CD13 activation assembles phosphoinositide (PI) signaling complexes to regulate the actin cytoskeleton

Abstract

Activation of transmembrane CD13 enables diverse cellular processes such as cell–cell and cell–matrix adhesion, endocytosis, and recycling of cell surface proteins by assembling and tethering protein complexes at the plasma membrane. Here, we identify a novel CD13-dependent protein assembly that regulates phosphoinositide (PI) signal transduction to impact actin dynamics and induce cell protrusions capable of propagating signals to distant cells. In response to cellular stress, the CD13-expressing human Kaposi’s sarcoma-derived cell line (KS1767, KSCs) formed elongated protrusions that extend above the substrate and link non-adjacent cells, which is significantly diminished in CD13^KO KSCs. Activation of CD13 with stimulating mAbs markedly induced protrusion formation with a striking accumulation of CD13 and actin at the base. Further, these membrane-delimited bridges in WT KSCs can transfer Ca²⁺ signals between connected cells via Connexin 43⁺ gap junctions. Mechanistically, CD13-mediated protrusion formation requires activation of CD13, Src, FAK and Cdc42 to tether the IQGAP1 and ARF6 complex at the membrane. This activates phosphatidylinositol-4-phosphate-5-kinase (PI5K) to increase local phosphatidylinositol 4,5-bisphosphate (PI(4,5)P2) levels, promoting actin-polymerization and membrane protrusion. Therefore, CD13 is a novel molecular PIP regulator, modulating signal transduction and downstream cellular processes, including actin cytoskeleton dynamics and membrane organization to facilitate intercellular communication.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-35022-6.

Introduction

Acquiring, interpreting, and exchanging molecular information is essential for the survival of all multicellular organisms. Communication between cells, as well as between cells and their microenvironment, is fundamental to preserving tissue homeostasis and restoring normal function upon tissue damage. Cells decipher extracellular signals primarily via specific membrane receptors and molecules that receive and transmit instructions to the cytoskeleton through the lipid bilayer. The cytoskeleton then interprets and integrates the various signals to dictate distinct cellular functions, including cell shape, motility, and division in response to external stimuli¹, allowing the process of healing to proceed. Therefore, interactions between the membrane and cytoskeleton are crucial to understanding how cytoskeletal remodeling is coordinated across cells and tissues. While significant mechanistic insights have emerged as to how extracellular signals impact membranes to influence cytoskeletal dynamics, understanding these interactions in the context of injured tissues and their unpredictable extracellular environments has yet to be fully explored. Indeed, the microenvironment of damaged tissues is highly toxic and proteolytic and properly translating external clues to elicit the appropriate cellular response is critical for both homeostasis and tissue repair.

One prominent example of cytoskeletal reaction to stress is the formation of protrusions that promote communication between cells via transfer of signals, signaling molecules, nucleic acids, organelles and membrane components. However, the extracellular and intracellular signals and specific molecules that either trigger or specifically inhibit the induction, formation and function of these various collaborative protrusions are largely unknown². Protrusions have been shown to form in response to metabolic stresses such as a deficiency in serum proteins or excessive glucose levels, but few specific molecules have been identified that induce protrusion formation^2,3. Logically, both the membrane and the cytoskeleton must play a role in protrusion formation. Indeed, inhibition of actin polymerization abrogates protrusion formation and various actin-binding, membrane tethering, signal regulators and actin-regulatory proteins have been implicated in protrusion formation and assembly such as Cdc42, IQGAP1, LST1, RalA, M-sec and filamin in various cell types ^4–9.

Similarly, membrane organization and function are controlled by membrane lipids that sort the membrane into dynamic subdomains, thus assembling signaling platforms^10–13. Membrane lipids themselves, particularly phosphatidylinositol or sphingolipid species, also induce signaling cascades and regulate the activity or membrane localization of various proteins via direct interaction to impact a variety of cellular functions^12–16. Emerging evidence indicates that the cytoskeletal makeup and the nature of the membrane receptors are highly context and cell-type-dependent, which, combined with the lack of specific markers, makes it a challenge to definitively distinguish among the various types of protrusions^2–4,17,18. Similarly, the mechanisms guiding protrusion formation show a fair degree of cell specificity and whether a common, unifying process exists remains unclear. Further investigation is clearly warranted to provide tools to answer these questions and to fully exploit these novel mechanisms of cell–cell communication.

Our observations have established that the transmembrane molecule CD13 is remarkably multifunctional and regulates diverse processes such as filopodia formation, mesenchymal stem cell (MSC) rescue, maintenance of the stem cell niche, endocytic trafficking, angiogenesis, membrane organization, cell–cell fusion and cell–matrix and cell–cell adhesion^19–30. Indeed, our recent studies demonstrated that CD13 recruits a complex containing IQGAP1/ARF6/EFA6 to the plasma membrane and tethers this complex to the actin cytoskeleton via α-actinin^29,31. Importantly, CD13 is upregulated and activated at sites of tissue damage and ensuing inflammation, suggesting it contributes to tissue repair. In the present study, we show that CD13 is required for the formation of long, thin protrusions above the substrate that connect distant cells and contain actin and important signaling molecules. Interestingly, activation of CD13 by the crosslinking mAb 452 enhances protrusion formation, and cells engineered to lack CD13 form significantly fewer protrusions, suggesting an active role for CD13 in protrusion formation. Morphologically, both CD13 and actin are highly abundant and co-localize at the base as well as along the length of the protrusions under both stressed and induced conditions, consistent with CD13 contributing to actin organization in protrusions. The CD13-interacting proteins ARF6, IQGAP1, and specific PIP species PI(4,5)P accumulate at the base and/or along the length of protrusions in WT KSCs, while α-actinin, and IQGAP1 are mislocalized in the absence of CD13. To determine the potential CD13 effects on membrane lipid levels, we demonstrated that treatment with CD13-crosslinking mAb 452 increased PIP5K activity, and thus likely PI(4,5)P2 levels, in WT but not in CD13^KO cells, thus establishing a novel connection between CD13 and lipid signaling. Mechanistically, treatment with peptides designed to interfere with CD13’s cytoplasmic signaling domain diminished protrusion formation, as did inhibition of IQGAP1, Cdc42, FAK, or Src signaling²⁸, suggesting that CD13 activation-dependent signal transduction mediates protrusion formation. Finally, we illustrate that CD13 expression is necessary for the transfer of calcium signals through gap junctions, which occurs in WT but not in CD13^KO KSCs. Collectively, these investigations support the notion that CD13 is a unique regulatory node that coordinates the assembly, location and function of intracellular signaling complexes to promote various forms of intercellular communication.

Results

CD13 expression facilitates protrusion formation.

Throughout our studies of the single-pass, transmembrane protein CD13 in various contexts, we routinely observe alterations in the cytoskeleton when CD13 is perturbed^{19,20,25,28–30,32}. Furthermore, these studies have identified CD13 as a membrane tether localizing signaling complexes at the plasma membrane. In this context, CD13 serves as a scaffold that organizes signaling molecules and cytoplasmic- and cytoskeletal-binding adaptor proteins into functional complexes, prompting us to investigate the precise connection between CD13 and cytoskeletal dynamics in various cell types. In response to stress by serum deprivation or high glucose, we observed the formation of abundant, CD13-expressing, actin-containing protrusions (white arrows, Fig. 1a; Figure S1) in the human Kaposi’s sarcoma-derived cell line (KSC), consistent with a role for CD13 in their formation and/or function. To further investigate this notion, we CRISPR-engineered KSC cells to delete CD13 (CD13^KO KSCs) or scrambled guide RNA controls (WT KSCs), cloned and tested independent lines²⁹. The number of potential protrusions/cell (protrusion index)^33,34 in WT KSCs increased under metabolic stress conditions induced by serum starvation (SS) or high glucose (HG) for 1 h (4.5 g/L glucose, CM vs. SS/HG; 3.5 vs. 2.5, Fig. 1b) as compared to complete medium control, while the protrusion index was significantly lower in CD13^KO KSCs under all conditions tested, confirming that CD13 contributes to cytoskeletal protrusion formation. Furthermore, we validated CD13-dependent protrusion formation in the established endothelial cell line HUVEC, where both SS and CD13 crosslinking mAb 452 treatments enhanced actin⁺ protrusion formation compared to CM (Figure S2).

Fig. 1.

CD13 is required for protrusion formation in the presence of stressors. (a) WT KS1767 (WT KSC) and CD13^KO KS1767 (CD13^KO KSC) cells were treated with stressors (SS; serum starvation, HG; high glucose) for 1 h. WT KSC protrusions stained positive for actin (phalloidin; red), and CD13 (green). White arrows highlight protrusions. Scale bar: 10 μm. (b) The protrusion index was calculated by dividing the total number of protrusions within a FOV by the total number of cells. WT cells form significantly more protrusions under stress conditions compared to CD13^KO KSCs. Both stress conditions significantly induce protrusion formation compared to a complete media (CM) control in WT, but not CD13^KO cells. (c) WT KSCs form protrusions above the substrate, while CD13^KO protrusions are present closer to the substrate. (d) WT KSC protrusions are longer than CD13^KO KSC protrusions in length. (e) Protrusions observed in KSCs are not microtubule based as inhibiting microtubule polymerization with nocodazole did not significantly impact the protrusion index. (f) KSC protrusions are actin-based, as inhibiting actin polymerization with either cytochalasin B or D significantly inhibited protrusion formation. Data are represented as mean ± standard error and consists of at least three independent experiments with at least 5–10 images taken per treatment condition. *p ≤ 0.05.

Further characterization of these actin protrusions by examining z-stack images indicated that the protrusions in CD13^KO KSC were generally located close to the substrate, while the majority of WT protrusions were elevated, spanning distances up to 2.4 µm above the culture surface. Z-stacks were acquired beginning at 0.24 µm and continued in 0.24 µm increments, reaching a total stack height of 2.64 µm, with each 0.24 µm section considered as a bin. Protrusions spanning multiple z-sections were assigned to the bin containing most of its length. Long, actin-containing protrusions were present in the middle sections of the z-stack in WT cells (1.3—2.4 μm in WT vs. 0.24 μm—0.72 μm in CD13^KO), above the substrate (Fig. 1c, S3) and were appreciably longer than in CD13^KO cells (15 μm vs. 10 μm) (Fig. 1d). To confirm that these are actin-based³⁵, we treated WT and CD13^KO KSCs with inhibitors of microtubule (nocodazole) or actin (Cytochalasin B, D) polymerization and measured the protrusion index. We found that protrusions were significantly decreased upon treatment with actin, but not microtubule inhibitors (Fig. 1e, f), indicating that these protrusions are actin-based and that CD13 contributes to the organization of the actin cytoskeleton¹.

CD13 activation specifically induces protrusion formation

Cell surface receptor activation is often triggered by treatment with specific crosslinking monoclonal antibodies^36,37. We have shown that in monocytic cells CD13 is activated by the crosslinking mAb 452, resulting in a rapid, Src-dependent phosphorylation of tyrosine 6 (Y⁶) on CD13’s cytoplasmic tail²⁸. CD13 activation coordinates its subsequent interaction with signaling mediators and actin-binding proteins, ultimately connecting the actin cytoskeleton to the plasma membrane to promote cell–cell and cell-extracellular matrix adhesion^28,32. To investigate the effect of CD13 cross-linking/activation on the cytoskeleton, we treated WT or CD13^KO KSCs with the mAb 452 (1.8 µg/mL, 1 h) and found a significant induction in protrusion formation as compared to vehicle or serum-stressed controls, suggesting that not only is CD13 necessary for protrusion formation, but its activation markedly promotes it (Fig. 2a-b and S4). By contrast, treatment with either a non-activating anti-CD13 antibody (WM-47, 0.1 μg/mL)²⁷, or a control antibody against a distinct cell-surface protein CD31³⁸ (Fig. 2, CD31, 1 μg/mL) showed minimal effect on WT KSC protrusion formation, emphasizing that specific activation of CD13 underlies protrusion formation. Inhibition of CD13’s enzymatic activity by treatment of KSCs with Bestatin (100 μg/mL) also showed no significant effect, indicating that CD13-mediated protrusion formation is independent of its aminopeptidase activity^24,39 (Fig. 2b).

Fig. 2.

CD13 crosslinking induces protrusion formation. (a) WT KSCs were treated with indicated mAbs or control for 1 h prior to staining for actin (phalloidin; red), CD13 (green) and DAPI (blue). Arrows highlight protrusions. (b) WT KSCs formed significantly more protrusions in comparison to CD13^KO KSCs in all treatment conditions. Treatment with activating CD13 mAb 452 (1.8 µgmL) significantly induced protrusion formation in WT KSCs. CD31 crosslinking antibody, WM4.7 (non-activating CD13 mAb; 1:100) and bestatin (an inhibitor of CD13’s aminopeptidase activity; 100 μg/mL) had no significant impact on protrusion formation in WT KSCs. Scale bar: 10 μm. Data are represented as mean ± standard error and consists of at least three independent experiments with at least 5–10 images taken per treatment condition. *p < 0.05.

CD13 co-localizes with actin at the base of protrusions

CD13 and actin were clearly colocalized along the membrane and at the base of the actin protrusions. To quantify this observation, we compared the overall intensity of expression of CD13 and actin along the total membrane. We performed line scans by measuring the pixel intensities of either CD13 or phalloidin expression in equivalent increments along the entire cell membrane by ImageJ software. (Fig. 3a–b). Representative overlays normalized to the peak average pixel intensity of CD13 (black) and phalloidin (blue) line scans are shown in Fig. 3c–d. The relative distance between lines on the graphs corresponds to the differences in intensity between CD13 and phalloidin expression in the same area of the membrane. These measurements indicated that the intensity of both CD13 and actin expression was significantly higher at the base of the projections in serum-starved conditions, which was further enhanced by mAb 452 treatment (Fig. 3c-e). Furthermore, CD13 expression also correlated with cholesterol (filipin stain) in SS conditions (r = 0.68), which again significantly increased upon mAb 452 treatment (r = 0.85), and the curves aligned noticeably more closely, likely due to CD13 activation-mediated cytoskeletal rearrangement (Figure S5-c). Similarly, staining with the lipid raft marker Cholera Toxin B (CTB) was significantly higher in WT membrane at the site of protrusion initiation and extended along the length (Figure S5d-e). Taken together, this data suggests that CD13 activation coordinates the assembly of actin and cholesterol to promote protrusion formation, prompting further investigation into potential mechanisms underlying CD13-mediated cytoskeletal effects.

Fig. 3.

CD13 and actin co-localize at the base of protrusions. (a) Representative images of actin (red; top) or CD13 (green; bottom) staining of serum-starved WT KSC protrusions with quantification (b). Normalized average fluorescence measurements for both actin and CD13 were calculated by selecting a cell in which the entire cell body could be seen within the FOV. Membrane areas were delimited by boxes of uniform size (white boxes) systematically moved around the entire cell membrane and measured with ImageJ. Membrane areas containing protrusions were classified as ‘protrusion membrane’, while the others, ‘non- protrusion membrane’. Protrusion membrane fluorescence was then normalized to non-protrusion membrane fluorescence. In all conditions, there was significant localization of both actin and CD13 at the base of protrusions compared to the rest of the cellular membrane. Treatment of cells with CD13 mAb 452(1.8 μg/mL) for 1 h significantly increased the amount of protein at the base of protrusions in comparison to SS control. (c) Representative images and (d) line scans of normalized average pixel intensity of CD13 (black) and phalloidin (red) under complete media (r = 0.62), serum-starved (r = 0.84) or mAb 452 (r = 0.88) treatment conditions. (e) Quantification of Pearson Correlation Coefficient (r) for CD13 and phalloidin under complete media (CM) or 452 mAb treatment conditions. 452 mAb treatment significantly increased CD13 and phalloidin correlation in comparison to complete media (CM) control. Scale bar: 10 μm. Data are represented as mean ± standard error and consists of at least three independent experiments with at least 5–10 images taken per treatment condition. Membrane measurements were taken from 50–100 cells per experiment. *p < 0.05.

CD13’s cytoplasmic tail coordinates protrusion formation

We have shown that although CD13’s cytoplasmic tail contains only 8 amino acids, it interacts with a number of other membrane-tethered and cytoplasmic proteins²⁸, and blocking these interactions with interfering peptides impairs subsequent CD13-dependent functions, such as cell migration^28–30. To determine whether protrusion formation involves CD13’s cytoplasmic tail, we loaded KSCs with the CD13-blocking (MAKGFYI; 1 μg/mL) or a control peptide (KFMYGAI; 1 μg/mL) using BioPORTER. Blocking the CD13 cytoplasmic tail interactions resulted in a reduction in protrusions as compared to control peptide or vehicle (Fig. 4a–b: Figure S6a) but did not affect CD13 expression levels (Fig. 4c). The migration of cells transfected with the CD13-blocking peptide was also significantly impaired compared to control peptide in an in vitro scratch assay (Figure S6b-c), confirming that the peptide blockade is functioning correctly. Therefore, CD13-dependent protrusion formation is mediated via protein interaction with its cytoplasmic tail.

Fig. 4.

CD13’s cytoplasmic tail is required for protrusion formation. (a) WT and CD13^KO KSCs were loaded with the CD13-blocking or control (1 μg/mL) peptide and stained for actin (phalloidin; red), CD13 (green) and DAPI (blue). (b) The presence of the blocking peptide significantly reduced WT KSC protrusion formation in comparison to vehicle or control peptide, while CD13^KO cells were unaffected with either peptide. 63x, oil. Scale bar: 10 µm. (c) Flow cytometric analysis of CD13 surface expression on peptide-loaded WT and CD13^KO KSCs shows neither peptide had a significant impact on CD13 surface expression. Scale bar: 10 μm. Data are represented as mean ± standard error and consists of at least three independent experiments with at least 5–10 images taken per treatment condition. *p < 0.05.

CD13-mediated protrusion formation involves FAK/Src signaling, small GTPases and PI(4,5)P2 generation

We have demonstrated that protein interactions with CD13’s cytoplasmic tail can occur in both constitutive and activation-dependent manners, triggering overlapping and complex signaling cascades to control diverse cell functions^{19,20,25,28–30,32}. Intriguingly, two of these pathways have been previously implicated in protrusion formation by other groups: FAK/Src signaling^40,41 and activation of the ARF6 small GTPase⁴². We had previously shown that the phosphorylation of FAK and Src kinases is reduced in CD13^KO compared to resting WT cells and in CD13^KO mice in vivo in a skeletal muscle ischemic injury model^19,20,28. Furthermore, activation of CD13 in human monocytic cell lines induced marked stimulation of these kinases, prominent alterations in the actin cytoskeleton and increased cell–cell and cell-ECM adhesion^{19,20,25,28–30,32}. Treatment of KSC with inhibitors of FAK Y397 (FAK14; 20 μM) and Src Y416 (PP2; 10 μM) phosphorylation for 1 h abrogated protrusion formation in WT to the level of the CD13^KO KSCs (Fig. 5a, S7). This phenotype could not be rescued by treatment with the activating mAb 452, implying that CD13 mediates protrusion formation via FAK-Src signal transduction pathways downstream of CD13 activation (Figure S7).

Fig. 5.

CD13-mediated FAK/Src/IQGAP1/ARF6 signaling promotes protrusion formation. (a) Treatment with inhibitors of major actin-organizing proteins such as FAK (FAK14; 20 μM), Src (PP2; 10 μM) or Cdc42 (ML141; 20 μM) significantly inhibits protrusion formation in WT KSCs, indicating that these proteins are critical for regulating actin cytoskeletal proteins in the formation of protrusions. (b). Representative images showing IQGAP1 or ARF6 (green) and phalloidin (actin; red) staining under CM or crosslinking (mAb 452; 1.8 µg/mL) conditions. (c-d) Quantification of Pearson Correlation Coefficient (PCC) between IQGAP1 and phalloidin (c) and ARF6 and phalloidin (d) staining following treatment with either CM or mAb 452. Treatment with mAb 452 significantly increased the correlation between IQGAP1 and phalloidin as well as ARF6 and phalloidin at the membrane. Data are represented as mean ± standard error and consists of at least three independent experiments with at least 5–10 images taken per treatment condition. *p < 0.05.

CD13 also directly regulates endocytosis and endosomal trafficking by tethering a complex containing the scaffolding protein IQGAP1, the ARF6 member of the Arf family of GTPases, and the actin-binding protein α-actinin to the plasma membrane²⁹. In the absence of CD13, these proteins are mislocalized, resulting in aberrant recycling of cell surface proteins²⁹. Independent studies⁴² demonstrated that in a neuronal cell line, ARF6 participates in a Rab GTPase signaling cascade that results in protrusion formation. Pertinent to these findings, immunofluorescence of serum-stressed KSCs showed IQGAP1 and ARF6 expression (Fig. 5b–d) within and at the base of protrusions, suggesting that CD13 again localizes these proteins to the membrane where, in this case, it facilitates protrusion formation, implicating CD13 association with IQGAP1/ARF6 in protrusion formation.

In addition to inducing FAK and Src signaling, IQGAP also recruits the Rho GTPase Cdc42/Rac1-WASP-Arp2/3 complex to the plasma membrane^43–45, enabling the protrusion of actin polymers. Similarly, it has been shown that active Cdc42/Rac1 is critical for protrusion formation in macrophages, which also express high levels of CD13⁴⁵. In agreement with these data, and similar to actin polymerization, inhibition of Cdc42-GTPase (ML141; 20 μM) for 1 h, significantly reduced the number of protrusions in WT KSCs (Fig. 5a).

Finally, the phosphatidylinositol species PI(4,5)P2 has been implicated in protrusion formation⁴⁶ and is a key regulator of actin assembly, cell membrane dynamics, focal adhesion formation and signal transduction⁴⁷. ARF6 recruitment to the plasma membrane activates PIP5K (phosphatidylinositol-4-phosphate 5-kinase) to generate PI(4,5)P2⁴⁸. We have shown that CD13 is important for assembling the proteins for endocytosis and therefore, hypothesize that CD13’s recruitment of ARF6 may also modulate PI(4,5)P2 generation. mAb 452 activation of CD13 in WT KSCs significantly increased PIP5K activity using an ATP depletion assay (Echelon Biosciences) as compared to control or serum-stressed conditions (Fig. 6a), indicating that activation of CD13 activates PIP5K to generate PI(4,5)P2 (Fig. 6b–c). In agreement with previous studies, PI(4,5)P2 localized in protrusions as visualized by transfection with PH-GFP, a plasmid encoding a labeled pleckstrin homology (PH) domain of PLCγ that specifically binds to PI(4,5)P2^49,50. PH-GFP/PI(4,5)P2 localization significantly correlates with that of CD13 in mAb 452-treated WT KSCs (PCC r = 0.91) in comparison to CM control (PCC r = 0.75, Fig. 6d–f).

Fig. 6.

CD13 crosslinking induces PIP5K and PI(4,5)P2 generation. (a) The relative PIP5K activity was measured using an activity assay kit (Echelon Biosciences) in both WT and CD13^KO KSCs. WT cells showed higher levels of PIP5K activation in comparison to CD13^KO cells in serum starvation (SS) and mAb 452 (1.8 µg/mL) treated conditions. Treatment of WT KSCs with mAb 452 significantly increased PIP5K activity over CM control. (b-f) Representative images and Pearson correlation coefficients for CD13 (red; black line) and PI(4,5)P2-GFP (green; blue line). (b, c) Images with zoomed (i) insets showing CD13 (red) and PI(4,5)P2-GFP (green) at the base of protrusions in complete media (b) or mAb 452 treatment (c). (d–f) Pearson correlation coefficients and line scans (d, f) were calculated between CD13 (r = 0.9) and PI(4,5)P2 (r = 0.72). There is a significant increase in coexpression of CD13 and PI(4,5)P2 upon mAb 452 treatment. Data are represented as mean ± standard error and consists of at least three independent experiments with at least 5–10 images taken per treatment condition. *p < 0.05.

CD13-dependent protrusions connect cells and transfer calcium signals.

We observed that CD13-dependent protrusions appeared to connect distant cells, raising the possibility that these may be tunneling nanotubes (TNTs), a subset of protrusions shown to connect and transfer subcellular proteins or organelles between cells^51–56. However, repeated attempts to visualize transfer of cytoplasmic cargo between CD13-activated or resting WT KSCs via protrusions were unsuccessful, and thus these protrusions are not open-ended TNTs. Alternatively, the closed-ended TNTs are a second type of TNTs that propagate Ca²⁺ fluxes to physically separated cells upon mechanical stimulation². We performed Ca²⁺ transfer experiments using real-time Fura2-AM Ca²⁺ imaging in the presence of 2 mM EGTA without calcium to ensure any Ca²⁺ influx did not occur through endogenous channels. Mechanical stimulation of Fura-2 AM-loaded WT and CD13^KO KSCs using a patch electrode induced a robust increase in the intracellular Ca²⁺ concentration in stimulated cells of both phenotypes (Fig. 7a, red cell ‘touching’), although positive transfer of the intracellular Ca²⁺ flux to surrounding cells occurred at a significantly faster rate in the WT compared to CD13^KO KSCs. This significantly higher Ca²⁺ concentration was sustained for 120 s post-perturbation, eventually equalizing by the 600 s time point (Fig. 7a–d). Image enlargement revealed connecting protrusions in WT but not in CD13^KO KSCs, consistent with Ca²⁺ transfer facilitated by protrusions (Fig. 7e). Finally, positive immunostaining for the gap junction protein connexin 43 (Cx43) at the base and along the length of the protrusion suggests that the CD13-dependent protrusions may be closed-ended TNTs (Fig. 7f). Consistent with previous studies⁵⁵, Cx43 expression was detected at both ends of the protrusion (arrows), possibly due to the presence of multiple, closely-spaced protrusions. Further investigation is necessary to precisely define and thoroughly characterize these protrusions by high resolution scanning and transmission EM. This data indicates that the CD13-dependent protrusions connect cells, are capable of transferring Ca²⁺, and that the efficiency of Ca²⁺ transfer is compromised in the absence of CD13, consistent with closed-ended TNTs.

Fig. 7.

CD13 promotes transfer of Ca²⁺ via protrusions. (a) KSCs were loaded with Fura-2 AM (2.5 µM) and stimulated with a patch electrode before observing Ca²⁺ transfer between cells, via protrusions, over a period of 2 min (120 s). White arrows indicate the cell stimulated. (b) Comparison of individual cell traces measuring Ca²⁺ flux in WT versus CD13^KO KSCs following stimulation. (c) Averaged cell traces of WT and CD13^KO KSCs showing a significant decrease in Ca²⁺ flux in CD13^KO KSCs that lack protrusions. (d) Measurement of the number of responsive cells (cells increasing their Ca²⁺ content that are not the stimulated cell) in WT versus CD13^KO KSCs shows a significant decrease in the number of cells responding to the stimulus. (e) Zoomed representative images of Ca²⁺ transfer via protrusions in WT, but not CD13^KO, KSCs. Showing that protrusions are responsible for the propagation of Ca²⁺ between cells in this environment. White arrows indicate stimulated cell. (f). Connexin 43 (green) expressing GJs are detected along the length and at both ends of the protrusion (phalloidin) in serum starved WT KSC. Arrows indicate Cx43⁺ GJs. Data are represented as mean ± standard error and consists of at least three independent experiments with at least 3 images taken per treatment condition. *p < 0.05. Scale bar is 100 μm.

Together, these data establish a pathway in which activation of CD13 orchestrates FAK and Src phosphorylation, CD13-dependent recruitment of IQGAP1/ARF6 at the plasma membrane, triggering downstream Cdc42 activation and actin polymerization, PIP5K activation, ultimately resulting in the generation of PI(4,5)P2 and cytoskeletal changes required to induce the formation of actin protrusions (Fig. 8)²⁹.

Fig. 8.

Proposed mechanism of CD13-dependent protrusion formation. (1) Crosslinking of CD13 with mAb 452 induces (2) its phosphorylation by Src and FAK kinases, which has been shown to activate Cdc42, and (3) the assembly of a membrane-bound complex containing IQGAP and ARF6, which also stimulates Cdc42 and actin polymerization. In addition, by tethering IQGAP at the membrane, CD13 enables (4) activation of ARF6 to ARF6-GTP. In turn, (5) activated ARF6 activates PI-5 K, enabling (6) the conversion of PI-4P to PI(4,5)P2, which promotes (7) formation of membrane protrusions containing (8) Cx43, allowing transfer of (9) Ca²⁺ to the neighboring cell.

Discussion

In the current study, we demonstrated that the cell surface molecule CD13 mediates the formation of actin protrusions in a human Kaposi’s sarcoma-derived cell line. Our studies show that CD13 acts by assembling the machinery necessary to activate signaling pathways regulating protein and lipid phosphorylation and activation, actin polymerization and membrane organization. Furthermore, we discovered a novel function for CD13 in the critical membrane phosphoinositide regulatory pathway, specifically in regulating the activation of PIP5K, which is involved in lipid phosphorylation and PIP2 generation. Phosphoinositides impact various cellular processes via their interaction with a vast array of protein partners and other membrane phospholipids. They are themselves regulated by PI kinases and phosphatases to control the cytoskeleton, membrane dynamics, and intracellular vesicular trafficking, all of which likely play a role in protrusion formation and transport. Membrane PIP2 interacts with actin-binding proteins to control actin network organization^{10,11,57–60}. Pertinent to our study, PIP2 binds to α-actinin, as does active CD13²⁸, likely contributing to protrusion formation. Similarly, in collaboration with Cdc42, PIP2 promotes N-WASP/Arp2/3 signaling to control actin polymerization and protrusion formation⁴⁵. PIP2 also regulates the activation of FAK, a critical step in many signaling pathways controlling migration and adhesion. We have shown that the absence of CD13 reduces activation of both FAK and Src kinases and cell migration^28,29 , perhaps due to higher PIP2 levels upon CD13 activation. PIP2 has been shown to interact with and activate FAK in an ATP-dependent manner to mediate focal adhesion turnover and FAK/integrin signaling⁴⁴. Finally, while phosphoinositides comprise only 1% of the plasma membrane, PIP2 plays a prominent role in membrane dynamics and organization via its binding proteins. Many of these are activated by signaling induced by cell surface receptor activation¹¹ and contribute to membrane curvature sensing, membrane domain segregation, endocytosis and exocytosis⁶¹. We have established that CD13 is a regulator of receptor-mediated endocytosis^24,39 and integrin vesicular trafficking²⁹ but have not explored the contribution of phosphoinositides to these CD13 functions. Further investigation into how these multifunctional proteins interrelate in these cellular processes will clearly uncover novel common mechanisms regulating protrusion formation.

Accumulating evidence demonstrates that not surprisingly, protrusion formation in general and TNT formation and cargo transfer in particular requires contributions from a large number of proteins of various functions including GTPases [RalA^5,8], vesicular tethering, transport and recycling proteins [Rabs 8 and 11⁶², M-Sec⁸), cytoskeletal regulators [RASSF1A³³], actomyosin-motor proteins [myo10^63,64] and ER chaperones [ERp29⁶¹] among others. Logically, these components must be carefully assembled into multimolecular complexes by scaffolding proteins and localized to specific areas of the cell to enable these processes. One such multimolecular complex has been identified to coordinate TNT formation, where overexpression of the membrane-bound LST1 (lymphocyte-specific transmembrane protein) in HeLa cells induces TNT formation by recruiting a complex containing the small GTPase RalA, its GEF RalAGPS2, the actin-binding scaffold protein filamin and the exocyst-homologous protein M-Sec to the plasma membrane ⁵. We have recently described a similar role for CD13 in the recruitment and tethering of a complex containing analogous components; the GTPase ARF6 and its GEF EFA6, the scaffold protein IQGAP1 and actin-binding α-actinin at the plasma membrane to regulate β1-integrin endosomal recycling ²⁹. Interestingly, these complexes may be functionally linked: integrins have been shown to activate RalA (a component of the LST1 complex) that in turn stimulates ARF6 (a component of the CD13 complex) to drive exocyst-dependent delivery of membrane microdomains to the plasma membrane ⁶⁵, perhaps to supply membrane to the growing TNT. In agreement with this mechanism, we find that in CD13^KO KSCs (which are impaired in TNT formation) ARF6 is in its inactive GDP form and β1-integrin accumulates in cytoplasmic vesicles rather than recycling to the membrane, perhaps interfering with integrin-Ral-ARF6 activation and subsequent membrane expansion. Furthermore, IQGAP1, β1-integrin, FAK and Src proteins are found along the length of the actin protrusions (not shown), consistent with a functional role for this CD13-tethered recycling complex in actin-based protrusion formation, which is under current investigation.

In addition to facilitating protrusion formation, we have recently reported that CD13 also negatively regulates cell–cell fusion, in part by limiting the formation of fusion-promoting, actin-based protrusions located at the base of the cell⁶⁶. Specifically, we found a significant increase in micro protrusion-like structures (visible under electron microscopy) attached to the substratum in CD13^KO macrophages under fusion-promoting conditions. These structures contributed to exaggerated macrophage fusion to form the multinuclear giant cells (MGCs) that characterize the inflammatory reaction induced by foreign implants⁶⁷. While this observation may seem at odds with our current study, which describes CD13 as a stimulator of protrusion formation, we find that KSCs lacking CD13 also exhibit increased numbers of short protrusions at the base of the cell (data not shown). The length of these structures and their location at the base of the cell suggest that they are similar to filopodia, although further characterization is needed. It is possible that under cellular stress, CD13 acts as a molecular switch, promoting the formation of protrusions to propagate stimulating signals and rescue distant cells, rather than facilitating filopodia formation. Further investigation into CD13 and its impact on actin dynamics may provide valuable insights into the specific molecules and cellular machinery involved in regulating various types of membrane protrusions. This includes filopodia, which are actin-rich projections involved in sensing the environment and cell movement; fusion-promoting protrusions, which facilitate cell–cell fusion processes; and TNTs, which serve as conduits for intercellular communication and material exchange. Understanding these distinct membrane structures could enhance our knowledge of how cells interact, migrate, and communicate under different physiological and pathological conditions.

Methods:

Cell culture

The KS1767 cell line (KSC) was derived from a human Kaposi’s Sarcoma lesion⁶⁸ and expresses cytokeratin and low levels of CD31 (Fig.S1b-c), consistent with an epithelial cell phenotype. Cells were cultured in DMEM (Gibco; 11965–092) supplemented with 1% Penicillin–Streptomycin (P/S) (Gibco; 15140122, 100U/mL) and 10% HI-FBS (Gibco; 16250078). Cells were detached using 5 mL 0.25% Trypsin–EDTA (Gibco; 25300054), collected into 10 mL fresh serum-containing DMEM, spun at 1200 rpm for 5 min and counted using trypan blue and a Countess II. For the continuation of cells in culture, 2 × 10⁶ cells were seeded per T175 flask with 25 mL DMEM supplemented with P/S and heat-inactivated FBS. For experiments, cell seeding density varied depending on the application from 1 × 10⁵–3 × 10⁵ cells/coverslip. Cells seeded onto coverslips were resuspended in DMEM supplemented with P/S and heat-inactivated FBS. Cells then settled overnight, or for 2-3 h, depending on the application, prior to treatments.

Scrambled control (WT KSC) and CD13^KO KSC Kaposi’s sarcoma-derived cells (CD13^KO KSC) were generated by CRISPR-Cas9 gene editing as described previously²⁹.

Antibody and ligand treatment of cells

Treatment	Working concentration	Catalogue #
CD13 mAb	1.8 μg/mL	Isolated by Shapiro Lab
Bradykinin	200 nM	Sigma; B3259
WM4.7	0.1 μg/mL	Sigma; C8589
CD31	1:1000 (1 μg/mL)	Invitrogen; MA3100
Bestatin	100 μg/mL	Sigma; B8385
Cytochalasin B	250 nM	Cayman Chemical; 11,328
Cytochalasin D	250 nM	ThermoFisher; PHZ1063
FAK14	20 μM	Sigma; SML0837
PP2	10 μM	Selleck Chem; S7008
ML141	20 μM	Tocris; 4266
Nocodazole	150 nM	Selleck Chem; S2775

Immunofluorescence and labeling

Antibodies/stains	Dilution/concentration, catalogue #
CD13 mAb	1:100 (1.8 μg/ml), isolated
Phalloidin-TRITC	1:100 (1 μg/mL), Sigma; #P1951
Phalloidin-488	1:1000 (0.3U/mL), ThermoFisher; #A12379
Alexa Fluor secondary antibodies (Life Technologies)	1:1200 (1.6 μg/mL) 488: #A11029, #R37114 594: #A11037, #A21203 640: #A21235
DAPI	1:500, Invitrogen; #D1306

KSCs were seeded at a density of 7.5 × 10⁴ cells/coverslip. Cells attached to coverslips overnight and washed two times with PBS followed by the addition of any treatments for 1 h. Cells were washed two times with PBS, fixed for 20 min in 4% PFA at room temperature and permeabilized in 0.01% Triton-X for 5 min at room temperature. Coverslips were then blocked for 1 h at room temperature in PBS/5% w/v BSA supplemented with 5% serum from the host of the secondary antibody. Primary antibodies were added in PBS/5%BSA/5%host serum and coverslips were incubated overnight at 4 °C. The cells were then washed twice with PBS and incubated for 1 h at room temperature with Alexa Fluor secondary antibodies and DAPI. The coverslips were then washed three times with PBS and mounted on glass slides with Prolong Gold Antifade mounting media (100 µl/coverslip, Molecular Probes; P10144). The mounting media was cured overnight at room temperature and protected from light.

Microscopy

Slides were visualized at excitation wavelength of 488 nm (Alexa 488), 543 nm (Alexa 594 or TRITC) and 405 nm (DAPI), and imaged using an inverted fluorescent Zeiss LSM880 microscope or a confocal Zeiss Axio Observer.Z1 with a 63X oil objective. Images and Z-stacks were taken using the ZEN 2 software by Zeiss.

Image analysis

Protrusion index was calculated as an average number of protrusions/eligible cell. Cells were considered eligible for analysis if at least 90% of the cell body was within the FOV, there was sufficient room surrounding the cell to form a protrusion (e.g. a cell of interest could not be counted if it was touching other cells on all sides). Total protrusions from eligible cells were then counted per FOV and normalized to the number of total eligible cells to get the protrusion index.

Measurement and correlation of protein co-localization by immuno-fluorescence

In this analysis, the expression levels (MFI) of membrane actin or CD13 were measured in uniform increments along the entire membrane of ~ 60 individual cells. To measure the overall expression levels of these proteins at membrane areas abutting protrusions vs. areas with no protrusions, the MFI value of each increment was normalized to the cell’s overall maximum pixel intensity and averaged for each group (e.g. normalized average MFI at the base of protrusions vs. no protrusions), represented as bar graphs. To assess the degree of co-expression, the relative pixel intensity of actin and CD13 was plotted for each point along the cell membrane, illustrated as line graphs. Coincident curves represent areas of co-expression, verified by the increased correlation of pixel intensity measurements between CD13 and phalloidin, calculated by the Pearson Correlation Coefficient (PCC). Graphs depict a representative set of 60 measurements along the membrane.

BioPorter and CD13 peptide generation

Peptides were synthesized by Pierce Custom Peptides (ThermoFisher Scientific). The following sequences were used: MAKGFYI (CD13-blocking) and KFMYGAI (control). Each peptide was synthesized with 98% purity. 1 mg of each peptide was resuspended in 500µL PBS, aliquoted and stored at -80C. For importing into the cell, BioPorter was used (Genlantis; BP502401) according to manufacturer’s instructions. BioPorter is a protein delivery reagent that utilizes an unspecified lipid formulation that allows direct translocation of proteins into living cells. After covalently binding to the protein, the reagent transports the lipid-protein complex into the target cell. Briefly, Bioporter was resuspended in 100% methanol and 2.5µL/reaction was used per well in a 24-well plate. Peptides were diluted in PBS to 0.1µg/mL, combined with the BioPorter and the volume was brought up to 250µL in serum-free media. Cells were then incubated with the mixture for 4 h at 37 °C, followed by fixation and staining as previously described.

Measurement of PIP5K activity

0.3 × 10⁶ WT or CD13^KO KSC were seeded onto 6-well plates in complete media or serum-free media + /− mAb 452. Lysates were collected and PIP5K activity assay was performed following manufacturer’s instructions (Echelon Biosciences, K-5700).

Wound healing assay²⁹

WT or CD13^KO KSC were cultured to confluence on glass coverslips. A wound was created by scratching with a 10 µl pipette tip along the center of the coverslip. Cells were washed with complete medium and treated with CD13 blocking or control peptide as described above. Cells were allowed to migrate and close the wound for 0–18 h.

Immunoblot analysis

Cell lysates were harvested in 1% NP40 lysis buffer containing 1X cOmplete Protease Inhibitor cocktail (Roche). Samples were analyzed by SDS-PAGE and the separated proteins were transferred onto nitrocellulose membrane using the TransBlot Turbo system (Biorad; 1,704,158). Membranes were blocked in 1XTBST containing 5% bovine serum albumin, treated with primary Ab followed by appropriate secondary Ab. Blots were developed using Clarity Max™ Western ECL Substrate (Biorad; 1705062) and imaged by ChemiDoc Imaging system (Biorad). GAPDH was used as loading control.

Ca²⁺ flux measurement

Changes of intracellular Ca²⁺ was measured using ratio Ca²⁺ imaging as described previously⁶⁹. In brief, Fura-2 AM (Thermal Fisher Scientific; F1221) in presence of 2 mM EGTA was dissolved in DMSO to make a stock concentration at 1 mM. Pre-warmed Neurobasal® Medium (Thermal Fisher Scientific; 21103–049) was used to dilute Fura-2 AM to a working concentration at 2.5 µM, and 0.02% Pluronic™ F-127 (Thermal Fisher Scientific; P3000MP) was added to facilitate loading of Fura-2 AM. Cells plated on 25 mm glass coverslips were washed using pre-warmed PBS for 3 times, and then incubated with 2 ml of Fura-2 AM working solution for 30 ~ 45 min at 37 °C. Non-incorporated dye was washed away using HEPES-buffered Saline Solution (HBSS) containing : 20 mM HEPES, 10 mM glucose, 1.2 mM MgCl₂, 1.2 mM KH₂PO₄, 4.7 mM KCl, 140 mM NaCl, 1.3mM Ca²⁺ (pH 7.4). Patch electrodes were pulled from borosilicate glass and fire-polished to a resistance of ∼3 MΩ when filled with HBSS solutions.

Statistical analysis

Statistical analysis was performed using unpaired, two-tailed Student’s t test and results are representative of mean ± SEM as indicated. Differences at p ≤ 0.05 were considered significant.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

EM, PZ, LY, LHS and MG designed the research plan. EM, BA, RS, NT, FM, PZ and MG conducted experiments and analyzed data. EM, LHS and MG wrote the manuscript. All authors reviewed the manuscript.

Funding

This work was supported by National Institutes of Health grant R21AI15 and Research Excellence grant, University of Connecticut to LHS and MG.

Data availability

The datasets generated and analyzed during the current study are available from the corresponding authors upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.