regulatory functions were examined separately from others (i.e. response genes). Combined results show that about 33% of the genes in the ordination plots were involved in global regulation, with 14% of these encoding transcription factors/transcriptional regulators and 19% encoding cell signalling proteins such as enzymes, messengers and receptors (electronic supplementary material, table S3). The response genes accounted for 67% of the total.

The genes were assigned to 22 categories (electronic supplementary material, File S8) with some genes having multiple categorizations. For example, the Eukaryotic translation initiation factor 3 Subunit J-B (Eif3j2) gene was assigned to protein synthesis and cancer categories [18].

Genes in the following functional categories were investigated: stress, immunity, inflammation, apoptosis, solute/ion/protein transport, embryonic development, epigenetic regulation and cancer. We focused on these categories because they were common to both organisms, they contained multiple genes and they might provide possible explanations for the postmortem increases in transcript abundances (e.g. epigenetic gene regulation, embryonic development, cancer). The transcriptional profiles were plotted by category and each profile was ordered by the timing of the increased abundance and peak abundances. This allowed comparisons of transcript dynamics as a function of postmortem time for both organisms. For each category, we provided the name and function of the gene and compared transcript dynamics within and between the organisms.

3.2. Stress response

In organismal death, transcripts from stress response genes were anticipated to significantly increase in abundance because these genes are activated in life to cope with perturbations, recover homeostasis [19] and stabilize the cytoskeleton [20]. The stress response genes were assigned to three groups: heat shock protein (Hsp), hypoxia-related and ‘other’ responses such as oxidative stress.

3.2.1. Hsp

In the zebrafish, Hsp gene transcripts that significantly increased in abundance included: Translocated promoter region (Tpr), Hsp70.3 and Hsp90 (figure 3). The Tpr gene encodes a protein that facilitates the export of the Hsp mRNA across the nuclear membrane [21] and has been implicated in chromatin organization, regulation of transcription, mitosis [22] and controlling cellular senescence [23]. The Hsp70.3 and Hsp90 genes encode proteins that control the level of intracellular calcium [24], assist with protein folding and aid in protein degradation [25].

Figure 3. Increased abundance of stress response gene transcripts by postmortem time (h) and stress category: (a) zebrafish and (b) mouse. Green, intermediate value red, maximum value.

In the mouse, the Hsp gene transcripts included: Tpr, Hsp-associated methyltransferase (Mettl21) and Heat shock protein 1 (Hspe1) (figure 3). The Mettl21 gene encodes a protein modulating Hsp functions [26]. The Hspe1 gene encodes a chaperonin protein that assists with protein folding in the mitochondria [27].

The timing and duration of the Hsp transcript abundances varied by organism. In general, the increase in transcript abundance of Hsp genes occurred much later in the zebrafish than the mouse (4 h versus 0.5 h postmortem, respectively). There were also differences in transcript abundance maxima of Hsp genes since they reached maxima at 9–24 h in the zebrafish, while they reached maxima at 12–24 h in the mouse. Previous studies have examined the increase of Hsp70.3 transcripts with time in live serum-stimulated human cell lines [28]. In both the zebrafish and human cell lines (figure 1a), the Hsp70.3 gene transcript reached a maximum abundance at approximately 12 h indicating the same reactions occur in life and death.

3.2.2. Hypoxia

In the zebrafish, hypoxia-related gene transcripts that significantly increased in abundance included: Carbonic anhydrase 4 (Ca4c), Nuclear factor (NF) interleukin-3 (Nfil3), Hypoxia-inducible factor 1-alpha (Hiflab) and Arginase-2 (Arg2) (figure 3). The Carbonic anhydrase 4 (Ca4c) gene encodes an enzyme that converts carbon dioxide into bicarbonate in response to anoxic conditions [29]. The Nfil3 gene encodes a protein that suppresses hypoxia-induced apoptosis [30] and activates immune responses [31]. The Hiflab gene encodes a transcription factor that prepares cells for decreased oxygen [32]. The Arg2 gene encodes an enzyme that catalyses the conversion of arginine to urea under hypoxic conditions [33]. Of note, the accumulation of urea presumably triggered the increase of Slc14a2 gene transcripts at 24 h, as reported in the Transport Section (below).

In the mouse, the hypoxia-related gene transcripts that significantly increased in abundance included: Methyltransferase hypoxia-inducible domain (Methig1) and Sphingolipid delta-desaturase (Degs2) (figure 3). The Methig1 gene encodes methyltransferase that presumably is involved in gene regulation [34]. The Degs2 gene encodes a protein that acts as an oxygen sensor and regulates ceramide metabolism [35]. Ceramides are waxy lipid molecules in cellular membranes that regulate cell-growth, death, senescence, adhesion, migration, inflammation, angiogenesis and intracellular trafficking [36].

The increased abundance of Ca4c transcripts in the zebrafish putatively indicates a build up of carbon dioxide 0.1–1 h postmortem in the zebrafish presumably due to lack of blood circulation. The increased abundance of the Nfil3 transcripts in the zebrafish and Methig1 transcripts in the mouse suggests hypoxic conditions exist within 0.5 h postmortem in both organisms. The increased abundance of the other hypoxia gene transcripts varied with postmortem time, with increases of Hiflab, Arg2 and Degs2 transcripts at 4 h, 12 h and 24 h, respectively.

3.2.3. Other stress responses

In the zebrafish, gene transcripts that significantly increased in abundance included: Alkaline ceramidase 3 (Acer3), Peroxirodoxin 2 (Prdx2), Immediate early (Ier2), Growth arrest and DNA-damage-inducible protein (Gadd45a), Zinc finger CCH domain-containing 12 (Zcchc12), Corticotropin-releasing hormone receptor 1 (Crhr1) and Zinc finger AN1-type domain 4 (Zfand4) (figure 3). The Acer3 gene encodes a stress sensor protein that mediates cell-growth arrest and apoptosis [37]. The Prdx2 gene encodes an antioxidant enzyme that controls peroxide levels in cells [38] and triggers production of Tnfa proteins that induce inflammation [39]. The Ier2 gene encodes a transcription factor involved in stress response [40]. The Gadd45a gene encodes a stress protein sensor that stops the cell cycle [41], modulates cell death and survival, and is part of the signalling networks in immune cells [42]. The Zcchc12 gene encodes a protein involved in stress response in the brain [43]. The Crhr1 and Zfand4 genes encode stress proteins [44,45].

While the Acer3, Prdx2 and Ier2 transcripts increased within 0.3 h postmortem, indicating a changed physiological state, the Gadd45a transcript increased at 9 h and the other transcripts (Zcchc12, Crhr1 and Zfand4) increased at 24 h postmortem.

In the mouse, gene transcripts that significantly increased in abundance included: Membrane-associated RING-CH 4 (March4), Homocysteine-responsive endoplasmic reticulum-resident ubiquitin-like domain member 2 (Herpud2), Prohibitin-2 (Phb2), Gadd45a and Two-oxoglutarate and iron-dependent oxygenase domain-containing 1 (Ogfod1) (figure 3). The March4 gene encodes an immunologically-active stress response protein [46]. The Herpud2 gene encodes a protein that senses the accumulation of unfolded proteins in the endoplasmic reticulum [47]. The Phb2 gene encodes a cell surface receptor that responds to mitochondrial stress [48]. The Ogfod1 gene encodes a stress-sensing protein [49].

Note that the stress gene transcripts in the mouse all increased within 1 h postmortem and remained at high abundance for 48 h.

3.2.4. Summary of stress response

In both organisms, organismal death increased the abundance of heat shock, hypoxia and ‘other stress’ gene transcripts, which varied in their timing and duration within and between organisms. Consider, for example, the Tpr and Gadd45a genes, which were common to both organisms. While the transcript abundance for the Tpr gene significantly increased within 0.5 h postmortem in both organisms, the transcript abundance for the Gadd45a gene increased at 9 h in the zebrafish and 0.5 h in the mouse. In addition, the transcriptional profile of the Tpr gene was more variable in the zebrafish than that of the mouse since the transcripts increased in abundance at 0.3 h, 9 h and 24 h postmortem, which suggests that they might be regulated through a feedback loop. By contrast, the transcriptional profile of Tpr gene in the mouse increased at 0.5 h and peaked at 12 and 24 h postmortem.

Taken together, the significant increase in transcript abundance of stress genes in both organisms is presumably to compensate for a loss of homeostasis.

3.3. Innate and adaptive immune responses

In organismal death, an increase of immune response gene transcripts was anticipated since vertebrates have evolved ways to protect the host against infection in life, even under absolutely sterile conditions [50]. Inflammation genes were excluded from this section (even though they are innate immune genes) because we examined them in a separate section (below).

In the zebrafish, gene transcripts that significantly increased in abundance included: Early growth response-1 and -2 (Egr1, Egr2), Interleukin-1b (Il1b), l -amino acid oxidase (Laao), Interleukin-17c (Il17c), Membrane-spanning 4-domains subfamily A member 17A.1 (Ms4a17.a1), Mucin-2 (Muc2), Immunoresponsive gene 1 (Irg1), Interleukin-22 (Il22), Ubl carboxyl-terminal hydrolase 18 (Usp18), ATF-like 3 (Batf3), Cytochrome b-245 light chain (Cyba) and Thymocyte selection-associated high mobility group box protein family member 2 (Tox2) (figure 4). The Egr1 and Egr2 genes encode proteins that regulate B- and T-cell functions in adaptive immunity [51,52]. The Il1b gene encodes an interleukin that kills bacterial cells through the recruitment of other antimicrobial molecules [53]. The Laao gene encodes an oxidase involved in innate immunity [54]. The Il17c and Il22 genes encode interleukins that work synergically to produce antibacterial peptides [55]. The Ms4a17.a1 gene encodes a protein involved in adaptive immunity [56]. The Muc2 gene encodes a protein that protects the intestinal epithelium from pathogenic bacteria [57]. The Irg1 gene encodes an enzyme that produces itaconic acid, which has antimicrobial properties [58]. The Usp18 gene encodes a protease that plays a role in adaptive immunity [59]. The Batf3 gene encodes a transcription factor that activates genes involved in adaptive immunity [60]. The Cyba gene encodes an oxidase that is used to kill microorganisms [61]. The Tox2 gene encodes a transcription factor that regulates natural killer (NK) cells of the innate immune system [62].

Figure 4. Abundance of immunity gene transcripts by postmortem time (h): (a) zebrafish and (b) mouse. Green, intermediate value red, maximum value. Some transcripts were represented by two different probes (e.g. Il1b, Laao).

Increases of immunity gene transcripts in the zebrafish occurred at different times with varying durations. While transcripts of genes involved in adaptive immunity increased in abundance at 0.1–0.3 h (Egr), 9 h (Ms4a17.a1) and 24 h (Usp18, Batf3) postmortem, transcripts of genes involved in innate immunity increased at 4 h (Il1b), 9 h (Laao, Il17c), 12 h (Muc2, Irg1) and 24 h (Il22, Cyba,Tox2), indicating a multi-pronged and progressive approach to deal with injury and the potential of microbial invasion.

In the mouse, gene transcripts that significantly increased in abundance included: Catalytic polypeptide-like 3G (Apobec3g), CRISPR-associated endonuclease (Cas1), Perforin-1 (Prf1), Immunoglobulin heavy variable 8–11 (Ighv8-11), C4b-binding protein (C4b), Complement component C7 (C7), T-cell receptor alpha and delta chain (Tcra/Tcrd), High affinity immunoglobulin gamma Fc receptor I (Fcgr1a), Defensin (Defb30), Chemokine-4 (Ccr4), Interleukin-5 (Il5), NK cell receptor 2B4 (Cd244), Cluster of differentiation-22 (Cd22), Lymphocyte cytosolic protein 2 (Lcp2), Histocompatibility 2 O region beta locus (H2ob) and Interferon-induced transmembrane protein 1 (Ifitm1) (figure 4). The Apobec3g gene encodes a protein that plays a role in innate anti-viral immunity [63]. The Cas1 gene encodes a protein involved in regulating the activation of immune systems [64–67]. The Prf1, C7 and Defb30 genes encode proteins that kill bacteria by forming pores in plasma membrane of target cells [68–70]. The Ighv8-11 gene encodes an immunoglobulin of uncertain function. The C4b gene encodes a protein involved in the complement system [71]. The Tcra/Tcrd genes encode proteins that play a role in the immune response [72]. The Fcgr1a gene encodes a protein involved in both innate and adaptive immune responses [73]. The Ccr4 gene encodes a cytokine that attracts leucocytes to sites of infection [74]. The Il5 gene encodes an interleukin involved in both innate and adaptive immunity [75,76]. The Cd244 and Cd22 genes encode proteins involved in innate immunity [77]. The Lcp2 gene encodes a signal-transducing adaptor protein involved in T cell development and activation [78]. The H2ob gene encodes a protein involved in adaptive immunity. The Ifitm1 gene encodes a protein that has anti-viral properties [79].

Most of the transcripts of immune response genes increased in abundance within 1 h postmortem in the mouse (n = 14 out of 16 genes), indicating a more rapid response than that of the zebrafish.

3.3.1. Summary of immune response

The increase in transcript abundance of immune response genes in both organisms included innate and adaptive immunity components. An interesting phenomenon observed in the mouse (but not zebrafish) was that four genes (C7, Tcra/Tcrd, Fcgr1a and Defb30) reached transcript abundance maxima at two different postmortem times (i.e. 1 h and 12 h) while others reached only one. The variability in the gene transcript abundances suggests possible regulation by feedback loops.

3.4. Inflammation response

The increased abundance of transcripts of inflammation genes in organismal death was anticipated because inflammation is an innate immunity response to injury. In the zebrafish, inflammation gene transcripts that increased in abundance included: Egr1, Egr2, Il1b, Tumour necrosis factor receptor (Tnfrsf19), Haem oxygenase 1 (Hmox1), Tumour necrosis factor (Tnf), G-protein receptor (Gpr31), Interleukin-8 (Il8), Tumour necrosis factor alpha (Tnfa), NF kappa B (Nfkbiaa), MAP kinase-interacting serine/threonine kinase 2b (Mknk2b) and Corticotropin-releasing factor receptor 1 (Crhr1) (figure 5). The Egr1 and Egr2 genes encode transcription factors that are pro- and anti-inflammatory, respectively [51,52,80]. The Il1b gene encodes a pro-inflammatory cytokine that plays a key role in sterile inflammation [81,82]. The Tnfrsf19 gene encodes a receptor that has pro-inflammatory functions [83]. The Hmox1 gene encodes an enzyme that has anti-inflammatory functions and is involved in haem catabolism [84,85]. The Tnf and Tnfa genes encode pro-inflammatory proteins. The Gpr31 gene encodes a pro-inflammatory protein that activates the NF-κB signalling pathway [86]. The Il8 gene encodes a cytokine that has pro-inflammatory properties [87]. The Nfkbiaa gene encodes a protein that integrates multiple inflammatory signalling pathways including Tnf genes [88]. The Mknk2b gene encodes a protein kinase that directs cellular responses and is pro-inflammatory [89]. The Crhr1 gene modulates anti-inflammatory responses [90].

Figure 5. Abundance of inflammation gene transcripts by postmortem time (h): (a) zebrafish and (b) mouse. Inflammation, pro-, + anti, −. Green, intermediate value red, maximum value. The Il1b and Mknk2b genes were represented by two different probes.

The increased abundance of the pro-inflammatory Egr1 transcript at 0.1 h was followed by an increase of anti-inflammatory Egr2 transcript at 0.2 h, suggesting the increase of one transcript was effecting another (figure 5). Similarly, the increased abundance of pro-inflammatory Il1b transcript at 4 h postmortem was followed by: increased abundance of pro-inflammatory Tnfrsf19, Tnf, Gpr31 and Il8 transcripts and the anti-inflammatory Hmox1 transcript at 9 h, the increased abundance of pro-inflammatory Tnfa, Nfkbiaa and Mknk2b transcripts at 12 h, and the increased abundance of anti-inflammatory Crhr1 transcripts at 24 h. Of note, while none of the pro-inflammatory gene transcripts increased in abundance past 24 h, the anti-inflammatory Crhr1 gene remained at high abundance at 48 h. It should also be noted that the Il1b, Il8 and Tnfa gene transcripts have been reported to be increased in traumatic impact injuries in postmortem tissues from human brains [91].

In the mouse, inflammation gene transcripts that increased in abundance included: mitogen-activated protein kinase (Map3k2), TNF receptors (Tnfrsf9, Tnfrs14), B-cell lymphoma 6 protein (Bcl6), C-C chemokine receptor-type 4 (Ccr4), Prokineticin-2 (Prok2) and platelet-activating factor receptor (Pafr) (figure 5). The Map3k2 gene encodes a kinase that activates pro-inflammatory NF-κB genes [89]. The Tnfrsf9 and Tnfrs14 genes encode receptor proteins that have pro-inflammatory functions [83]. The Bcl6 gene encodes a transcription factor that has anti-inflammatory functions [92]. The Ccr4 gene encodes a cytokine receptor protein associated with inflammation [74]. The Prok2 gene encodes a cytokine-like molecule, while the Pafr gene encodes a lipid mediator both have pro-inflammatory functions [93,94].

Most inflammation-associated gene transcripts increased in abundance within 1 h postmortem and continued to be abundant for 12–48 h. The anti-inflammatory Bcl6 gene transcripts increased in abundance at two different times, 0.5–6 h and 24 h, suggesting that their abundances might be regulated by a feedback loop. It should also be noted that pro-inflammatory Map3k2 and Tnfrs14 gene transcripts were not at high abundance after 24 and 12 h, respectively, which also suggests regulation by a putative feedback loop from the Bcl6 transcript product.

3.4.1. Summary of inflammation response

In both organisms, some transcripts that increased in abundance have pro-inflammatory functions while others have anti-inflammatory functions. It is possible that the increases in transcript abundances are regulated by feedback loops involving an initial inflammatory reaction followed by an anti-inflammatory reaction to repress it [95]. The variation in the gene transcript abundances of these inflammatory genes suggests the underlying regulatory network is still active in organismal death.

3.5. Apoptosis and related genes

Since apoptotic processes kill damaged cells for the benefit of the organism as a whole, we anticipated a significant increase in the abundance of apoptosis gene transcripts in organismal death.

In the zebrafish, apoptosis gene transcripts that increased in abundance included: Jun (Jdp2, Jun), Alkaline ceramidase 3 (Acer3), Fos (Fosb, Fosab, Fosl1), IAP-binding mitochondrial protein A (Diabloa), Peroxiredoxin-2 (Prdx2), Potassium voltage-gated channel member 1 (Kcnb1), Caspase apoptosis-related cysteine peptidase 3b (Casp3b), DNA-damage-inducible transcript 3 (Ddit3), BCL2 (B-cell lymphomas 2)-interacting killer (Bik) and Ras association domain family 6 (Rassf6) (figure 6). The Jdp2 gene encodes a protein that represses the activity of the transcription factor activator protein 1 (AP-1) [96]. The Acer3 gene encodes an enzyme that maintains cell membrane integrity/function and promotes apoptosis [97]. The Fos genes encode proteins that dimerize with Jun proteins to form part of the AP-1 that promotes apoptosis [98,99]. The Diabloa gene encodes a protein that neutralizes inhibitors of apoptosis (IAP)-binding protein [99] and activates caspases [100]. The Prdx2 gene encodes antioxidant enzymes that control cytokine-induced peroxide levels and inhibit apoptosis [101]. Although the Kcnb1 gene encodes a protein used to make ion channels, the accumulation of these proteins in the membrane promotes apoptosis via a cell signalling pathway [102]. The Casp3b encodes a protein that plays a role in the execution phase of apoptosis [103]. The Ddit3 gene encodes a transcription factor that promotes apoptosis. The Bik gene encodes a protein that promotes apoptosis [104]. The Rassf6 gene encodes a protein that promotes apoptosis [105].

Figure 6. Abundance of apoptosis gene transcripts by postmortem time (h): (a) zebrafish and (b) mouse. Apoptosis, pro-, + anti, −. Green, intermediate value red, maximum value. The Fosb gene was represented by two different probes.

In the zebrafish, transcripts of both anti-apoptosis Jdp2 and pro-apopotosis Acer3 genes increased in abundances within 0.1 h postmortem (figure 6). These increases were followed by increases of five pro-apoptosis gene transcripts and one anti-apoptosis gene transcript within 0.3–0.5 h. The transcriptional dynamics varied among the genes. Specifically, (i) the increased abundance of the Fosb gene transcript stopped after 1 h, (ii) the transcripts of the Diabloa and Fosab genes reached abundance maxima at 0.5–4 h and then their abundances decreased after 9 h for the Diabloa and after 24 h for the Fosab genes, (iii) the Jun gene transcripts reached two maxima (one at 0.5 and another at 4–12 h)—then its abundance decreased after 24 h, and (iv) the transcript of the Prdx2 gene showed a continuous increase in abundance until reaching a maximum at 24 h and then the abundance decreased. The remaining genes were pro-apoptosis and their transcripts increased in abundance after 1–24 h postmortem. The transcripts of the Ddit3 and Rassf6 genes were very different from the other transcripts because they increased in abundance at one sampling time (12 h and 24 h, respectively) and then decreased. Apparently none of the transcripts of apoptosis genes increased in abundance after 24 h, in contrast to genes in other categories (e.g. transcripts of some of stress and immunity genes increased in abundance up to 96 h postmortem).

In the mouse, apoptosis gene transcripts that increased in abundance included: BCL2-like protein 11 (Bcl2L11), Casein kinase IIa (Csnk2a1), Interleukin 15 receptor subunit a (Il15ra), Myocyte enhancer factor 2 (Mef2a), F-box only protein 10 (Fbxo10), Sp110 nuclear body protein (Sp110), TGFB-induced factor homeobox 1 (Tgif1), Intersectin 1 (Itsm1), the Ephrin type-B receptor 3 (Ephb3) and the p21 protein-activated kinase 4 (Pak4) (figure 6). The Bcl2L11 gene encodes a protein that promotes apoptosis [106]. The Csnk2a1 gene encodes an enzyme that phosphorylates substrates and promotes apoptosis [107]. The Il15ra gene encodes an anti-apoptotic protein [108]. The Mef2a gene encodes a transcription factor that prevents apoptosis [109]. The Fbxo10 gene encodes a protein that promotes apoptosis [110]. The Sp110 gene encodes a regulator protein that promotes apoptosis [111]. The Tgif1 gene encodes a transcription factor that blocks signals of the transforming growth factor beta (TGFβ) pathway, and therefore is pro-apoptosis [112]. The Itsn1 gene encodes an adaptor protein that is anti-apoptosis [113]. The Ephb3 gene encodes a protein that binds ligands on adjacent cells for cell signalling and suppresses apoptosis [108]. The Pak4 gene encodes a protein that delays the onset of apoptosis [114].

In the mouse, transcripts for the pro- and anti-apoptosis genes increased in abundance within 0.5 h postmortem however, with the exception of Bcl2L11, most reached transcript abundance maxima at 12–48 h postmortem (figure 6). The Bcl2L11 transcripts reached abundance maxima at 1 and 6 h postmortem.

3.5.1. Summary of apoptotic response

In both organisms, transcripts of both pro- and anti-apoptosis genes increased in abundance in organismal death. However, the timings of the increases, the transcript maximum abundance and the duration of the increased abundances varied by organism. The results suggest the apoptotic genes and their regulation are distinctly different in the zebrafish than the mouse, with transcripts of the mouse genes having increased abundance to 48 h postmortem while those of the zebrafish having increased abundance to 24 h. Nonetheless, the pro- and anti-apoptosis genes appear to be inter-regulating each another.

3.6. Transport gene response

Transport processes maintain ion/solute/protein homeostasis and are involved in influx/efflux of carbohydrates, proteins, signalling molecules and nucleic acids across membranes. Transcripts of transport genes should increase in abundance in organismal death in response to dysbiosis.

In the zebrafish, transport-associated gene transcripts that increased in abundance included: Solute carrier family 26 anion exchanger member 4 (Slc26a4), Potassium channel voltage-gated subfamily H (Kcnh2), Transmembrane emp24 domain-containing protein 10 (Tmed10), Leucine-rich repeat-containing 59 (Lrrc59), the Nucleoprotein TPR (Tpr), Importin subunit beta-1 (Kpnb1), Transportin 1 (Tnpo1), Syntaxin 10 (Stx10) and Urea transporter 2 (Slc14a2) (figure 7). Of note, the four Tmed10 transcripts shown in figure 7 each represents a profile targeted by an independent probe. The transcription profiles of this gene were identical indicating high reproducibility of the Gene Meter approach. The Slc26a4 gene encodes prendrin that transports negatively charged ions (i.e. Cl − , bicarbonate) across cellular membranes [115]. The Kcnh2 gene encodes a protein used to make potassium channels and is involved in signalling [116]. The Tmed10 gene encodes a membrane protein involved in vesicular protein trafficking [117]. The Lrrc59, Tpr, Tnpo1 and Kpnb1 genes encode proteins involved in trafficking across nuclear pores [118–121]. The Stx10 gene encodes a protein that facilitates vesicle fusion and intracellular trafficking of proteins to other cellular components [122]. The Slc14a2 gene encodes a protein that transports urea out of the cell [123].

Figure 7. Abundance of transport gene transcripts by postmortem time (h): (a) zebrafish and (b) mouse. Green, intermediate value red, maximum value. The Tmed10 gene was represented by four different probes.

The transcripts of Slc26a4, Kcnh2, Lrrc59 and Tpr genes initially increased in abundance within 0.3 h postmortem and remained in high abundance for 12–24 h. The transcripts of the Tnpo1 gene increased in abundance twice, at 4 and 12 h, suggesting putative regulation by a feedback loop. The transcripts of the remaining genes increased in abundance at 24 h. The increased abundance of the Slc14a2 gene transcript suggests a build up of urea in zebrafish cells at 24–96 h postmortem, which could be due to the accumulation of urea under hypoxic conditions by the Arg2 gene (see Hsp stress response section).

In the mouse, transport-associated gene transcripts that increased in abundance included: Calcium-binding mitochondrial carrier protein (Aralar2), Sodium-coupled neutral amino acid transporter 4 (Slc38a4), SFT2 domain-containing 1 (Sft2d1), Uap56-interacting factor (Fyttd1), Solute carrier family 5 (sodium/glucose co-transporter) member 10 (Slc5a10), Mitochondrial import receptor subunit (Tom5), Translocated promoter region (Tpr), ATP-binding cassette transporter 12 (Abca12), Multidrug resistant protein 5 (Abc5), LIM and SH3 domain-containing protein (Lasp1), Chromosome 16 open reading frame 62 (C16orf62), Golgi transport 1 homologue A (Golt1a), ATP-binding cassette transporter 17 (Abca17), Nucleotide exchange factor (Sil1), Translocase of inner mitochondrial membrane 8A1 (Timm8a1), Early endosome antigen 1 (Eea1) and Potassium voltage-gated channel subfamily V member2 (Kcnv2) (figure 7). The Aralar2 gene encodes a protein that catalyses calcium-dependent exchange of cytoplasmic glutamate with mitochondrial aspartate across the mitochondrial membrane and may function in the urea cycle [124]. The Slc38a4 gene encodes a symport that mediates transport of neutral amino acids and sodium ions [125]. The Sft2d1 gene encodes a protein involved in transporting vesicles from the endocytic compartment of the Golgi complex [126]. The Fyttd1 gene is responsible for mRNA export from the nucleus to the cytoplasm [127]. The Slc5a10 gene encodes a protein that catalyses carbohydrate transport across cellular membranes [128]. The Tom5 gene encodes a protein that plays a role in importation to proteins destined for mitochondrial sub-compartments [129]. The Abca12, Abca17 and Abc5 genes encode proteins that transport molecules across membranes [130–132]. The Lasp1 gene encodes a protein that regulates ion transport [133]. The C16orf62 gene encodes a protein involved in protein transport from the Golgi apparatus to the cytoplasm [134]. The Golt1a gene encodes a vesicle transport protein [126]. The Sil1 gene encodes a protein involved in protein translocation into the endoplasmic reticulum [135]. The Timm8a1 gene encodes a protein that assists importation of other proteins across inner mitochondrial membranes [136]. The Eea1 gene encodes a protein that acts as a tethering molecule for vesicular transport from the plasma membrane to the early endosomes [137]. The Kcnv2 gene encodes a membrane protein involved in generating action potentials [138].

Within 0.5 h postmortem, transcripts of genes involved in: (i) ion and urea regulation (Aralar), (ii) amino acid (Slc38a4), carbohydrate (Slc5a10) and protein (Sft2d1, Tom5) transport, (iii) mRNA nuclear export (Fyttd1, Tpr) and (iv) molecular efflux (Abca12, Abc5) increased in abundance in the mouse. The transcription profiles of these genes varied in terms of transcript abundance maxima and duration. While the transcripts of Aralar, Sft2d1, Slc38a4, Fyttd1 and Slc5a10 reached abundance maxima at 1 h, those of Tom5, Tpr, Abca12 and Abc5 reached maxima at 12–24 h postmortem. The duration of the increased abundance also varied for these transcripts since most remained at high abundances for 48 h postmortem, while the Sft2d1, Fyttd1 and Slc5a10 transcripts were at high abundances from 0.5 to 12+ h. The shorter duration of increased abundance suggests prompt gene repression. The transcript abundances of Lasp1, C16orf62, Golt1a and Abca17 increased at 1 h postmortem and remained elevated for 48 h. The transcripts of Sil1, Timm8a1 and Eea1 increased in abundance at 6 h, while those of Kcnv2 increased at 24 h postmortem and remained elevated for 48 h.

3.6.1. Summary of transport genes

The increased abundance of transcripts of transport genes suggests attempts by zebrafish and mice to reestablish homeostasis. Although the transcripts of half of these genes increased in abundance within 0.5 h postmortem, many increased at different times and for varying durations. While most of the transcripts of transport genes in the zebrafish were not abundant after 24 h, most transcripts of transport genes in the mouse remained abundant for 24–48 h postmortem.

3.7. Developmental control genes

An unexpected finding in this study was the increased abundance of transcripts of developmental control genes in organismal death. Developmental control genes are mostly involved in regulating developmental processes from early embryo to adult in the zebrafish and mouse therefore, we did not anticipate their transcripts to become more abundant in organismal death.

In the zebrafish, development-associated gene transcripts that increased in abundance included: LIM domain-containing protein 2 (Limd2), Disheveled-associated activator of morphogenesis 1 (Daam1b), Meltrin alpha (Adam12), Hatching enzyme 1a (He1a), Midnolin (Midn), Immediate early response 2 (Ier2), Claudin b (Cldnb), Regulator of G-protein signalling 4-like (Rgs4), Proline-rich transmembrane protein 4 (Prrt4), Inhibin (Inhbaa), Wnt inhibitory factor 1 precursor (Wif1), Opioid growth factor receptor (Ogfr), Strawberry notch homolog 2 (Sbno2) and Developing brain homeobox 2 (Dbx2) (figure 8). The Limd2 gene encodes a binding protein that plays a role in zebrafish embryogenesis [139]. The Daam1b gene regulates endocytosis during notochord development [140]. The Adam12 gene encodes a metalloprotease-disintegrin involved in myogenesis [141]. The He1a gene encodes a protein involved in egg envelope digestion [142]. The Midn gene encodes a nucleolar protein expressed in the brain that is involved in the regulation of neurogenesis [143,144]. The Ier2 gene encodes a protein involved in left–right asymmetry patterning in the zebrafish embryo [145]. The Cldnb gene encodes a tight junction protein in larval zebrafish [146]. The Rgs4 gene encodes a protein involved in brain development [147]. The Prrt4 gene encodes a protein that is predominantly expressed in the brain and spinal cord in embryonic and postnatal stages of development. The Inhbaa gene encodes a protein that plays a role in oocyte maturation [148]. The Wif1 gene encodes a WNT inhibitory factor that controls embryonic development [149]. The Ogfr gene plays a role in embryonic development [150]. The Sbno2 gene plays a role in zebrafish embryogenesis [151]. The Dbx2 gene encodes a transcription factor that plays role in spinal cord development [152].

Figure 8. Abundance of development gene transcripts by postmortem time (h): (a) zebrafish and (b) mouse. Green, intermediate value red, maximum value. The Cldnb gene was represented by two different probes.

Although the abundances of Limd2, Daam1, Adam12 and He1a transcripts increased in the zebrafish within 0.1 h postmortem, other gene transcripts in this category increased from 0.3 to 24 h postmortem, reaching abundance maxima at 24 h or more.

In the mouse, development-associated gene transcripts that increased in abundance included: MDS1 and EVI1 complex locus protein EVI1 (Mecom), MAM domain-containing glycosylphosphatidylinositol anchor 2 (Mdga2), FYVE, RhoGEF and PH domain-containing 5 (Fgd5), RNA-binding motif protein 19 (Rbm19), Chicken ovalbumin upstream promoter (Coup), Single-minded homolog 2 (Sim2), Solute carrier family 38, member 4 (Slc38a4), B-cell lymphoma 6 protein (Bcl6), Sema domain transmembrane domain (TM) cytoplasmic domain (semaphorin) 6D (Sema6d), RNA binding motif protein 45 (Rbm45), Transcription factor E2F4 (E2f4), Long chain fatty acid-CoA ligase 4 (Lacs4), Kallikrein 1-related peptidase b3 (Klk1b3), Sema domain, immunoglobulin domain, TM and short cytoplasmic domain (Sema4c), TGFB-induced factor homeobox 1 (Tgif1), Interferon regulatory factor 2-binding protein-like (Irf2bpl), Ephrin type-B receptor 3 (Ephb3), Testis-specific Y-encoded-like protein 3 (Tspyl3), Protein ripply 3 (Ripply3), Src kinase-associated phosphoprotein 2 (Skap2), DNA polymerase zeta catalytic subunit (Rev3l), MKL/myocardin-like 2 (Mkl2) and Protein phosphatase 2 regulatory subunit A (Ppp2r1a) (figure 8). The Mecom gene plays a role in embryogenesis and development [153]. The Mdga2 gene encodes immunoglobins involved in neural development [154]. The Fgd5 gene is needed for embryonic development since it interacts with hematopoietic stem cells [155]. The Rbm19 gene is essential for preimplantation development [156]. The Coup gene encodes a transcription factor that regulates development of the eye [157] and other tissues [158]. The Sim2 gene encodes a transcription factor that regulates cell fate during midline development [159]. The Slc38a4 gene encodes a regulator of protein synthesis during liver development and plays a crucial role in fetal growth and development [160,161]. The Bcl6 gene encodes a transcription factor that controls neurogenesis [162]. The Sema6d gene encodes a protein involved in retinal development [163]. The Rbm45 gene encodes a protein that has preferential binding to poly(C) RNA and is expressed during brain development [164]. The E2f4 gene is involved in maturation of cells in tissues [165]. The Lacs4 gene plays a role in patterning in embryos [166]. The Klk1b3 gene encodes a protein that plays a role in the developing embryos [167]. The Sema4c gene encodes a protein that has diverse function in neuronal development and heart morphogenesis [168,169]. The Tgif1 gene encodes a transcription factor that plays a role in trophoblast differentiation [170]. The Irf2bpl gene encodes a transcriptional regulator that plays a role in female neuroendocrine reproduction [171]. The Ephb3 gene encodes a kinase that plays a role in neural development [172]. The Tspyl3 gene plays a role in testis development [173]. The Ripply3 gene encodes a transcription factor involved in development of the ectoderm [174]. The Skap2 gene encodes a protein involved in actin reorganization in lens development [175]. The Rev3l gene encodes a polymerase that can replicate past certain types of DNA lesions and is necessary for embryonic development [176]. The Mkl2 gene encodes a transcriptional co-activator that is involved in the formation of muscular tissue during embryonic development [177]. The Ppp2r1a gene plays a role in embryonic epidermal development [178].

The transcripts of Mecom, Mdga2, Fgd5, Rbm19, Coup, Sim2, Slc38a4, Bcl6, Sema6d, Rbm45, E2f4 and Lacs4 genes in the mouse significantly increased in abundance within 0.5 h postmortem but the other transcripts increased from 1 h to 48 h reaching abundance maxima at 12 h or more.

3.7.1. Summary of developmental control genes

In organismal death, there is progressive increase in transcript abundances of some developmental control genes suggesting that they are no longer silenced. A possible reason for these increased abundances is that the postmortem physiological conditions resemble those of earlier developmental stages.

3.8. Cancer genes

There are a number of databases devoted to cancer and cancer-related genes. Upon cross-referencing the genes found in this study, we discovered a significant overlap. The genes found in this search are presented below.

In the zebrafish, transcripts of the following cancer genes significantly increased in abundance: Jdp2, Xanthine dehydrogenase (Xdh), Egr1, Adam12, Myosin-IIIa (Myo3a), Fosb, Jun, Integrin alpha 6b (Itga6), Ier2, Tpr, Dual specificity protein phosphatase 2 (Dusp2), Disintegrin and metallopeptidase domain 28 (Adam28), Tnpo1, Ral guanine nucleotide dissociation stimulator-like (Rgl1), Carcinoembryonic antigen-related cell adhesion molecule 5 (Ceacam1), Fosl1, Il1b, Hif1a, Serine/threonine-protein phosphatase 2A regulatory (Ppp2r5d), DNA replication licensing factor (Mcm5), Gadd45, Myosin-9 (Myh9), Casp3, Tnf, Il8, Cyclic AMP-dependent transcription factor (Atf3), small GTPase (RhoA), Mknk2, Ephrin type-A receptor 7 precursor (Epha7), ETS-related transcription factor (Elf3), Nfkbia, Kpnb1, Wif1, RAS guanyl-releasing protein 1 (Rasgrp), Ras association domain-containing protein 6 (Rassf6), Cyba, DNA-damage-inducible transcript 3 (Ddit3), Serine/threonine-protein kinase (Sbk1) and Tyrosine-protein kinase transmembrane receptor (Ror1) (figure 9).

Figure 9. Abundance of cancer gene transcripts by postmortem time (h): (a) zebrafish and (b) mouse. Green, intermediate value red, maximum value. Bold gene name means it was found in more than one cancer database. The Rgl1 gene was represented by two different probes.

In the mouse, transcripts of the following cancer genes significantly increased in abundance: Retinoblastoma-like protein 1 (Rbl1), Elongation factor RNA polymerase II (Ell), Bcl-2-like protein 11 (Bcl2l11), Sal-like protein 1 (Sall1), Map3k2, Bcl6, Tnfrsf9, CK2 target protein 2 (Csnk2a1), Transcription factor E2f4 (E2f4), Zinc finger DHHC-type containing 14 (Zdhhc14), Tpr, RAS p21 protein activator 1 (Rasa1), Gadd45, Prohibitin (Phb2), Serine/threonine-protein phosphatase PP1-gamma catalytic (Ppp1cc), Lasp1, G protein-coupled receptor kinase 4 (Grk4), LIM domain transcription factor (Lmo4), Protein phosphatase 1E (Ppm1e), Protein sprouty homolog 1 (Spry1), Multiple PDZ domain protein (Mpdz), Kisspeptin receptor (Kiss1), Receptor-type tyrosine-protein phosphatase delta precursor (Ptprd), Small effector protein 2-like (Cdc42), AT-rich interactive domain-containing protein 1A (Arid1a), Lymphocyte cytosolic protein 2 (Lcp2), DNA polymerase zeta catalytic subunit (Rev3l), Tnfrsf14, Integrin beta-6 precursor (Itgb6), Triple functional domain protein (Trio), ATPase class VI type 11C (Atp11c) and Serine/threonine-protein phosphatase 2A regulatory (Ppp2r1a) (figure 9).

3.8.1. Summary of cancer genes

Genes analysed under this category were classified as ‘cancer genes’ in a Cancer Gene Database [10] (figure 9). The timing, duration and peak transcript abundances differed within and between organisms. Note that some gene transcripts had two abundance maxima. In the zebrafish, this phenomenon occurred for Adam12, Jun, Tpr, Dusp2, Tnpo1 and Hif1a genes and in the mouse, Bcl6, Tnfrs9, Lasp1, Cdc42 and Lcp2 genes, and is consistent with the notion that the transcript abundances are being regulated through feedback loops.

3.9. Epigenetic regulatory genes

Epigenetic regulation of gene expression involves DNA methylation and histone modifications of chromatin into active and silenced states [179]. These modifications alter the condensation of the chromatin and affect the accessibility of the DNA to the transcriptional machinery. Although epigenetic regulation plays an important role in development, modifications can arise stochastically with age or in response to environmental stimuli [180]. Hence, we anticipated that epigenetic regulatory genes would be involved in organismal death.

In the zebrafish, transcripts of the following epigenetic genes significantly increased in abundance: Jun dimerization protein 2 (Jdp2), Chromatin helicase protein 3 (Chd3), Glutamate-rich WD repeat-containing protein 1 (Grwd1), Histone H1 (Histh1l), Histone cluster 1, H4-like (Hist1h46l3) and Chromobox homolog 7a (Cbx7a) (figure 10). The Jdp2 gene is thought to inhibit the acetylation of histones and repress expression of the c-Jun gene [181]. The Chd3 gene encodes a component of a histone deacetylase complex that participates in the remodelling of chromatin [182]. The Grwd1 gene is thought to be a histone-binding protein that regulates chromatin dynamics at the replication origin [183]. The Histh1l gene encodes a histone protein that binds the nucleosome at the entry and exit sites of the DNA and the Hist1h46l3 gene encodes a histone protein that is part of the nucleosome core [184]. The Cbx7a gene encodes an epigenetic regulator protein that binds non-coding RNA and histones and represses gene expression of a tumor suppressor [185].

Figure 10. Abundance of epigenetic gene transcripts by postmortem time (h): (a) zebrafish and (b) mouse. Green, intermediate value red, maximum value. Bold gene name means it was found in more than one cancer database. The Jdp2 gene was represented by two different probes.

The transcripts of both Jdp2 and Chd3 genes increased in abundance within 0.3 h postmortem, and reached abundance maxima at 0.5 h. Note that two different probes targeted the Jdp2 transcript. The transcript of the Grwd1 gene increased in abundance at 1 h and 24 h postmortem. The transcript of the histone genes increased in abundance at 4 h postmortem, reaching abundance maxima at 24 h. The transcript of the Cbx7a gene increased in abundance at 12 h, reaching an abundance maximum at 24 h. The transcript abundances of these genes decreased after 24 h.

In the mouse, transcripts of the following epigenetic genes significantly increased in abundance: Tubulin tyrosine ligase-like family member 10 (Ttll10), Histone cluster 1 H3f (Hist1h3f), Histone cluster 1 H4c (Hist1h4c), YEATS domain-containing 2 (Yeats2), Histone acetyltransferase (Kat7) and Probable JmjC domain-containing histone demethylation protein 2C (Jmjd1c) (figure 10). The Ttll10 gene encodes a polyglycylase involved in modifying nucleosome assembly protein 1 that affects transcriptional activity, histone replacement and chromatin remodelling [186]. The Hist1h3f and Hist1h4c genes encode histone proteins that are the core of the nucleosomes [187]. The Yeats2 gene encodes a protein that recognizes histone acetylations so that it can regulate gene expression in the chromatin [188]. The Kat7 gene encodes an acetyltransferase that is a component of histone-binding origin-of-replication complex, which acetylates chromatin and therefore regulates DNA replication and gene expression [189]. The Jmjd1c gene encodes an enzyme that specifically demethylates Lys-9 of histone H3 and is implicated in the reactivation of silenced genes [190].

The transcripts of the Ttll10, Yeats2 and histone protein genes increased in abundance 0.5 h postmortem and reached maxima at different times, with the Ttll10 transcript reaching a maximum at 1 to 6 h, the histone transcripts reaching maxima at 6 and 12 h postmortem, and the Yeats2 transcript reaching maxima at 12–24 h postmortem (figure 10). The transcripts of the Kat7 and Jmjd1c genes increased in abundance at 24 h, reaching abundance maxima at 48 h postmortem. Note that the transcripts of the histone genes were no longer abundant after 24 h postmortem.

3.9.1. Summary of epigenetic regulatory genes

The increased abundance of transcripts of genes encoding histone proteins, histone–chromatin modifying proteins, and proteins involved in regulating DNA replication at the origin were common to the zebrafish and the mouse. These findings indicate that epigenetic regulatory genes are still modifying chromatin structure in organismal death and thus change the accessibility of transcription factors to the promoter or enhancer regions.

3.10. Percentage of gene transcripts with significant abundance by postmortem time

The percentage of gene transcripts was defined as the number of gene transcripts with abundances greater than the control over the total number of transcripts with significant abundance in a category at a specific postmortem time. A comparison of the percentage of gene transcripts by postmortem time of all gene categories revealed similarities between the zebrafish and the mouse. Specifically, most gene transcripts increased in abundance between 0.5 and 24 h postmortem, and after 24 h the transcript abundance drastically dropped (figure 11, ‘All genes’). It should be noted that the same pattern was found in stress, transport and development categories for both organisms. However, in the zebrafish, the immunity, inflammation, apoptosis and cancer categories differed from the mouse. Specifically, the gene transcripts in the immunity, inflammation and cancer categories increased in abundance much later (1–4 h) in the zebrafish than the mouse, and the duration of elevated abundances was much shorter. For example, while 90% of the transcripts for genes in the immunity and inflammation categories increased in abundance in the mouse within 1 h postmortem, less than 30% of the transcripts in the same categories were abundant in the zebrafish (figure 11), indicating a slower initial response. It should be noted that while the number of transcripts of immunity genes reached abundance maxima at 24 h postmortem in both organisms, the number of inflammation genes reaching abundance maxima occurred at 1–4 h in the mouse and 24 h in the zebrafish. The significance of these results is that the inflammation response occurs rapidly and robustly in the mouse while in the zebrafish it takes longer to establish, which could be attributed to phylogenetic differences. There were significant differences in the transcript abundances of apoptosis genes between the zebrafish and the mouse. In the mouse, the percentage of transcripts of apoptosis genes reached 100% at 1 h postmortem and remained sustained for 48 h postmortem, while the percentage of transcript genes with increased abundance in the zebrafish never reached 70% and the abundances abruptly decreased after 12 h.

Figure 11. Percentage of transcripts with increased abundances by postmortem time and category. Number of total genes by organism and category are shown. ‘All genes’ refer to the gene transcripts that significantly contributed to the ordination plots. Mouse is red and zebrafish is black.

3.11. Upregulation or differential mRNA stability?

Since equal amounts of RNA were used for all time points (see below), although degradation was ongoing, it is theoretically possible that the apparent increase in the abundance of a subset of transcripts is actually due to a higher stability of these transcripts compared to the background of degrading transcripts. Hence, the question arises whether higher transcript abundances are due to upregulation after organismal death or complex decay profiles leading to relative enrichments. To determine whether the significant increases were due to such an enrichment, the expected profile of a hypothetical stable non-degrading cRNA was calculated for the zebrafish, mouse liver and mouse brain. In theory, the abundance of a stable non-degrading cRNA transcript determined by the Gene Meter approach should positively correlate to the amount of total cRNA delivered to the DNA microarray. Below is a rationale and the approach used to identify stable non-degrading cRNAs in the transcript pool.

As outlined in the Material and methods section, the amount of sample taken from an animal was approximately the same and the homogenization volume was the same. The electronic supplementary material, tables S1 and S2 show the quantity of total RNA extracted from a tissue (x, ng µl −1 ). Since a fixed amount of RNA was taken into labelling, the volume of the homogenized sample was proportional to 1/x, i.e. the effective quantity of tissue taken into the microarray analysis was proportional to 1/x.

Let us assume there was a subset of stable RNAs, while all other RNA molecules were degrading. Hence, the quantity of the stable gene transcript would be directly proportional to the amount of tissue taken into the experiment, 1/x. In order to provide an expected concentration–time profile for the assumed stable cRNA, one can compare (on the log2 scale) the 1/x values for all time points. To make it relative to the live control, the log2 value of the control will be subtracted from each time point. The obtained profile will be the expected profile (fold change) of a stable non-degrading cRNA.

Taking into account the above considerations, we found that the expected profile of the stable zebrafish cRNA would have an eightfold increase at 96 h postmortem (figure 12). By contrast, at 48 h postmortem, we found that the expected profile of the stable cRNA in mouse liver would have an approximately fourfold decrease, while the stable cRNA in the mouse brain would have an approximately twofold decrease, because less tissue was taken into the microarray analysis (since the total RNA yield increased). It is important to note that these expected stable mouse mRNA profiles (lower two panels of figure 12) would not have been selected by the statistical procedure for identifying transcriptional profiles that significantly increased in abundance relative to the live controls.

Figure 12. Expected fold change of a putatively stable cRNA by postmortem time. Fold change was determined by subtracting the log2 of the inverted concentration in µl ng −1 of the extracted cRNA of the live controls from the inverted concentration of extracted cRNA at each sampling time.

In the zebrafish, the potentially enriched transcripts (due to their stability) were identified by correlating their abundances to the expected fold change of the stable cRNA. In theory, potentially enriched transcripts should be positively correlated to the expected fold change of the putatively stable transcript. Alternatively, transcripts that are not enriched due to stability effects should be negatively correlated or not correlated at all to the expected fold change. The correlations of the 548 gene transcripts (i.e. those that were significantly increased in abundance with postmortem time) ranged from highly negative to highly positive (figure 13). Note higher frequency gene transcripts on the left side of the histogram indicate that most transcripts were negatively correlated to the expected fold change, while the lower frequency on the right side of the histogram indicates a small portion of the gene transcripts were putatively enriched than other transcripts. A standard statistical table, using d.f. = n − 2 with direction, revealed that correlations greater than 0.685 were statistically significant at α = 0.01 (three bars on the right side of the histogram). Hence, 45 of the 548 gene transcripts were found to be significantly correlated to the expected fold change of the stable cRNA.

Figure 13. Distribution of the correlations of the expected fold change and relative gene transcript abundance by postmortem time for the zebrafish. We only considered probes targeting gene transcripts that significantly increased with postmortem time relative to the live controls (n = 548). Correlations above 0.685 were significant at α = 0.01 and indicate the possibility of enrichment of the stable cRNA.

The 45 gene transcripts that were putatively enriched are shown in table 1. Figure 14 shows the transcriptional profiles of three selected gene transcripts compared to expected fold change profile as shown in the top three panels. The transcriptional profile of a negatively correlated gene transcript served as a control. None of the gene transcripts from the mouse samples were enriched because the amount of cRNA in the tissue extract increased or stayed about the same with postmortem time.

Figure 14. Comparison of expected fold change (grey) based on total cRNA extracted relative to live control versus specific gene transcript profiles (black) by postmortem time. (ac) Relative transcript abundances that are highly correlated with expected fold change and are therefore putatively enriched and/or stable. (d) A transcript profile that is negatively correlated with expected fold change and therefore neither enriched nor stable.

Table 1. Positively correlated zebrafish probes and expected fold change at α = 0.01. −, non-annotated gene. The profiles of the probes in bold are shown in figure 14.