Genomic clones were isolated under high-stringency hybridization conditions from human genomic libraries using cDNA probes for the nine human PKC isotypes. Genomic PKC clones confirmed by polymerase chain reaction (PCR) (up to five clones per gene) were subsequently used to determine the human chromosomal location of the PKC gene module using FISH. Fluorescent doublets were observed in the majority of the metaphases on each of the two sister chromatids at the positions indicated in Figure 1a, which shows the composite karyotype of the nine distinct PKC genes. No other chromosomal site exhibited significant fluorescent signals (data not shown). The fluorescence intensity of up to eight signal-carrying chromosomes was measured to create a highly significant average green-to-red profile. This profile was linearly interpolated to the size of standard chromosomal ideograms, allowing an objective assignment of fluorescent signals to a chromosomal band. The diagrammatic presentation of fluorescent spot distribution on metaphase human chromosomes is shown in Figure 1b.

Thus, the chromosomal location of all members has been fine-mapped to confirm and/or correct the existing data set. As Table 1 shows, out of the nine human PKC members, five isotypes - PKCα, β, δ, ζ and ι - had not been correctly assigned to individual human chromosomes in the existing literature [5,7,8,9,10,11,12,13]. A comparison of our chromosomal locations for PKC genes with the genome-draft assignments (HUGO), confirmed our results overall; however, our detailed FISH analysis provided enhanced mapping resolution.

Using HUGO and bioinformatics tools, we have dissected the genomic organization, that is the characterization of the exon/intron structure, of the nine PKC isotype genes (Figure 3). The PKC loci range in size from approximately 24.4 kilobases (kb) (PKCγ) to 480 kb (PKCα), and are composed of 14 to 18 coding exons and 13 to 17 introns, varying in size from 94 to 188,435 base pairs (bp). The exons are small to intermediate in size, ranging from 32 to 381 nucleotides. All intron-exon junctions appear to comply with the GT-AG rule. Considering the common evolutionary origin of this gene family (as shown by a cDNA-based phylogenetic tree of PKC and its closest relative, PKD, Figure 3a), it is not surprising that some of their exon structures are identical (and nearly homologous in their amino-acid sequence) within the PKC subfamilies (Figure 3b). The AUG translation initiation site for the ORFs of PKCα, β, γ, ε, ζ and η is located in exon 1, for PKCδ, θ and ι one intron is located in the 5'-untranslated region (5'-UTR) and only the subsequent exons determine the functional domains of the PKC proteins. Nevertheless, the existence of additional exons within the 5'-UTR cannot be excluded at this point.

These structural genomic data can be used to represent the phylogenetic relationships of the PKC genes. The organization of the regulatory and catalytic subdomains of PKC has been remarkably preserved during evolution: using the exon structure of PKCα as a reference, exons 10-15 within the catalytic domain share the highest similarity in size, organization and primary structure among conventional (c)PKC and novel (n)PKCs, but not atypical (a)PKCs. aPKCζ, and ι are very distinct from cPKC and nPKC within the exon structure of the catalytic domain. Along this line, aPKCι and aPKCs appear particularly conserved between each other in both their regulatory and catalytic domain structures. Within the regulatory domain, the subfamilies cPKC, nPKC and aPKC are clearly distinct from each other, and there is even a split within the nPKCs into the D-forms, PKCδ and θ, and the E-forms, PKCε and η. Also, cPKCα and cPKCβ appear more closely related to each other than to cPKCγ. Overall, the data suggest that gene duplication followed by intron loss and/or insertion and exon shuffling have been important in the evolution of the PKC gene family. This careful in silico dissection of the gene fine structure is also a prerequisite to an efficient search for potential abnormalities in the PKC genes.

The existence of alternative splice variants of PKC within certain tissues is an attractive option for providing further heterogeneity in PKC, reflecting the diverse pathways of signal transduction within these cells.

Alternative splicing is currently known for at least one human PKC isotype. Two distinct PKCβ cDNA clones have been isolated that encode sequences of 671 and 673 amino acids, respectively, which differ from each other only in the carboxy-terminal regions of approximately 50 amino acids [14,15]. Characterization of the PKCβ chromosomal gene provided direct evidence of the existence of two adjacent carboxy-terminal exons that could be alternatively spliced to generate two types of PKCβ [16]. Importantly, these splice variants, PKCβI and βII, appear to be expressed in a tissue-selective manner, suggesting that their functions are different.

A mouse variant of PKCδ - PKCδII - has recently been described [17]. This has a 78 bp (26 amino acid) insertion into the caspase-3-sensitive site in the V3 domain of the original PKCδI sequence, which renders the PKCδII isoform insensitive to caspase-3. PKCδI is detected in most tissues, whereas PKCδII was expressed selectively in the testis, ovary, thymocytes, brain and kidney [18].

In addition, a carboxy-terminal truncated form of PKCδIII exists in rats, which has an 83 bp out-of-frame insertion at the same site in the V3 region [18]. Genomic DNA analysis revealed that the difference between mouse PKCδII and rat PKCδIII is due to a different sequence at the critical 5' donor splice sites (data not shown). PKCδIII, which represents just the regulatory domain, might show a dominant-negative effect against PKCδI, and thus alternative splicing involving this variant might modulate signaling pathways.

Two forms of rat PKCζ RNA, with different 5' ends, have been reported. The major form (PKCζ wild type) is expressed ubiquitously, whereas the smaller form - protein kinase M ζ (PKMζ) - which encodes just the catalytic domain of the enzyme without most of its regulatory domain, is predominant in normal brain and certain rat prostate tumors. The rat PKCζ gene locus appears to contain two alternative promoters from which it can be transcribed to give two transcripts with 5'-end heterogeneity [19].

Finally, a PKCθII cDNA clone has been isolated from mouse testis, encoding a unique 5' sequence of 20 amino acids and the PKCθI (= wild type) sequence of 444 amino acids. The transcription of PKCθII RNA is initiated from the PKCθII-specific exon, which is located between exons 7 and 8 of the PKCθI gene, indicating that alternative splicing is the mechanism by which PKCθII is generated. PKCθII is expressed exclusively in the testis in an age-dependent manner with sexual maturation. Consistent with its lack of a C1 regulatory domain, PKCθII is constitutively active and may have a crucial role in spermatogenesis [20].

The authentic PKC gene loci present in a single copy in the human genome have been analyzed in detail for single-nucleotide polymorphisms (SNPs) using resources from the National Center for Biotechnology Information website [21]. Among the 11 SNPs within the coding regions of the PKC gene module currently known from these databank searches, only two (one each in PKCδ and PKCη, respectively) give rise to non-synonymous codons and create missense mutations (Table 2). These missense mutations do not exert an effect on kinase function in any obvious way, even though an effect cannot be ruled out at this time. Overall, the human PKC genes appear not to be highly variable regions as characterized by SNP analysis; however, the available exonic SNPs appear sufficient to be used for haplotyping purposes in linkage and/or association studies.

Using the Cancer Genome Anatomy Project [22], a comparison of the chromosomal location of these nine fine-mapped PKC loci with human disease loci revealed chromosomal aberrations/breakpoints in these regions in, for example, patients with malignant disease (Figure 4). In particular, the locations 17q24 (PKCα), 19q13 (PKCγ), 3p21 (PKCδ), 2p21 (PKCε), 1p36 (PKCζ) and 3q26 (PKCι) are noteworthy, as deletions or balanced translocations involving these defined chromosomal regions are frequently described in naturally occurring malignancies. Thus, in theory, PKC-family genes could be affected as possible disease candidate genes in accidental recombinations during intrachromosomal rearrangement or even interchromosomal translocations, as in the deleterious joining of abl sequences to the immunoglobulin or bcr loci that leads to malignancy (via gain-of-function). However, many other genes involved in the regulation of cell activation, proliferation and apoptosis are also included in these chromosomal regions. Nevertheless, our mapping results suggest that some PKC isotypes could be candidate genes involved in human cancer.

Members of the PKC family have been consistently implicated in aberrant signaling responses contributing to malignant transformation, for example, as intracellular receptors for tumor-promoting phorbol esters, which have been shown to protect various cells, including T cells, from apoptosis [23,24]. As a result of this potentially transforming capacity of PKC-family genes, the aberrantly high expression levels of distinct PKC isotypes found in most tumor cell lines argue for a functional link between PKC and oncogenesis (G.B., unpublished work). Chronic recruitment of PKCθ (by an undefined mechanism) into the membrane fraction of malignant cells has been reported in cell lines derived from patients with T-cell leukemia [25]. Other, more definitive, results show that both Bcr-Abl and PKCι activity are necessary for resistance to apoptosis in hematopoietic K562 cells, supporting a functional role for PKCι in the survival of leukemia cells [26]. Recent studies, including our own [23,27], have directly linked distinct PKC isotypes to molecular pathways regulating apoptosis. But in spite of the putative role of PKCs in cellular growth control and differentiation, no clinical example of a causative role of PKCs in primary cell malignancy has yet been published. It remains to be seen whether the genetic dissection of naturally occurring malignant somatic cell mutants will eventually establish a link between PKCs and clinical disease.

A directed search for potential loss-of-function (and most likely recessive) mutations in the PKC loci, potentially associated with distinct genetic syndromes, will now be initiated. Along with information from biochemical work on signal transduction, tissue-specific expression patterns and phenotypes of the currently established mouse loss-of-function PKC isotype knockout lines ([28,29,30,31,32,33] and G.B., unpublished work), the genomic fine mapping reported here will enable genetic studies in defined groups of patients to search for functional PKC polymorphisms or mutations associated with familial genetic defects or abnormalities. Given the enormous increase in genetic and molecular databases, bioinformatics approaches should continue to improve and develop into useful tools for evaluating hypotheses, with only the very promising ones being subjected to empirical testing. This human genetic, and therefore long-term, approach may aid in delineating the basic physiological and possible pathophysiological functions of the PKC gene module, and may illuminate whether and eventually how PKCs are involved in human genetic disease.