Phosphatidylinositol 3-kinases (PI3Ks) were first identified as enzymes associated with various oncoproteins and growth factor receptors 1. Evidence soon accumulated that PI3Ks can provide a critical signal for cell proliferation, cell survival, membrane trafficking, glucose transport, neurite outgrowth, membrane ruffling, and superoxide production as well as actin reorganization and chemotaxis 23. The diverse range of PI3K functional effects reflects its activation by multiple receptors, the existence of a PI3K enzyme family, and multiple effector proteins that can interact with PI3K lipid products by distinct structural motifs 4. Mammalian PI3Ks are divided into three classes based on their structure and substrate specificity 5. Class I enzymes are acutely activated by a variety of cell-surface receptors, and are responsible for synthesis of intracellular phosphatidylinositol (3,4,5)-trisphosphate (PtdIns (3,4,5) P₃) 4. Class I PI3Ks are divided into two types. Class IA consists of a p110 catalytic subunit that associates with an SH2 domain-containing subunit: p85 (and its splice variants: p55 and p50). The p85 subunit is considered the most abundant of the class IA regulatory subunits. The majority of tyrosine kinase coupled transmembrane receptors can activate class IA PI3K, although there is a clear variation in the strength and duration of the resulting signals 2. The only class IB PI3K identified to date is the p110γ catalytic subunit complexed with a p101 regulatory protein and activated by heterotrimeric G proteins. Class IB PI3K is largely observed in white blood cells 6 with some expression in murine prostate, kidney, exocrine pancreas and salivary glands 7. PI3K-dependent accumulation of PtdIns (3,4,5) P₃ initiated a multitude of downstream signals that include suppression of apoptosis, reorganization of the actin cytoskeleton, cardiac myocyte growth, glycogen synthase stimulation by insulin, TNF-mediated neutrophil priming and superoxide generation, and leukocyte migration and adhesion to endothelial cells 28. In contrast to class I PI3Ks, which are mainly cytosolic, class II PI3Ks are predominantly associated with membrane fractions of cells 4. Enzymes of this class are larger than class I or III PI3Ks, bind adaptor proteins via proline rich regions present within their N-terminal sequence 9. The association of class II PI3K enzymes with cellular membrane compartments might suggest participation of these enzymes in sorting events or vesicle formation 10. However, it has recently become evident that a variety of membrane receptors, including tyrosine kinases [insulin, epidermal growth factor receptor (EGFR), platelet derived growth factor receptor (PDGFR)] integrins and a chemokine, monocyte chemotactic peptide-1 (MCP-1), can activate the class II kinases 111213. Based on work with cell lines and Northern blotting, expression of PI3K-C2α, PI3K-C2β isoforms was considered ubiquitous 1415, whilst PI3K-C2γ was mainly found in the liver 1617. Class III PI3K is homologous to the saccharomyces cervisiae Vps34p. It is a constitutively active enzyme whose product phosphatidylinositol-3-phosphate (PtdIns (3) P) plays an essential role in protein trafficking through the lysosome 18.

Extensive insight into the PI3K signalling network has been gained through the use of PI3K inhibitors, overexpression approaches, gene targeting studies and microinjection of PI3K antibodies into cells 2419. Whilst these methodologies have assigned roles for PI3K enzyme activity, there is little evidence for critical non-redundant functions of individual enzyme subclasses. Fundamental to the understanding of PI3K enzyme specificity is therefore a robust definition and comparison of the pattern of PI3K expression in human tissues. The existence of multiple classes of PI3K enzymes and several isoforms within each class (class IA/IB, PI3K-C2α, PI3K-C2β, PI3K-C2γ) suggests significant signalling redundancy potentially manifesting in tissue/cell type specific expression. Identification of tissues in which individual or multiple PI3K isoforms are expressed could provide clues to the role or regulation of PI3K enzymes in vivo and identify more relevant in vitro systems to study these enzymes. Furthermore, recognition of PI3K tissue expression has clinical relevance as it would assist in developing targeted therapies utilising PI3K inhibitors.

In this study, we analyze and describe the topographical and semi-quantitate the level of expression of three PI3K isoforms (IA PI3K (p85), PI3K-C2α and PI3K-C2β), which are considered to be widely distributed (data based on in vitro studies), in a range of normal human tissues using immunohistochemical techniques.

Most of the current literature relies on biochemical methods to provide evidence for the presence of PI3K mRNA or protein in human tissue or cell line lysates 141516. With a single report of p85 expression (in lung cancer) 20 and none relating to PI3K-C2α and PI3K-C2β, there was a need for a systematic study examining topographical expression and precise cellular localisation of PI3K enzymes in human tissues.

Our report illustrates the selective expression of PI3K enzymes in normal adult human tissues, appearing to be restricted to the differentiated cell lineages compared to the proliferative and stem cell compartments. In differentiated cells (epithelial and mesenchymal), expression of p85, C2α and C2β, either singly or in combination, was observed only in a limited repertoire of cells. Differences in expression and in spatial localisation of PI3K isoforms possibly reflect the diverse roles played by these enzymes in various tissues in response to different ligands. In hepatocytes, endocrine organs (thyroid and parathyroid glands and Islets of Langerhans), squamous epithelium and secretory epithelia in general, no p85 or class II PI3K enzymes were detected. This either reflects very low levels of expression under normal in vivo conditions (not detectable by methods employed in this study), or the predominance of alternative PI3K isoforms in these tissues. In the liver, for example, there is biochemical evidence of high expression of PI3K-p50α and PI3K-C2γ in the liver 161721. Marked responsiveness of p50α to insulin suggests that this adaptor plays an important role in the activation of hepatic PI3K 21. Another observation was that those cells expressing only two PI3K isoforms, co-expression of p85 and PI3K-C2α was not observed (Figures 3 and 4). This, together with the unique expression of PI3K-C2α in mesenchymal elements (endothelial cells and smooth muscle cells), myoepithelial and ductal cells suggests a more specialized role for PI3K-C2α compared to p85 or PI3K-C2β. Further in vitro biochemical studies are required to elucidate this role.

PI3K expression was analysed in tissues representative of self regenerating stem cell systems such as skin, intestines, bladder, uterine cervix and testis (which normally show growth and differentiation throughout life), quiescent (stable) tissues such as parenchymal cells of the liver, kidney, pancreas, vascular endothelial cells and thyroid (tissues that show little growth in adult life but can be driven to proliferate by an appropriate stimulus), and of non-dividing (permanent) terminally differentiated cells that have no reserve capacity for growth such as neurones (ganglion cells) and macrophages 22. In self-regenerating tissues, shedding of epithelial cells is compensated by a precursor stem cell population that rapidly divides then selectively differentiates to replace lost cells 23. Proliferation and differentiation are inversely correlated processes where differentiation of primitive into more specialized cells involves increased restriction in proliferative capacity, culminating in cell-cycle exit and arrest 24. Well-defined stem cell compartments have been identified in the GI tract, skin and testis (reviewed in 252627). In these organs, p85, PI3K-C2α and PI3K-C2β were only expressed in fully differentiated cells. With the exception of the apical p85 expression in proliferative endometrium, there was no identifiable expression of PI3K enzymes in stem/progenitor or proliferating cell compartments. The absence of either p85 or class II PI3K immunoreactivity in germinal centres of reactive lymphoid follicles is consistent with our observation that PI3K enzymes are expressed in differentiated, committed cells and not in proliferating or immature cells. PI3K expression in permanent non-dividing, non-epithelial cells such as ganglion cells and macrophages is indicative of a general rather than cell-specific growth regulatory mechanism. Based on in vitro data examining cell cycle progression in cell lines and embryonic stem cells 2829303132, we would have predicted an inverse pattern of expression to what was observed.

Our data suggest an association between PI3K and cellular differentiation. This association is best demonstrated in intestinal epithelium and testicular seminiferous tubules, both of which are models to study the mechanisms of tissue differentiation. In these tissues, spatially organised stem cell, proliferative and differentiated compartments create a "gradient" of cell differentiation. In the intestine and the testis, PI3K expression increased with differentiation. The increase in PI3K expression in these tissues was more dramatic in the case of p85 than for class II enzymes. Previous in vitro data linking PI3K activation to differentiation have been controversial. In the intestine, our data supports the study by Laprise et al 33 who showed that PI3K activity promotes intestinal epithelial cell differentiation and contradicts others 3435 who propose that PI3K inhibition contributes to intestinal cell differentiation. The expression pattern of p85 observed in the testis is supported by data from Kissel et al 36 who showed that Kit-mediated PI3K signalling is critical for the development of male germ cells during the pre-meiotic stages. When these signals were interrupted, male sterility developed due to failure of germ cell to enter meiosis and increased apoptosis of primary spermatocytes 36. Interestingly, we observed nuclear PI3K-C2β immunoreactivity in developing spermatocytes. This pattern of expression was confined to the testis. A role for nuclear PI3K signalling has been recently identified 37. Conditions where the presence of nuclear PI3K-C2β was identified, were associated with compensatory rat liver regeneration and differentiation of HL-60 cell line along granulocytic lineage commitment pathways 3839. The significance of PI3K-C2β expression in the testis and its nuclear localisation is currently unknown.

PI3K expression was observed in immune cells including macrophages and polymorphnuclear leucocytes. The intensity of expression in macrophages frequently exceeded that observed in epithelial cells. Interestingly, only a subset of CD68 positive macrophages (including Kuppfer cells) expressed PI3K enzymes. One possible explanation for this observation is that expression may be related to a specific phase in macrophage function. We are currently examining this possibility. Polymorphnuclear leucocytes expressed moderate-strong PI3K-C2α but not p85. This may be related to the relative abundance of the leukocyte-specific class IB PI3K isoform. In vitro observations show that PI3Ks are important regulators of phagocytosis 40. Class IA and III PI3Ks are thought to act consecutively in phagosome formation and maturation, and that their respective products, PtdIns (3,4,5) P₃ and PtdIns (3) P accumulate transiently at different stages 41. The involvement of class II PI3Ks, which can also produce PtdIns (3) P can not be excluded 41. Although the exact mechanisms involved in receptor coupling remain obscure, MCP-1-induced activation of PI3K-C2α may contribute to some of its cellular responses. More vigorous investigations are needed to determine whether activation of PI3K-C2α or the related PI3KC2β occurs in response to other chemokines and to determine the functional significance of class II PI3K activation in macrophages. Unexpectedly, B-lymphocytes and their precursors were non reactive. A role for p85 in B lymphocyte development and activation was demonstrated in p85α deficient mice in two independent studies 4243. However, the regulation of PI3K activation in p85 α-deficient mice is rather complex. In these knockouts, there is reduced expression of all class IA catalytic subunits with enhanced PI3K-dependent responses following stimulation 44454647. Therefore, extrapolating from these knockout studies to human lymphocyte function is difficult.

The results presented here in suggest a role of PI3K enzymes in differentiated cells, however, they do not rule out a role for these enzymes in proliferation. It is possible that proliferating cells express PI3K enzymes but at levels which were undetectable by methods used in this study. Other more sensitive methods such as real-time PCR could be employed in future studies to explore the above.

One of the limitations of any IHC study is antibody specificity and antigen preservation. Identifying reliable PI3K antibodies for use in IHC has been especially challenging. Several commercially available and in-house class I PI3K antibodies (p85 or p110) including those used in previous IHC studies 2048 did not give satisfactory results in our hands. This was either due to limited sensitivity or reduced specificity (high background or multiple bands on the Western blot). We selected 3 PI3K antibodies (class IA PI3K (p85), PI3K-C2α and PI3K-C2β) that recognised a protein of the appropriate size on Western blotting and identified their target in formalin-fixed paraffin embedded tissues. Reactivity in IHC was specific as evidenced by successful blocking of immunoreactivity by the antigenic peptide fragments to which the antibodies were raised. We have demonstrated that, in contrast with their expression in cell lines of differing origins, class IA PI3K adaptor, p85, and class II PI3K enzymes display a specific and selective tissue distribution. PI3K enzyme expression in fully differentiated cells argues against their role in cell proliferation. Studies have indicated that p85 may act as a signalling protein independent of p110 4950. It is, therefore, conceivable that p85 signals detected in this report reflect p110-related lipid kinase activity and other p110-unrelated functions. The use of reliable p110 antibodies or novel anti-PtdIns (3,4,5) P₃ antibodies 51 should allow more detailed and comprehensive mapping of the PI3K enzymes in normal and diseased human tissues, and provide insight into potential therapeutic targets of isoform specific PI3K antagonists.

None declared.

SSE: Performed experiments (Western blotting, PI3K-p85immunostaining), reviewed and interpreted the data, photomicrography and preparation of the manuscript.

JD: together with ENL had the original idea for the project. Provision of affinity purified antibody to class II PI3K enzymes. Practical supervision of the project. Contributed to the design and relevance of the study, reviewed and interpreted the data.

PA: contributed to the design and relevance of the study and reviewed and interpreted the data.

PT: performed experiments with class II PI3K antibodies including peptide blocking.

GS: reviewed and interpreted the data.

ENL: was the primary supervisor of the project, had the original idea for the project, together with JD contributed to the design and relevance of the study, reviewed and interpreted the data and is the corresponding author.

The conclusions were agreed by all authors.

All authors contributed sections of text to the manuscript, read the manuscript and agreed to the final version.