S100 and annexin proteins identify cell membrane damage as the Achilles heel of metastatic cancer cells

Abstract

Mechanical activity of cells and the stress imposed on them by extracellular environment is a constant source of injury to the plasma membrane (PM). In invasive tumor cells, increased motility together with the harsh environment of the tumor stroma further increases the risk of PM injury. The impact of these stresses on tumor cell plasma membrane and mechanism by which tumor cells repair the PM damage are poorly understood. Ca²⁺ entry through the injured PM initiates repair of the PM. Depending on the cell type, different organelles and proteins respond to this Ca²⁺ entry and facilitate repair of the damaged plasma membrane. We recently identified that proteins expressed in various metastatic cancers including Ca²⁺-binding EF hand protein S100A11 and its binding partner annexin A2 are used by tumor cells for plasma membrane repair (PMR). Here we will discuss the involvement of S100, annexin proteins and their regulation of actin cytoskeleton, leading to PMR. Additionally, we will show that another S100 member – S100A4 accumulates at the injured PM. These findings reveal a new role for the S100 and annexin protein up regulation in metastatic cancers and identify these proteins and PMR as targets for treating metastatic cancers.

Keywords:

• actin
• annexins
• Annexin A2
• breast cancer
• cancer
• plasma membrane repair
• S100 proteins
• S100A11
• Metastasis

Introduction

The phospholipid bilayer of the plasma membrane surrounds and physically separates the interior structures of the cell from the extracellular environment. It is selectively permeable to ions and organic molecules and yet maintains an essential osmotic barrier to the outside. Loss of this barrier function due to PM injury poses critical threat to single as well as multicellular organisms and thus cells rapidly repair PM lesions. Even in the shielding tissue environment where the tissue architecture provides the protective environment to restrict the injury by damaging forces, PM disruptions are frequent. Prevention of damage can be achieved by dynamic adaptations at a single cell or tissue level to limit the level of imposed stress, e.g., by actively modulating the epithelial layer to relieve mechanical forces. Still, many cells experience plasma membrane ruptures on a recurring basis that they need to cope with to maintain cell and tissue integrity. This is particularly evident in cells from, e.g., skin, lungs, gastrointestinal tract, skeletal and heart muscles, which reside in areas with high mechanical activity and hence increased damage frequency. Muscle and lung cells offer a good example of this as they are routinely wounded as a result of exercise and over stretching.^1,2 Defect in the muscle cell's ability to repair has been shown to result in muscular dystrophy,^3-6 where poor repair of injured muscle cell membrane leads to cell death and tissue inflammation. Poor plasma membrane repair (PMR) is also associated with Niemann-Pick type A,⁷ diabetes,⁸ and Chediak-Higashi⁹ syndrome and therapies targeting PMR have been shown to be effective in treating muscle and lung injuries.^10-12 While much of the focus has been on disease resulting from poor PMR, we have recently identified that improved PMR is an important contributor for cancer metastasis.¹³ Human breast cancer cells that turn metastatic through up-regulation of truncated EGF receptor (ErbB2) also need to up-regulate their PMR machinery, and absence of enhanced PMR compromises their invasive ability. Here we will briefly discuss the mechanisms involved in PMR, with a focus on the involvement of annexins, actin and S100 proteins. Additionally, we will describe the involvement of this process in cancer metastasis and discuss the potential for harnessing this novel aspect of tumor metastasis for developing new therapeutic approaches to target metastatic cancers.

The Membrane Repair Machinery

Studies over the past 2 decades have revealed that plasma membrane repair is a complex and active process that require membrane replacements, fusion events and cytoskeletal reorganization.¹⁴ PMR is triggered by Ca²⁺ influx at the injury site due to over a thousand-fold gradient of calcium that exists across the plasma membrane.¹⁵ The calcium influx triggers a versatile repair system that involves replacing or patching the injured membrane.¹⁶ Fusion of intracellular vesicles around the wound perimeter to form a patch was revealed using the sea urchin egg.¹⁷ This formed the basis for the patch model, according to which Ca²⁺ entry at the wound site triggers recruitment and homotypic fusion of vesicles at the wound site which then fuses with the plasma membrane to seal the wound.^17,18 In the sea urchin eggs these vesicles were identified as the yolk granules. In case of mammalian cells identification of lysosomes as the vesicles that undergo Ca²⁺-triggered fusion with the injured membrane led to the proposal that these are the patch forming vesicles in the mammalian cells.^19,20 Additionally, injury-triggered fusion of non-secretory vesicles called enlargeosomes,²¹ and accumulation of mitochondria at the site of PM injury²² have also been found to be required for PMR.

In cells wounded by small pore forming toxins, PMR progresses by the replacement of the damaged cell membrane, such that the injured membrane area is physically removed by endocytosis,²³ blebbing,²⁴ or ectocytosis.²⁵ Shedding and endocytosis of damaged cell membrane by vesicles has also been shown for lesions formed by the membrane attack complex (MAC).^26,27 Lysosome exocytosis has been shown to be important for removal of injured membrane through the secretion of Acid Sphingomyelinase (ASMase).²³ Secreted ASMase hydrolyses the plasma membrane sphingomyelin to ceramide causing endocytosis of the pores in the plasma membrane formed by the pore forming toxins.²³ Cells from NPA patients, lacking ASMase protein, or Limb Girdle Muscular dystrophy 2B (LGMD2B) patients (lacking dysferlin protein), who show slow and poor injury-triggered ASMase secretion both show compromised PMR.^4,23 The PM repair deficit in these patient cells can be rescued by providing sphingomyelinase.⁴ ASMase is also known to trigger cell membrane shedding and could contribute to the shedding of membrane by ectocytosis.^25,28,29 Thus, ASMase secreted by lysosome exocytosis may be involved in all the 3 processes (exocytosis, endocytosis, and ectocytosis) involved in PMR.

Recently, it was identified that vesicular shedding of damaged plasma membrane is not limited to pore forming toxins, but small focal injury can also trigger Endosomal Sorting Complex Required for Transport (ESCRT) III-mediated shedding of the injured PM.³⁰ We find that shedding of damaged membrane is required even following larger focal injury. This process is initiated by the EF hand calcium binding protein Apoptosis linked gene (ALG)-2, which accumulates at the injured PM and through binding its partner ALIX (ALG-2 interacting protein X) it facilitates accumulation of ESCRT III complex, resulting in cutting and shedding of damaged cell membrane.³¹ Annexins are another group of calcium-binding protein that are implicated in sensing PM injury and formation of membrane blebs to shed the membrane damaged by pore forming toxin.³² The members of the annexin protein family interact with the S100 proteins and anionic phospholipids to promote membrane segregation, vesicle trafficking, and vesicle fusion in a Ca²⁺-dependent manner.^33,34 Annexin and S100 proteins are also known regulators of actin cytoskeleton.^35,36

Actin cytoskeleton associated with the plasma membrane is another key regulator of PM repair. The cortical actin causes membrane tension, which prevents spontaneous resealing upon injury.³⁷ Ca²⁺ influx at the injury site triggers depolymerization of cortical actin as observed in Xenopus oocytes and Drosophila Embryos,³⁸ which is followed by resynthesis of the cortical actin. This progresses in way of a contractile actomyosin ring that forms a “purse string” - a circular structure and constricts circumferentially, in a manner coincident with the recruitment of filamentous actin (F-actin) and myosin-II at the wound borders.³⁹ Disruption of the actin cytoskeleton by either Cytochalasin D or Latrunculin B prevents actomyosin ring assembly and impairs wound closure.³⁹ PMR in Drosophila embryos also involves formation of actomyosin complex and a plasma membrane plug that is rapidly recruited from the surrounding edges of the membrane.³⁸ Additionally, intracellular vesicles are recruited to the wound perimeter enabling formation of a membrane patch within the actin ring to seal the wound.^14,38

Annexin and S100 Proteins in Membrane Repair

Annexin protein family, believed to have originated a billion years ago, has members in all major eukaryotic phyla.⁴⁰ Annexins have evolved extensively and independently in several eukaryotic lineages into a varied family.⁴⁰ This family of proteins appear to be instrumental in dealing with membrane stress - plant annexins are up-regulated during abiotic stress response and help to cope with it.^41,42 In humans there are 12 different annexin proteins with orthologs in most vertebrates.^40,33 They contain a unique COOH-terminal core domain that consists of 4 preserved structural repeats on which type-2 Ca²⁺ binding domains are located. Upon binding Ca²⁺, annexins bind the negatively charged phospholipids of the membrane to form a ternary complex bridging adjacent membranes.⁴³ The NH2-terminal region of annexins vary in length and sequence between family members and enables the individual protein to interact with distinct cytoplasmic partners such as the calcium binding S100 proteins.⁴⁴

S100 proteins are small (10-14 kDa), EF-hand-type Ca²⁺-binding proteins that upon Ca²⁺ activation exert both intracellular and extracellular functions. S100 genes are exclusively found in vertebrates and are clustered on chromosome 1q21 in humans (S100A1–S100A16).⁴⁵ The majority of the protein family members form symmetric noncovalent homodimers - a unique feature of the S100 proteins among the EF-hand protein family.⁴⁶ These proteins undergo a change in conformation upon binding calcium, which exposes a hydrophobic domain that can interact with the NH2-terminal region of specific annexins.⁴⁴ This interaction facilitates close apposition of adjacent phospholipid membranes and promotes membrane fusion.³³ Additionally, several pair of S100-annexin complexes such as S100A10 and annexin A2 (S100A10–ANXA2) can bind cytoskeletal components and have been associated with intracellular vesicle fusion.⁴⁷ S100A10 and ANXA2 are known to exist as a heterotetrameric complex where an S100A10 dimer resides in the center of the complex, interconnecting 2 Annexin A2 molecules.⁴⁴ Similarly, annexin A1 and S100A11 (also called S100C or calgizzarin) interact in a temporal Ca²⁺-dependent manner.⁴⁸ Several other pairs of annexin and S100 proteins have been discovered and it seems plausible that some S100 proteins can bind several annexins to exert their biological roles.⁴⁹

Annexins were first implicated in the process of PMR by gene expression analysis of muscle from LGMD2B mouse model, which identified injury-dependent interaction of ANXA1 and ANXA2 proteins with dysferlin.⁵⁰ Dysferlin is the muscle protein, lack of which results in poor repair of muscle fibers.³ Due to the known role of annexins in aggregating membranes it was proposed that interaction of dysferlin and annexins may facilitate PMR through aggregation and fusion of intracellular vesicles.⁵⁰ However, work using zebrafish suggests that these dysferlin vesicles may be derived from the PM.⁵¹ The first demonstration of a role of annexins in PMR in human cells was offered for ANXA1.⁵² ANXA1 was shown to accumulate at the wound perimeter and use of ANXA1 inhibiting antibody, peptide, or dominant-negative mutant inhibited PMR.⁵² Subsequently role of ANXA1 was identified in repair of plasma membrane injured by pore forming toxins, here ANXA1 was shown to facilitate shedding of the injured plasma membrane through the formation of blebs containing the membrane pores.²⁴ Annexins A5 and A6 have also been implicated in the process of PMR.^51,53 The ability of ANXA5 to assemble into 2-dimensional arrays at the sites of membrane injury in response to Ca²⁺ has been implicated in preventing wound expansion by keeping the membrane edges together. This is prevented in cells lacking ANXA5 causing PMR defect. Addition of exogenous recombinant ANXA5 protein rescues this PMR defect.⁵³ ANXA6 on the other hand forms a structure termed as “repair cap” at the site of injury, which facilitates healing of injured muscle fibers.⁵⁴ ANXA6 was also shown to be involved in shedding of microvesicles containing portions of the PM with the lesions formed by streptolysin O. Here, ANXA6 was shown to be activated and recruited to the site of injury at lower Ca²⁺ concentration as compared to ANXA1 suggesting a potential for sequential recruitment of annexins to facilitate PMR.⁵⁵

Enhanced PMR is Needed for Tumor metastasis - Role of Annexin and S100 Proteins

Cellular membranes of cancer cells are known to be destabilized – we have previously shown lysosomal membrane of tumor cells are more fragile^56,57 and that death induced by lysosomal membrane pemebilization is prominent in cancer cells.⁵⁶ This fragility has been harnessed for the development of novel therapeutics.^58,59 Similar to lysosomal membrane, PM of cancer cells are more unstable and have reduced stiffness, which is caused by an increase in saturated phospholipids.⁶⁰ The PM stiffness is inversely correlated with the ability of the cancer cell to migrate and invade 3-dimensional matrix.⁶⁰ The greater motility of metastatic cells and increase in mechanical stress due to invasion of the tissue matrix can increase PM damage. We recently identified this to be the case, as PM of invasive cancer cells suffered greater damage.¹³ In response, we find these invasive cancer cells enhance their PMR by up regulating expression of S100A11 and annexin A2 proteins.¹³ Annexins and S100 proteins are commonly up regulated in variety of cancers.^61,62 Specifically, S100A11 is overexpressed in several tumors, where it is associated with metastasis and poor disease prognosis.^63-65 S100A11 is enriched in pseudopodia of metastatic cancer cells and is required for forming actin-dependent pseudopodial protrusions, which facilitates migration of the tumor cells.⁶⁶

To study the involvement of PM damage and repair in cancer cells as they enhance their motility and invasiveness, we made use of the MCF7 human breast cancer cell model. Here invasiveness was modulated by ectopically expressing a truncated ErbB2 oncogene (p95ErbB2).⁶⁷ The p95ErbB2 oncogene mimics constitutively active cleaved form of ErbB2 oncoprotein commonly found in aggressive breast cancers.⁶⁸ Increased invasiveness of these cells resulted in greater damage to the PM of these cells. Signaling through p95ErbB2 caused increase in the expression of the S100A11, a protein that we find enhances repair of the injured PM. Upon injury, S100A11 co-accumulates with ANXA2 at the site of PM repair. Co-accumulation of these proteins at the repair site is mutually dependent and is independent of ANXA1 - the other binding partner of S100A11. While ANXA1 is also recruited to the injured cell membrane, it localizes away from S100A11-ANXA2 complex. Greater Ca²⁺-dependent binding of S100A11 with ANXA2 as compared with ANXA1⁶⁹ could be responsible for this differential response of the 2 annexins. In resting cell, ANXA1, ANXA2 and S100A11 are all predominantly cytosolic and the PM is supported by a layer of cortical actin. The PM-associated cortical F-actin is necessary to support the plasma membrane, but it also creates tension that can inhibit passive resealing of the PM after injury.⁷⁰ Following injury, cortical actin depolymerizes at the site of injury, arguably by actin severing proteins. This reduces cortical tension and avoids further damaging the injured PM. Also, by the collapse of the neighboring wounded edges these membranes are brought together to facilitate their eventual fusion. As such, removal of cortical actin can facilitate repair of the wounded membrane.⁷¹ We find that PM injury-triggered ANXA2 and S100A11 accumulation is followed by a rapid buildup of F-actin at the site of repair.¹³ S100A11–ANXA2 can bind F-actin and decrease the depolymerization rate of preformed actin filaments.^72,73 Thus, these proteins preserve existing F-actin and allow buildup of new F-actin around the injury site ( Fig. 1 ). Absence of either of these proteins prevents F-actin buildup following injury and prevents PMR, similar to the drugs that alter actin polymerization or depolymerization.^13,74 Cortical F-actin deploymerization around the injury site is followed by the recruitment of ANXA1 at the damaged PM and excision of this damaged membrane. Loss of ANXA2 and S100A11 and pharmacological inhibition of F-actin buildup prevent PMR by blocking excision of the damaged part of the PM marked by ANXA1. Thus, actin remodeling is required for the tumor cells to excise the injured part of the PM indicating that ANXA2 and S100A11 facilitate PMR, in part, by regulating F-actin buildup and excision of the damaged part of the PM at the site of injury.¹³

Figure 1. Mechanism and need of plasma membrane repair in metastatic tumor cells. In the process of invading tumor stroma the metastatic tumor cells experience increased damage to their PM which is facilitated by the increased expression of proteins such as Annexin A2 and S100A11. These proteins regulate Ca²⁺-dependent F-actin growth, which provides support to the repairing plasma membrane and potentially facilitates actin based accumulation of repair vesicles derived from e.g. PM or endosomes providing membrane to repair the damaged membrane. Lack of annexin A2 or S100A11 proteins compromises these activity and result in poor PM repair of the invading cell. This triggers calcium overload and leakage of the cytoplasmic content eventually leading to the death of the invading cell. Thus regulating these and other proteins involved in PM repair in metastatic cells could provide an avenue to target metastatic potential of the tumor cell. ECM: Extra cellular matrix.

ANXA2 has Ca²⁺ dependent phospholipid binding ability, which enables aggregation of endosomes and other vesicles.⁷⁵ We found that upregulation of S100A11 and ANXA2 does not enhance injury triggered fusion of the endosomes/lysosomes.¹³ However, in light of the ability of these proteins to nucleate polymerization of cortical F-actin at the endosomes⁷⁶ they may facilitate accumulation of endosomal vesicles at the site of injury. The buildup of the cortical F-actin together with presence of vesicular endomembrane at the wounded edges of the plasma membrane will facilitate PMR at the repair site marked by the S100A11–ANXA2 complex and excise the damaged part of the PM marked by ANXA1 ( Fig. 1 ). The buildup of cortical actin is analogous to F-actin drawstring formation during healing of injured Xenopus oocyte cell membrane.³⁹ Thus, by analogy we believe F-actin buildup may pull the wounded membrane edges together during excision and vesicle aggregation and F-actin assembly mediated by the S100A11–ANXA2 complex may help with PMR by facilitating vesicle fusion and cortical actin buildup. Additionally, it may help with vesicle fusion, as Ca²⁺-regulated F-actin dynamics at the cell membrane facilitates vesicle fusion.⁷⁷ These findings demonstrate that invasive cancer cells are dependent on efficient PMR system to cope with elevated rate of cell injury and that they rely on the S100A11–ANXA2 complex to facilitate plasma membrane repair.

PMR as Target for Therapeutic Intervention

While the diseases linked to altered PMR are caused primarily due to a defect in this process,³⁷ tumor metastasis offers a counter example to such diseases.¹³ This presents an interesting therapeutic scenario since lessons learned from most of the PMR associated diseases could be applied to control tumor metastasis. In one such recent study of LGMD 2B, we identified that dysferlin protein facilitates injury-triggered secretion of the lysosomal enzyme Acid Sphingomyelinase (ASMase).⁴ This leads to poor repair of LGMD2B patient muscle cells, a defect that is rescued by providing this enzyme function exogenously.⁴ Such a role of ASMase suggests that inhibiting ASMase would be an attractive therapeutic target against cancers. Thus, it is interesting that in an independent study we identified acid sphingomyelinase (ASM) inhibition as a target for cancer therapy.⁵⁸ While in the use of ASMase inhibitors we focused our attention on destabilization of the lysosomal membrane, our findings with the LGMD2B patient cells suggest that such an inhibition may have additional unrealized benefit in the cancer cells by inhibiting PMR. Targeting this and other regulators of PMR is a novel therapeutic approach to control tumor metastasis, one whose potential is yet to be properly explored. In order to realize this potential it would be valuable to identify specific molecular regulators of PMR that tumor cells rely upon. Our results that inhibiting PMR by depletion of S100A11 in the metastatic cells makes them unable to invade 3D tissue matrix offers support for the utility of such a targeted strategy.¹³

S100 and annexins proteins are among the most frequently dysregulated proteins in neoplasia and are overexpressed in various cancers.⁶¹ Thus members of these proteins families are attractive candidates to compromise PMR in tumor cells. As discussed above members of the annexin family - annexin A1, A2, A5 and A6 have already been implicated in PMR and over expression of some of these annexins is directly correlated with aggressive clinical stage in colorectal, pancreatic and brain tumors and linked to metastatic progression.⁷⁸ Further, changes in annexin A3 and A4 expression has been associated with chemo resistance in ovarian cancer cells.^79,80 These findings make a case for a potential therapeutic approach against tumor metastasis that could involve targeting PMR by regulating expression of specific annexins. However, in view of the sequence homology of the annexins and use of multiple annexins in PMR, suggests the likelihood of functional redundancy between the family members. In line with this, changes in the expression of one annexin can profoundly affect the expression levels of another suggesting a strict functional fail-safe mechanism.⁸¹ Thus, efficient pharmacological strategies to target annexins may require that several family members are inhibited simultaneously, e.g. by targeting the conserved annexin core domain. Alternatively, annexin function can be compromised by restricting the interaction with its S100 protein binding partners, e.g., by blocking the interaction between ANXA2 and S100A11. Additionally, in light of our identification of actin as a key mediator of PMR through the action of S100A11– ANXA2 complex proteins, regulation of actin polymerization offers another set of targets. Such targets have an additional advantage of inhibiting metastasis by inhibiting the motility and thus invasiveness of the cancer cells.

Aside from annexins, S100 proteins offer another potential target. These proteins are implicated in multiple stages of cancer and are commonly upregulated and associated with tumor progression in various cancers.⁶² However, with S100A11 as an exception¹³ their direct regulatory role in PMR has not yet been characterized. Other S100 family members may also function as Ca²⁺ triggered switches that can regulate PMR through interaction with other annexins at the plasma membrane. To this end, we looked at the response of S100A4, which is a well-documented metastasis-promoting protein that is overexpressed in various types of cancers and exerts its function both intra- and extracellularly.^82,83 Our analysis of S100A4 by live cell imaging of PMR in HeLa cells show that injury triggers S100A4 co-accumulation with S100A11 at the injury site ( Fig. 2A ). In a manner similar to S100A11, S100A4 accumulation precedes the accumulation of F-actin at the site of repair ( Fig. 2B ). It remains to be established whether S100A4 can facilitate F-actin accumulation or PMR in invasive cancer cells. If so, in view of the prevalence of metastatic cancers associated with S100A4 overexpression, S100A4 would be an attractive candidate to broadly target inhibition of tumor metastasis by regulating PMR.

Figure 2. S100A4 and S100A11 are recruited to the site of injury before actin build-up. HeLa cells were transiently transfected to express (A) S100A4-GFP and S100A11-RFP or (B) S100A4-GFP and the F-actin reporter protein Utrophin-mCherry. The cells were injured focally by a pulsed laser (white arrow) and the response of the proteins was monitored live as the cells underwent repair in the same way as we described previously.¹³ The plot shows the kinetics of accumulation of individual proteins at the repair site (blue arrow) marked by the white box.

A common feature of the S100 proteins is the presence of evolutionarily primitive Ca²⁺-binding EF-hand domains. Calpains are another member of the EF -hand protein family implicated in cancer progression and PMR.^84,85 We recently identified yet another member of the EF -hand protein family (Apoptosis linked gene 2 - ALG-2) that can regulate PMR.³¹ While Calpains regulate various aspects of tumor progression,⁸⁴ their involvement in PMR in cancer cells has not been elucidated. In case of ALG-2 we found that it facilitates PMR by enabling shedding of damaged cell membrane through the Ca²⁺-triggered accumulation of the ESCRT III complex at the injured cell membrane.³¹ The ESCRT machinery is involved in membrane curvature and cleavage.⁸⁶ While the role of ESCRT machinery in cancer is disputed,⁸⁷ metastatic tumor cells are long known to make use of membrane cleavage to shed their PM.^88,89 Shedding of PM by tumor cells is gaining wider recognition due to their role in the process of cell-cell communication through the use of extracellular vesicles.⁹⁰ Thus, our finding that PMR of metastatic cells involves excision and shedding of PM may extend beyond cell survival to extracellular communication that facilitates tumor growth and metastasis. We hope this new aspect of tumor metastasis and its potential involvement in tumor cell invasion and signaling will attract strong translational interest leading to the development of new cancer therapeutic strategy targeting this Achille's heel of cancers.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgment

We thank Birgitte Grum-Schwensen and Noona Ambartsumian for the S100A4-GFP plasmid, University of Copenhagen.