Integrin Signaling, Cell Survival, and Anoikis: Distinctions, Differences, and Differentiation

Cell survival and apoptosis implicate an increasing complexity of players and signaling pathways which regulate not only the decision-making process of surviving (or dying), but as well the execution of cell death proper. The same complex nature applies to anoikis, a form of caspase-dependent apoptosis that is largely regulated by integrin-mediated, cell-extracellular matrix interactions. Not surprisingly, the regulation of cell survival, apoptosis, and anoikis furthermore implicates additional mechanistic distinctions according to the specific tissue, cell type, and species. Incidentally, studies in recent years have unearthed yet another layer of complexity in the regulation of these cell processes, namely, the implication of cell differentiation state-specific mechanisms. Further analyses of such differentiation state-distinct mechanisms, either under normal or physiopathological contexts, should increase our understanding of diseases which implicate a deregulation of integrin function, cell survival, and anoikis.


Introduction
A complex assortment of regulatory systems is required for the embryogenesis and ontogeny, as well as for the maintenance, renewal, and repair, of the comprehensive array of organs and tissues which allow multicellular organisms like mammals to function and survive. One of such critical regulatory systems is programmed cell death (PCD) [1][2][3][4][5][6], best defined as being "a sequence of events based on cellular metabolism that lead to cell destruction" [1]. Such a laconic definition of PCD includes nevertheless three distinct processes of cell death: apoptosis (caspase-induced or caspase-independent), autophagic cell death (caspaseinduced or caspase-independent), and programmed necrosis (caspase-induced or caspase-independent) [1,2,[7][8][9][10]. The present paper will focus on specific aspects of the "caspaseinduced apoptosis" side of PCD, particularly with regards to integrin-mediated signaling.
Thus, the fate of a cell depends principally on a tightly modulated balancing act between the anti-and proapoptotic activities from multiple Bcl-2 homologs [6,8,11,12,[14][15][16][17]. Such a balance is established at its base through a modulation of the expression of various homologs according to the specific survival stimulus, cell type, and species concerned. However, post-transcriptional and/or post-translational modulations (e.g., alternative splicing, phosphorylation, sequestration) also contribute significantly to this balance [11,12,[14][15][16][17][24][25][26]. For instance, the phosphorylation of Bad on either of the S112, S136, or S155 residues inactivates its sensitizer functions, whereas phosphorylation of at least two of these residues furthermore leads to its degradation by proteasome. Another example is the case of Bid, which is normally expressed as an inactive precursor, and which consequently requires cleavage into its tBid form in order to enact its activator functions. Also of note are Bim and Bmf, two homologs that are typically sequestered in microtubules and actin microfilaments, respectively, therefore requiring their liberation from such captivity in order to perform their pro-apoptotic functions [11,12,[14][15][16][17][24][25][26][27][28].
It is of note that the establishment of the apoptosome (and consequent CASP-9 activation) constitutes what is generally recognized as the intrinsic, or "common," pathway of apoptosis. This mode is typically induced by pro-apoptotic stimuli such as the loss of survival signals or critical failures in the operation of cellular processes (e.g., high-error DNA replication, plasma membrane peroxidation, DNA damage, protein misfolding, mitochondrial dysfunction) [1-12, 17, 18]. However, there is another mode of induction of apoptosis known as the extrinsic (or "death receptor," "receptormediated") pathway. This mode is induced by one type of pro-apoptotic stimulus, namely, the binding of "death ligands" (e.g., tumor necrosis factor (TNF)-α, FasL, or TNFrelated apoptosis-inducing ligand/TRAIL) to specific "death receptors." Once activated, these receptors lead to the recruitment of adaptor proteins such as FADD ("Fas-associated Death Domain protein"), which in turn recruit the precursor/inactive form of the initiator caspase CASP-8 in order to form the DISC ("Death Inducing Signaling Complex"). Consequently, this results in the activation of CASP-8 [10-12, 17, 29-32]. In cells with a high density of death receptors and/or high expression of pro-CASP-8, the DISC-induced activation of CASP-8 will be of sufficient intensity to bring about a subsequent amplifying activation cascade of executioner caspases, therefore rendering the process irreversible. However, in cells with a low density of death receptors and/or low expression of pro-CASP-8, the DISC-induced activation of CASP-8 will be instead of low intensity, in turn leading to low-intensity executioner caspase activation. In this situation, CASP-8 (and/or executioner caspases) will then have to (a) cleave Bid into its activator-functional tBid form, (b) destabilize actin microfilaments and microtubules by cleavage to allow the release of Bmf and Bim, and/or (c) cleave microtubule-sequestered Bim in order to directly release it. These altogether lead to a subsequent shift in the balance of anti-and pro-apoptotic Bcl-2 homolog activities, in effect causing a translocation of the death signal to mitochondria for the formation of the apoptosome and activation of CASP-9-and thus only then rendering the process irreversible [10-12, 14-17, 29-32]. The relevance of these considerations will become apparent in the following sections.
Integrins can be divided into three main functional groups: cell-cell adhesion integrins, vascular integrins, and cell-ECM adhesion integrins. It is those integrins that have the β1 subunit in common which constitute the majority of receptors for ECM components [39][40][41][42][43]. Also of this group is the α6β4 integrin, which is expressed exclusively in epithelial cells [39,44]. The repertoire of integrins (and variants) expressed by a given cell is dependent on the contexts of ECM composition, tissue-type, and species concerned [33][34][35][36][37][38][39][40][41][42][43][44]. Taking into account that some ECM components may be bound by more than one integrin, it is therefore not surprising that integrin-mediated cellular responses to the ECM are quite varied and multifaceted [39][40][41][42][43][44]. To this effect, the initial observations that a cell's anchorage to its ECM constitutes a critical factor for its survival [45,46]   Additionally, the stimulation of the PI3-K/Akt and/or Raf/MEK/Erk pathways generally results in an upregulation of Bcl-2 expression for the anti-apoptotic homologs and a down-regulation for the pro-apoptotic homologs, as well as an upregulation of IAP expression and other apoptosis suppressor molecules, such as c-Flip. Alternatively, Fak and/or Src will participate in the recruitment of paxillin, talin, vinculin to assemble focal adhesions in direct association with actin microfilaments. Similarly, ILK will be engaged by β1 integrins to assemble and regulate focal adhesions along with its PINCH and parvin partners, whereas ILK can also contribute in microtubule assembly/stability. The overall stabilization of the cytoskeleton enables the sequestration of the pro-apoptotic homologs Bmf and Bim. Note that various other survival signaling pathways stimulated by integrin β1/Fak/Src signaling (e.g., p130 CAS -Nck-PAK, and PKC) are not shown here for reasons of clarity and conciseness. Pα/β/γ, parvin α, β, γ; Pax, paxillin; PINCH1/2, particularly interesting new cysteine-histidine rich protein-1/-2; Tal, talin; Vin, vinculin.
Overall, signal transduction by integrins differs little from that of growth factor receptors with tyrosine kinase activity ("RTKs"). It is therefore not surprising that there is much chatter between RTK-and integrin-mediated signaling, the two often cooperating in the regulation of various cell processes such as proliferation, migration, and survival [12, 33, 34, 39, 43, 44, 58-73, 77, 79-87]. To this effect, integrins can enact "inside-out" activation of RTK signaling largely through their own engagement of Fak, Src and/or ILK [62-73, 77-79, 82-87]. Reciprocity being the rule, RTKs can perform inside-out activation of integrin signaling, mainly through their own engagement of Src [64-66, 69-71, 86, 87]. Therefore, the already wide range of integrin-mediated signals can be expanded vastly further through such crosstalk with various growth factor receptors.
The precise molecular mechanisms governing the promotion of cell survival by integrins remain to be fully elucidated. Nonetheless, PI3-K/Akt and Ras/Raf/MEK/Erk constitute the best known cell survival-promoting pathways among the assorted ones identified so far as being engaged by integrin-mediated Fak/Src signaling [12, 33-36, 47, 58-66, 68-70, 72, 73, 80, 81, 88-101]. With regards to the PI3-K/Akt pathway, Akt is well known for its numerous cell survival functions such as the inhibitory phosphorylation of Bax (on the S184 residue), Bad (on S136 preferentially, but on S112 as well), and Bim (on S87), as well as the phosphorylation of pro-CASP-9 (on S196) to suppress its activation and the inhibitory phosphorylation of GSK-3β ("Glycogen Synthase Kinase-3β"; on S9) which, when active, positively phosphorylates Bax (on S163) and negatively phosphorylates Mcl-1 (on S155, S159 and T163) ( Figure 2) [11, 12, 14-17, 24-26, 34, 81, 88-94, 101]. Although ILK contributes in the formation of integrin-mediated focal adhesions by binding directly to the cytoplasmic tail of the β subunits of activated receptors [76][77][78], it has also been shown to be PI3-K-dependent for its activation and capable of acting as an upstream contributor to the activation of Akt [82-84, 88, 89, 91]. However, with the recent evidence that ILK is a pseudokinase [102,103], its previously tacit kinase activation upon binding the cytoplasmic domain of integrin β subunits, as well as its kinase signaling roles in promoting integrin-mediated cell survival, remain therefore contentious [82-84, 102, 103]. At the very least, the scaffolding functions of ILK in focal adhesion assembly and actin microfilament linkage [76][77][78], as well as its implication in microtubule assembly [104], are likely to contribute in the sequestration of Bmf and Bim (Figure 2). Similarly, the said scaffolding functions of ILK can be used by integrins to cross-talk with RTKs, thus allowing for its contribution in RTK/integrin cooperative signaling in promoting cell survival [77,78,[82][83][84][85][86].
Hence, the importance of Fak and/or Src in the integrinmediated promotion of cell survival is intimately linked with their pivotal role in the engagement of pathways such as PI3-K/Akt and/or Ras/Raf/MEK/Erk (Figure 2), depending on the composition of the signaling cassettes they help engender and the specific integrin receptors implicated, as well as according to the cell type and species studied [12,34,43,47,48,[58][59][60][61][62][63][64][65][66][67][68][69][70][71][72][73]80]. Interestingly, Fak and Src can also contribute directly in the promotion of cell survival, aside from their usual roles in the integrin-mediated assembly of focal adhesions and engagement of survival-promoting pathways ( Figure 2). Indeed, Src can phosphorylate pro-CASP-8 (on Y380) to suppress its activation [108], whereas Fak can bind RIP1 ("Receptor-Interacting Protein-1"; a.k.a. RIPK1) in order to prevent its recruitment of FADD and consequent formation of the DISC [109,110]. All in all, it is therefore more than ever undisputable that a cell's integrinmediated anchorage to its ECM constitutes a powerful, multilayered, and complex device whose most significant purpose is to ensure survival. It should be noted here that the already complicated nature of the regulation of cell processes is strikingly enhanced by the implication of specific kinase isoforms and/or family kinase members, therefore often resulting in distinct-or even differing-roles by the pathways in which they participate. PI3-K isoforms complexes (4 catalytic subunit isoforms, 3 regulatory subunit isoforms), Akt isoforms (−1 to −3), RSK isoforms (−1 to −4), IRS isoforms (−1 to −4), or Shc (p46 Shc , p52 Shc , p66 Shc ) and Raf (A-to C-) family members, constitute but a few examples identified to date in humans [71, 94-96, 98, 101, 105-107, 111-113]. That these can be selectively expressed depending on the tissue and species, in addition to the fact that they can perform specific functions even within the same given cell type, further emphasizes the reality of the vast intricacies that underlie the regulation of cell processesincluding of course cell survival, whether integrin mediated or otherwise.

Integrins and Anoikis: Making Sure That Letting Go Really Means the End
Considering the importance of integrin-mediated cell-ECM interactions in driving cell survival, it is therefore not surprising that the disruption, or loss, of integrin binding induces PCD-namely, a form of caspase-dependent apoptosis that is termed anoikis (a.k.a. "detachment-induced apoptosis" or "integrin-mediated death") [10, 12, 33-36, 46, 47, 49, 60, 81, 101, 108, 114-120]. Like apoptosis and PCD in general, anoikis performs important roles during organogenesis, as well as in tissue maintenance and renewal [33-35, 38, 49, 79-81, 101, 117-122]. For example, the involution of mammary glands and the renewal of the epidermis, as well as that of the intestinal epithelium, implicate the induction of anoikis in obsolete cells [101,[121][122][123][124][125][126]. In this respect, it is now recognized that cells are endowed with an innate/default anchorage-dependent surveillance system meant to ensure that all integrins expressed by them do interact with their respective ECM ligands, thus inducing anoikis when that is not the case. In other words, any cell that strays accidentally or otherwise from its assigned position within its given tissue, either by interacting with a "wrong" ECM or by losing anchorage to its own ECM, is targeted for death [12,36,60,108,115,117]. Anoikis constitutes overall a "four-punch hit" against cell survival that implicates elements from both the intrinsic and extrinsic pathways of caspase-dependent apoptosis. The first "punch" comes from the deactivation of Fak and/or Src, leading to a disengagement of pathways such as PI3-K/Akt and Ras/Raf/Mek/Erk (Figure 3). Hence, the numerous pro-survival roles performed by these pathways (Figure 2) undergo failure. The second "punch" comes from the concomitant disassembly of anchoring focal adhesions (and/or hemidesmosomes), in large part by the loss of integrinmediated engagement of Fak, Src, and/or ILK (Figure 2), thus greatly destabilizing the cytoskeleton and consequently allowing the release of Bmf and Bim ( Figure 3). Additionally, the presence of p66 Shc (a member of the Shc family of adaptor proteins [113]) at focal adhesions can induce an elevated/sustained activation of the small GTPase RhoA, which in turn contributes likewise in the destabilization of the cytoskeleton through effectors that remain to be fully identified [127]. Interestingly, such "anoikis-sensitizing" function of p66 Shc appears to be independent of its better known ability to translocate to mitochondria that is typically part of its apoptosis-inducing roles [113,127].
The fourth "punch" delivered against cell survival by the loss of integrin binding consists in the induction of the extrinsic pathway of apoptosis-that is, CASP-8 activation (Figure 3) [12,36,81,101,108,115,[117][118][119][120]. Although the activation of CASP-8 constitutes an early/immediate event following the loss of integrin-mediated cell adhesion, the precise mechanisms that are responsible for such activation remain poorly understood. A part of the puzzle lies with previous observations that both pro-CASP-8 and activated CASP-8 are associated with the cytoplasmic tails of β1 and/or β3 subunits of unligated integrins [152,153]. Interestingly, pro-CASP-8 is found already associated with the cytoplasmic tails of integrin β subunits under healthy/adhering conditions [108,153], most likely due to the fact that pro-CASP-8 often complexes with Src following its inhibitory phosphorylation by the latter [108]. Nonetheless, the formation of a DISC at unligated integrins as the causal agent for CASP-8 activation remains contentious, despite reports from different cell types that CASP-8 activation can be FADDdependent in anoikis [109,110,[154][155][156][157][158][159][160] and that c-Flip can inhibit the process [156,160,161]. Indeed, with Fak and Src being downactivated following detachment from the ECM, RIP1 is freed from Fak [109,110], and pro-CASP-8 is no more negatively phosphorylated by Src (see previous section) [108,162]. Additionally, the concomitant downactivation of the PI3-K/Akt and/or Ras/Raf/MEK/Erk pathways leads to a down-regulation of c-Flip expression (see previous section and Figure 3) [12,60,81,[116][117][118]120]. Thus, RIP1, FADD and pro-CASP-8 can complex to form the DISC [109]. However, it turns out that Fak-freed RIP1 forms a FADD-dependent DISC at the cytoplasmic domain of the death receptor Fas [110]. In hindsight, that is to be expected given that, in some cell types, the expression of FasL can be upregulated following the loss of integrin binding, or that anoikis can be attenuated/blocked by the inhibition of Fas [12, 81, 109, 110, 115, 116, 118-120, 154, 155, 160, 161]. Yet in the end, RIP1 does not form a DISC at unligated integrins [110]. Furthermore, it has been shown that FADD is not found associated with the cytoplasmic tails of β3 subunit-containing unligated integrins in human umbilical vein endothelial cells undergoing anoikis, despite the fact that pro-CASP-8 and activated CASP-8 are [152].
Most perplexing is what has been reported recently in human keratinocytes. In these cells, bound β1 integrins contain the β1A subunit variant, whereas the β1B variant is expressed in the cytosol and apparently not part of any functional heterodimeric receptor [153]. As can now be expected (see above), pro-CASP-8 is found associated with the cytoplasmic tails of β1A subunits when those cells are adhering [153]. Upon loss of ECM binding, β1A integrins are internalized so as to colocalize with the cytosolic β1B subunits, whereby pro-CASP-8 somehow shuttles from the tails of the former to those of the latter in order to undergo Journal of Signal Transduction 9 activation in a FADD-independent manner [153]. Hence, how exactly pro-CASP-8 and/or activated CASP-8 associate with β subunits of unligated integrins, as well as the precise mechanisms responsible for CASP-8 activation at these sites, remain open questions. Although it is not yet established whether DAPK1 can contribute to the formation of a DISC [143][144][145][146][147][148], the fact that activated DAPK1 binds to the cytoplasmic tail of β subunits of integrins it has deactivated [149][150][151] definitively raises suspicions to that effect, especially in light of the previous observations in keratinocytes. Then again, it may simply be that the mechanisms of CASP-8 activation following the induction of anoikis differ according to the combined contexts of the cell type, of the integrin repertoire expressed (including variants), and of the species concerned-not unlike everything else that has been discussed herein so far.
In any case, the activation of CASP-8 following the loss of integrin binding is typically of low-intensity. As a result, this leads to a likewise low intensity activation of executioner caspases (Section 2). Hence, anoikis implicates the necessary following events: (a) caspase-mediated cleavage of Bid into its active tBid form; (b) caspase-mediated cleavage of pro-survival kinases (e.g., Fak, Akt) for deactivation; (c) caspase-mediated liberation of Bmf and Bim (Figure 3) [12,36,81,101,108,115,[117][118][119][120]. Moreover, MEKK1 ("Mitogen-activated protein kinase kinase kinase 1"), an upstream activator of the JNK pathway, may be cleaved by executioner caspases (namely, CASP-7) in order to generate a constitutive active form which is responsible for a consequent prolonged/sustained activation of JNK (JNK1 presumably, [128,137,138]), who then undertakes apoptotic phosphorylating functions (see above and Figure 3). Consequently, these events altogether effect a translocation of the death signal to mitochondria for the formation of the apoptosome and activation of CASP-9 ( Figure 3). Interestingly, anoikis in normal cells is characterized by a delay in the irreversible commitment to the process following the triggering of its initial pro-apoptotic events [12,46,79,81,101,114,120,128]. This explains why cells can be rescued from anoikis by reattachment within a limited time-span subsequent to their loss of anchorage [12,46,79,81,101,114,120,128]. Such a "window of anoikis reversibility" varies in length depending on the cell type and species, thus ranging from some fifteen minutes to four hours [12,46,79,81,101,114,120,128]. It is thought that a window of anoikis reversibility depends primarily on the time-span required for the death signal to reach mitochondria, according to the specific integrins (and variants) implicated, the precise determinants of cell survival/death in play (e.g., Bcl-2 homolog expression profiles, survival pathways/isoforms engaged, apoptotic kinases/isoforms involved, etc.), and the degree of cytoskeletal organization/cell polarity imparted not only by integrins (i.e., focal adhesions, hemidesmosomes), but as well by cell-cell interactions (e.g., E-cadherins). Indeed, the latter sensitize cells to anoikis by further linking/organizing the cytoskeleton, albeit contributing at the same time to cell survival signaling [12,46,81,114,115,128,163,164]. Additionally, it has been observed that the activation of pro-survival kinases such as Src, Akt and/or Erk1/Erk2 undergoes a short (five to fifteen minutes in length), transient up-activation following the loss of anchorage. While the mechanisms responsible for these "deathly gasps" remain unclear, they are first and foremost considered to act as a protection against transient detachments from the ECM, such as those required during normal cell processes like migration or cytokinesis. Accordingly, these may also contribute in defining a window of anoikis reversibility [12,46,79,81,114,120]. However, considering the recent evidence that an upactivation of MEK/Erk in normal cells can actually trigger the intrinsic apoptotic pathway [165,166], or activate DAPK1 and a consequent autophagic cell death [148,149,165,166], the question arises whether such "deathly gasps" may not in fact represent "poisoned fruits." Be that as it may, it is clear that once CASP-9 activation occurs, any window of reversibility is shut and anoikis becomes irreversibleso much so, in fact, that attempts from this point on to inhibit the process result in Beclin-1-driven autophagic cell death (see above) [2,[7][8][9][10][118][119][120]. To this effect, this may explain why the overexpression of Bcl-2 or Bcl-X L alone can protect cells from anoikis [12, 36, 53, 81, 114, 117-120, 128, 137, 167], as such overexpression would not only manage to block the translocation of the death signal to mitochondria following detachment, but furthermore prevent the trigger of autophagic cell death by an overwhelming inhibitory binding to Beclin-1.
Given that normal cells require a sound anchorage to their ECM in order to live, and in view of the "fourpunch hit" described above that occurs following the loss of integrin-mediated attachment, it is therefore usually considered that integrins suppress anoikis [12,81,101,[120][121][122][123]. This is well exemplified by the observations that the forced expression of dominant negative mutants of Fak and/or Src readily induces anoikis, whereas the forced expression of constitutive active mutants of either, or both, protects against it [12, 60, 63-73, 81, 99-101, 114, 119, 120, 168]. Likewise, similar results are obtained when mutants of kinases that are engaged by integrin/Fak/Src-mediated cell survival signaling are used, such as PI3-K or Akt [12, 60, 63-73, 81, 81, 88-99, 99, 100, 100, 101, 101, 114, 119, 120]. However, there is now evidence that some integrins may actually sensitize cells to anoikis. As example, the knockdown of the expression of α8β1 in human intestinal epithelial crypt cells results in the loss of vinculin at focal adhesions and confers a measure of anoikis resistance via an illicitly sustained activation of Fak [169]. Although it remains unclear by which mechanisms Fak activation is thus sustained, it is germane that vinculin has been shown to enforce the adhesion-dependent activation/deactivation of Fak [60,67,170,171] and that the gene disruption of vinculin in F9 mouse embryonic carcinoma cells renders them resistant to anoikis, also via a sustained activation of Fak [170]. Another example of anoikis-sensitizing integrin is αvβ3 in colon cancer cells, although it is not known how this integrin enacts such a sensitizing function [172,173]. It is thought that anoikis-sensitizing integrins may be part of the above-mentioned anchorage-dependent surveillance system, as a means to further warrant the disposal of cells that detach from their ECM [169]. However, it is already clear that anoikis-sensitizing functions will be undertaken by specific integrins according to the cell type. Indeed, αvβ3 is better known for its anoikis-suppressing roles in endothelial cells, among other cell types [174,175], whereas α8β1 has already been reported to suppress anoikis in myofibroblasts [176].
Overall, the preceding considerations altogether demonstrate the multilayered complexities that underlie the regulation of cell survival, apoptosis, and anoikis. Moreover, these include further mechanistic distinctions according to the contexts of the composition of the ECM, the specific integrins implicated, the cell type, and the species. However, studies in recent years in one tissue in particular, the human intestinal epithelium, have unearthed yet another level of intricacy in the regulation of these processes, namely, the implication of distinct mechanisms according to the state of cell differentiation.

Differentiation State-Specific Regulation of Cell Survival, Anoikis: Difference Is the Key
The intestinal epithelium is an elegant and valuable physiological system for understanding the functional connections between integrin-mediated cell-ECM interactions and the cell state [101,125,[177][178][179][180][181]. The continuous renewal of this simple columnar epithelium occurs along a well-defined unit, the crypt-villus axis. This unit consists generally in two cell populations: the proliferative, immature cells of the crypt and the differentiated cells of the villus [125,126]. As part of the dynamic process of intestinal epithelial renewal, obsolete enterocytes enter anoikis upon reaching the apex of the villi, as a means of exfoliation [101,125,126,180]. For their part, crypt cells can occasionally undergo apoptosis in order to evacuate daughter cells that are damaged or defective [101,125,126,180]. Such apparent contrast of destiny between undifferentiated and differentiated enterocytes along the crypt-villus axis, coupled with their specific profiles of expression of Bcl-2 homologs which are established during the differentiation process [101,125,[182][183][184][185][186], initially introduced the concept of a distinct modulation of cell survival and apoptosis according to the state of differentiation [101,125,184]. This concept so far has been demonstrated mostly in human intestinal epithelial cells [101,125]. For instance, the individual expression of Bcl-2 homologs is subjected to specific regulatory mechanisms depending on the differentiation status of enterocytes [167,[185][186][187]. As example, the PI3-K/Akt-1 and MEK/Erk pathways are selectively implicated in the promotion of enterocytic survival according to the state of differentiation, including with regards to their modulation of the expression/activity of Bcl-2 homologs (Figure 4) [167,[185][186][187]. In this respect, PI3-K/Akt-1, but not MEK/Erk, is critical for the survival of undifferentiated/crypt enterocytes, whereas both pathways are vital for the survival of differentiated/villus ones ( Figure 4) [167,[185][186][187].
The obvious follow-up question is whether the concept of a distinct modulation of cell survival and apoptosis according to the state of differentiation applies as well to integrinmediated cell survival and anoikis. In effect, human intestinal epithelial crypt and villus cells express differential profiles of integrins as they interact with specific basement membrane components, which are likewise deposited differentially along the crypt-villus axis [125, 177-181, 184, 188]. As expected when considering the normal fate of villus cells (see above), it turns out that differentiated/villus enterocytes are more susceptible to anoikis than their undifferentiated/crypt counterparts [185,189,190]. Although crypt cells express the α8β1 receptor as an anoikis-sensitizing integrin [169], the villus cells are nonetheless highly polarized in addition to bearing E-cadherin adherens junctions, in stark contrast to the former [163]. Moreover, such distinctions in anoikis susceptibility between crypt and villus enterocytes translate into differentiation state-specific mechanisms of integrinmediated cell survival and anoikis ( Figure 4) [167,185,186,[189][190][191]. For instance, β1 integrins, Fak, and Src distinctively modulate the expression/activity of Bcl-2 homologs depending on the enterocytic differentiation status (Figure 4) [167,185,186,190]. Furthermore, α2β1, α5β1, and α6Aβ1 are required for the survival of undifferentiated enterocytes, whereas α3β1, α6Bβ1 and α6Bβ4A are required for the survival of differentiated ones [185,186] (S. Thibodeau and P.H. Vachon, manuscript in preparation). Similarly, the engagement of PI3-K/Akt-1 is integrin β1/Fak/Srcdependent in undifferentiated cells, but Src-independent in differentiated ones, whereas the engagement of MEK/Erk remains integrin β1/Fak/Src-dependent regardless of the state of differentiation ( Figure 4) [167,185,186,[189][190][191]. To this effect, the integrin β1/Fak/Src/PI3-K/Akt-1 pathway antagonizes the apoptotic activation of p38β in undifferentiated enterocytes, whereas the integrin β1/Fak/PI3-K/Akt-1 and integrin β1/Fak/Src/MEK/Erk pathways both contribute in antagonizing that of p38δ, in differentiated ones ( Figure 4) [167,186,189].
Consequently, the regulation of integrin-mediated cell survival and anoikis is indeed subjected to differentiationstate-specific mechanisms ( Figure 4). However, the overall concept of differentiation state-specific controls of cell survival, apoptosis, and anoikis does not constitute a peculiarity of the intestinal epithelium, since it evidently applies to other tissues-albeit not without obligatory cell type-and/or species-dependent distinctions. For instance, skeletal muscle myoblasts require fibronectin and α5β1A in order to survive, whereas myocytes require instead laminin-211 and α7Bβ1D [50][51][52][53]. Furthermore, myoblasts depend on an integrindriven Fak/Src/MEK/Erk pathway for their survival, whereas myocytes depend instead on an integrin-driven Fyn/PI3-K/Akt-2 one [52,53,192,193]. A fundamental challenge now presenting itself lies in the search for answers to three broad questions regarding such differentiation statespecific mechanisms of cell survival, apoptosis and anoikis, namely: (a) the "why" for the existence of such distinct mechanisms according to the state of cell differentiation; (b) the "how" of these distinct mechanisms (i.e., the further functional identification of specific extracellular ligands, integrins and molecules/pathways involved, as well as their respective differentiation-specific roles in the suppression or induction of apoptosis/anoikis); (c) the "in what capacity" these same differentiation-specific mechanisms contribute in the emergence of diseases when they become deregulated.  Figure 4: Differentiation state-specific mechanisms of integrin β1/Fak/Src-mediated control of human intestinal epithelial cell survival and anoikis. In undifferentiated enterocytes (a), β1 integrin/Fak signaling recruits Src which acts as a cornerstone in the subsequent engagement of PI3-K/Akt-1 and MEK/Erk in the suppression of anoikis. However, in contrast to PI3-K/Akt-1, MEK/Erk is not required for survival due to its non-to-marginal roles in the regulation of the expression/functions of Bcl-2 homologs and its noninvolvement in suppressing the activation of the enterocyte undifferentiated state-selective apoptotic kinase isoform p38β. In differentiated enterocytes (b), MEK/Erk remains Src dependent but not PI3-K/Akt-1 (which however remains Fak-dependent). PI3-K/Akt-1 and MEK/Erk are now both required for survival, as they both play major roles in the regulation of Bcl-2 homolog expression/functions as well as in suppressing the activation of the enterocyte differentiated state-selective apoptotic kinase isoform p38δ. +, up-regulation of expression; −, down-regulation of expression.

Integrins, Anoikis, and Disease: Too Much or Not Enough, It's Still Unhealthy
As in the case of apoptosis, there is an increasing number of pathological disorders that are characterized by a deregulation of integrin-mediated cell survival, and anoikis signaling [35,39,44,49,117,118,120,194]. Certain forms of muscular dystrophy and epidermolysis bullosa constitute classic examples of pathologies that are caused primarily by a deregulated induction of anoikis [39,44,79,80,[195][196][197]. Accordingly, the need to understand the regulation of cell survival and apoptosis has gained much importance in cancer research, especially since the acquisition of a resistance to anoikis constitutes a critical step in tumor progression-particularly in the emergence of invasive and metastatic cells [86, 116, 118-120, 174, 188, 198-200]. Incidentally, cancer cells are notorious for exhibiting major alterations in their repertoire of expressed integrins [49,62,80,86,174,180,188,199,200], as well as in their surrounding ECM [35,49,180,188,200]. Moreover, metastatic cells are infamous for displaying a marked resistance to anoikis [86, 116, 118-120, 174, 188, 198-200]. To this effect, much attention has been given to Fak and Src in tumor progression [63][64][65][66][67][68][69][70][71][72][73]201], as well as to downstream pathways engaged by them, such as PI3-K/Akt and MEK/Erk [88][89][90][91][92][93][94][95][96][97][98][99]. Along with these, virtually every known player in the intrinsic and extrinsic pathways of apoptosis is likewise being scrutinized, in order to not only seek an eventual understanding of the acquisition of resistance to anoikis (and/or apoptosis) by cancer cells, but as well as to identify molecular targets that are susceptible to shut down such problematic resistance. Thus, anoikissensitizing integrins may be added to this expanding list, in view of the observations that anoikis-resistant colon cancer cells do not express α8β1, whereas forcing its expression reinstates in them a good measure of susceptibility to anoikis [169]. Similarly, colon cancer cells that are anoikissensitive acquire a resistance to the process following their loss of expression of αvβ3 [172]. However, the cross-talk between integrins and RTKs further complicates matters greatly, since the deregulated activity of an RTK can confer resistance to anoikis by maintaining/enhancing integrinmediated cell survival after the loss of attachment, via inside-out signaling [12,34,86,87,116,117,119,120,174,175]. As one example among many, colon cancer cells that strongly express EGF/TGF-α ("Epidermal Growth Factor/Transforming Growth Factor-α") display a resistance to anoikis by at least in part sustaining the activation of Src via autocrine stimulation of EGFR ("EGF Receptor"), therefore allowing Src to maintain functional interactions with Fak and consequently sustaining a Fak/Src-mediated suppression of anoikis [202]. In this respect, the deregulation of RTK signaling is well known to induce epithelialmesenchymal transition (EMT) in normal epithelial cells, a phenotypic change that is considered a prerequisite for the acquisition of anchorage-independent growth and resistance to anoikis, consequently leading to the emergence of invasive and metastatic cancer cells [62, 66-73, 85-87, 116, 119, 120, 200, 201, 203].
Hence, the task of elucidating the mechanisms of resistance to anoikis in cancer cells, let alone identifying commonalities, is evidently proving to be a daunting oneand that is without taking into account the inherent molecular differences between the various types of cancers, as well as the cellular and molecular heterogeneity that is invariably found within any given cancer. Nevertheless, increasing our knowledge of those determinants which control cell survival, apoptosis, and anoikis within a specific tissue will, in turn, add further to our comprehension of apoptosis/anoikis-linked diseases of said tissue, in addition to paving the way toward the design of targeted molecular therapeutic approaches that are both tissue-and diseaseselective. To this end, the inclusion of acquiring a better grasp of differentiation state-specific distinctions in the integrin-mediated regulation of cell survival and anoikis in such research endeavors would not only allow for a more complete understanding of a given tissue's normal physiology, but furthermore provide a full accounting of the physiopathological underpinnings of its apoptosis/anoikislinked diseases-including tumorigenesis and tumorigenic progression. Incidentally, cancerous intestinal epithelial cells display integrin/Fak/Src anoikis-suppressing signaling features that are similar to either their normal undifferentiated or differentiated counterparts, and do so regardless of their own degree of cell dedifferentiation [202]. Whether or not cancer cells of other tissues do likewise remains, however, to be confirmed.

Post-Mortem and Conclusions
Cells require survival signals in order to live, and interactions with their ECM through integrins constitute a critical contributor to this effect. Integrin-mediated cell survival and anoikis comprise a multifaceted and multilayered surveillance mechanism which is responsible for upholding the correct position of cells within their respective tissues, thereby sentencing to death any cell that would stray from its assigned position by either interacting with an inappropriately composed ECM or by losing anchorage to its own ECM. Thus, anoikis partakes greatly in the configuration and preservation of the proper functional organization of tissues, thereby ensuring the disposal of cells that assume a roguish behavior with regards to their precise integrin-ECM adhesion requirements for survival-whether this occurs accidentally in the course of normal physiological processes, such as tissue renewal, or otherwise.
It is undeniable that the regulation of cell survival and anoikis implicates distinct mechanisms according to the composition of the ECM, the repertoire of integrins (and variants) expressed, the degree of cell polarity (i.e., cytoskeletal organization) conferred at least in part by specific ECM-integrin interactions, the transduction pathways engaged and/or suppressed through signaling by individual integrins, the isoforms and/or family members participating in such pathways, the RTKs expressed which may, or may not, interfere with their intrusive cross-talk, and the various regulators/effectors of apoptosis that are expressed. Although there may be a measure of mechanistic commonality between some cell types, one should always be mindful of the reality that the integrin-mediated control of cell survival and anoikis constitutes a complex business that is not only tissue type-and species context-dependent, but as well differentiation state-distinctive. Taking into account these caveats should allow for an improved focusing on the precise determinants of cell survival of a given tissue within the whole of its physiological framework, therefore facilitating an eventual elucidation of the selective integrinmediated underpinnings that are implicated in tissuespecific apoptosis/anoikis-linked pathological disorders.