High spatial-resolved heat manipulating membrane heterogeneity alters cellular migration and signaling

Significance

The direct link between membrane heterogeneity and cell biological processes remains poorly understood, this is, at least in part, due to the shortage of high spatial-resolved molecule tools to interfere with the membrane heterogeneity. This work establishes a DNA-origami-based nanoheater system to manipulate the local membrane fluidity and membrane phase separation, which in turn triggers an integrin-associated cell migration change. This is significant because the high spatial-resolved, remote-controlled, domain-targeted, and reversible molecule tool could alter membrane heterogeneity through physical property change, featuring the capability to build a direct, unexplored correlation between membrane heterogeneity and cell function, which holds great potential in clinical applications.

Abstract

Plasma membrane heterogeneity is a key biophysical regulatory principle of membrane protein dynamics, which further influences downstream signal transduction. Although extensive biophysical and cell biology studies have proven membrane heterogeneity is essential to cell fate, the direct link between membrane heterogeneity regulation to cellular function remains unclear. Heterogeneous structures on plasma membranes, such as lipid rafts, are transiently assembled, thus hard to study via regular techniques. Indeed, it is nearly impossible to perturb membrane heterogeneity without changing plasma membrane compositions. In this study, we developed a high-spatial resolved DNA-origami-based nanoheater system with specific lipid heterogeneity targeting to manipulate the local lipid environmental temperature under near-infrared (NIR) laser illumination. Our results showed that the targeted heating of the local lipid environment influences the membrane thermodynamic properties, which further triggers an integrin-associated cell migration change. Therefore, the nanoheater system was further applied as an optimized therapeutic agent for wound healing. Our strategy provides a powerful tool to dynamically manipulate membrane heterogeneity and has the potential to explore cellular function through changes in plasma membrane biophysical properties.

The plasma membrane features intrinsic heterogeneity with diversities in local composition and biophysical properties (1–5). Of which, lipid rafts are small (10 to 200 nm), highly transient domains that are more tightly packed compared to the rest of the membrane (6–8). Lipid raft principles provide a protein sorting mechanism by selectively excluding or including proteins (9, 10). In this way, lipid rafts are structurally important to plasma membrane organization (11, 12).

Moreover, both in vitro reconstitution and cellular studies showed that the phase separation behavior of lipid mixtures triggers changes in membrane protein sequestration and oligomerization that influence protein function and downstream signaling (13–16). However, few analytical techniques and molecular tools have the appropriate spatial and temporal resolution to accurately probe and regulate the phase separation feature; therefore, the direct mechanistic correlation between lipid rafts and cell function or dysfunction remains unclear (17, 18).

Instead of regulating chemical compositions in lipid rafts, interfering with the fluidity gap between raft and nonraft regions could be another possible way to alter the phase separation (19, 20). Cell membrane fluidity is closely related to three factors, temperature, cholesterol content, and kinds of fatty acids in the phospholipids (21). Among them, we found that temperature could be physically manipulated by external forces. To achieve a domain-dependent membrane fluidity manipulation, a high spatial-resolved, targeted-heating nanodevice is the prerequisite. DNA origami technology enables fabricating nanostructures with well-defined topology and size that match lipid rafts (22–26). Furthermore, DNA origami features high biocompatibility and ease of modification (27–30), thereby, in principle, a size-matchable DNA origami possesses the ability to specifically seize and label solely lipid rafts.

Herein, we designed a nanoheater composed of a DNA origami scaffold, Au nanorod (NR), and aptamers that recognize different domains of cell membranes. Upon positioning at interested domains, harnessing the localized photothermal properties of Au NRs, the nanoheater specifically increases the domain temperature upon NIR laser illumination. In this way, for the first time, we altered plasma membrane fluidity with high spatial resolution, which in turn interferes with phase separation. Our results indicated that the heat-induced alternation in heterogeneity could lead to domain-dependent cell migration regulation. This high spatial-resolved heating strategy could be used to explore heterogeneity-associated cell signaling and help to solve some controversial debates about lipid rafts.

Results

Design and Preparation of the Nanoheater.

To achieve domain-dependent membrane fluidity control, we designed a nanoheater that anchors on rafts or nonraft regions and heats the anchored region on the membrane under NIR laser. In brief, the nanoheater consists of three functional units (Fig. 1A): 1) a rectangular DNA origami (90 nm × 60 nm) scaffold; 2) an Au NR (length: ∼40 nm; width: ∼12 nm) assembled on one face of the scaffold as a photothermal unit that can be heated multiple times; and 3) aptamers that possess specific recognition to either cholera toxin B (CTxB, raft) or transferrin receptor (TfR, nonraft) (SI Appendix, Fig. S1). Au NRs were attached to the origami scaffold by modifying with single-stranded DNA (ss-DNA) (SI Appendix, Fig. S2) and then hybridizing with a complementary ss-DNA hanging on the origami. The successful attachment of Au NRs on origami was confirmed by Agarose gel (SI Appendix, Fig. S3), atomic force microscopy (AFM) (Fig. 1C), and zeta potential measurements (Fig. 1D).

Fig. 1.

Design and attachment of nanoheaters on the cell membrane. (A) Schematic illustrating the design of nanoheater through the integration of a rectangular DNA origami scaffold, an Au nanorod (photothermal unit), and various aptamers (directing unit). (B) Schematic illustrating the positioning of nanoheater at lipid raft or nonraft domain through CTxB aptamer or TfR aptamer, respectively. 808 nm laser was then introduced to activate nanoheaters to heat the labeled domains. (C) AFM images of rectangular DNA origami (Top) and Au NR-loaded DNA origami (Bottom). (Scale bar: 120 nm.) Insets: enlarged images of marked areas. (D) Zeta potential of Au NRs, ss-DNA-modified Au NRs (Au NR-DNA), Au NR-loaded DNA origami (Ori-Au NR), and DNA origami. (E) Temperature changes of a buffer solution containing the nanoheater (2 nM) upon successively illuminating by 808 nm laser (2 W cm⁻²) for 1 min and then cooling down to room temperature. (F) Reconstructed dSTORM images of fixed MCF-7 cells upon treatments with Alexa647-conjugated CTxB and Alexa488-conjugated TfR antibody (Top), Alexa647-conjugated CTxB and FAM-labeled lipid raft-anchored nanoheater (Middle), Alexa647-conjugated TfR antibody and FAM-labeled non-raft-anchored nanoheater (Bottom). (Scale bar: 10 μm, enlarged: 2 μm.) (G) Western blot analysis of the expression levels of HSP 70 after different treatments as indicated.

Ganglioside GM1 is enriched in lipid rafts, and the high affinity of CTxB subunit towards ganglioside GM1 enables CTxB to be a widely used marker of lipid rafts (31, 32), while TfR has been reported to exist mainly in nonraft regions and thus is used to label nonraft in this study (33, 34). To start with, we treated MCF-7 cells with CTxB to mark lipid rafts, followed by adding nanoheaters to anchor the labeled rafts (35). Similarly, TfR aptamer was applied for attaching the non-raft-targeted nanoheater (Fig. 1B) (36, 37).

Au NRs are effective photothermal reagents due to their NIR plasmonic absorption. Additionally, the application of aptamer instead of antibody to conjugate nanoheaters with lipid domain helps to minimize the distance between the targeted biomolecules to achieve a possible high-spatial resolved local heating (38). In this study, the Au NRs have an aspect ratio of 3.3 and longitudinal localized surface plasmonic resonance (LSPR) peak at 818 nm (SI Appendix, Fig. S4), we thereby introduced an 808 nm laser to illuminate the nanoheater to rise local temperature. Then, 1 min of laser illuminating with an intensity of 2 W cm⁻² increased the solution temperature from 27 °C to 42 °C (SI Appendix, Fig. S5) and thus was selected as the optimized condition for lipid heating. Importantly, the laser-induced temperature change is completely reversible and can be repeated for several cycles (Fig. 1E).

We then positioned the nanoheaters on the membrane of MCF-7 cells. To begin with, Alexa647-labeled CTxB and Alexa488-labeled anti-TfR antibodies were used to light up raft or nonraft regions, respectively. The treated cells were then incubated with either a CTxB aptamer-carrying nanoheater or a TfR aptamer-carrying nanoheater. Since the antibody and aptamer of TfR share different binding domains (39), both aptamer and antibody conjugate to the TfR protein simultaneously. The distribution of lipid raft and nonraft domains on the cell membrane and colocalizations of nanoheater with their interested domains were investigated by a dual-color direct stochastic optical reconstruction microscopy (dSTORM). Mander’s coefficients (MOC) were used to quantitatively analyze the colocalization in the merged reconstructed dSTORM images (Fig. 1F) (40, 41). The calculated MOC for CTxB and TfR antibody was 0.223, indicating that the labeled raft and nonraft regions were independently distributed with a weak overlapping. While MOC for raft-anchored nanoheater and CTxB was 0.945, the non-raft-anchored nanoheater and TfR antibody was 0.950, suggesting a high level of spatial overlapping.

The nanoheater-anchored cells were then illuminated using an 808 nm NIR laser for 1 min, waited for 45 s, and then illuminated for another 15 s, and this procedure was repeated three times to achieve sufficient heating. To interrogate the efficient heating of cell membranes on cellular function, we investigated the expression of heat shot protein (HSP) 70, a family of conserved ubiquitously expressed proteins that are strongly upregulated by heat stress (42, 43). The upregulation of HSP 70 after NIR illumination (Fig. 1G) suggested the treated cells responded efficiently to heat stimuli regardless of anchoring domains. Additionally, our control experiments indicated that the DNA origami maintained its structure integrity within the heating temperature range (SI Appendix, Fig. S6), and the nanoheater still anchored on the cell membrane at room temperature for at least 2 h without being internalized (SI Appendix, Fig. S7), suggesting high stability of the nanoheater on the cell membrane.

Nanoheater-Induced Changes in Plasma Membrane Fluidity.

The localized photothermal effect of plasmonic nanoparticles has been widely used as noninvasive therapy for cancers (44–48). Thereby, although the NIR laser-induced solution temperature rise is only up to 42 °C, a relatively low temperature compared with that in photothermal therapy, we still assessed the cell viabilities after nanoheater anchoring and NIR illumination. Indeed, the cell viability assay indicated that 88% of treated cells were still alive (SI Appendix, Fig. S8). Membrane integrity was also evaluated, which showed the plasma membrane remains its integrity (SI Appendix, Fig. S9) (49). All these results indicated the present heating temperature did not lead to significant cell death and membrane disruption.

We then investigated the heat-induced fluidity change of the plasma membrane with a widely used membrane fluidity probe, N,N,N-trimethyl-4-(6-phenyl-1,3,5-hexatrien-1-yl)-benzenaminium, ​4-m​eth​ylb​enz​ene​​sul​fon​a​te ​(TM​A-DPH) (​Fig. 2 A, Inset). TMA-DPH locates close ​to ​the​ water/lipid borde​r a​nd ​its ​flu​ore​scence intensity responds to chan​ges​ in​ physical properties of the acyl chain region of the membrane that affect its ability to rotate (50); thus, the fluorescence anisotropy of TMA-DPH is closely associated with membrane fluidity. The obtained fluorescence polarization values (P value) of nanoheater-treated MCF-7 cells were demonstrated in Fig. 2A and SI Appendix, Fig. S10. After NIR laser illumination, P value for the raft-anchored cells decreased from 0.255 ± 0.001 to 0.191 ± 0.002, and for the non-raft-anchored cells decreased from 0.269 ± 0.001 to 0.195 ± 0.004, while P value of control cells without any treatment remained unchanged (SI Appendix, Fig. S11). The decreased P values indicated the NIR-illumination indeed improved the membrane fluidity of both raft-anchored and non-raft-anchored nanoheaters treated cells.

Fig. 2.

Membrane fluidity changes after nanoheater treatment. (A) Fluorescence polarization (P) values of TMA-DPH stained, nanoheater-treated MCF-7 cells in the presence (+) and absence (−) of NIR laser illumination. P was measured at λ_ex = 355 nm and λ_em = 430 nm. Inset: molecular structure of TMA-DPH. (B) Experimental setup of the FRAP assay under in situ NIR laser illumination. (C) Fluorescence recovery images of lipid raft (Top) and nonraft (Bottom) domains after photobleaching upon various treatments. The radius of the selected region is 2.5 μm for all images. (Scale bar: 10 μm.) (D) Plots of fluorescence intensity of the lipid raft (Left) and nonraft (Right) of the marked area in (C) versus time after photobleaching. Inset: diffusion coefficients of Alexa647 obtained from plots of normalized FRAP data. The mean and SDs were from at least three different samples with an analysis of 15 bleach spots for each experiment.

However, since TMA-DPH possesses no preferences toward different domains, it reflects only the fluidity of whole membranes but not associated with raft or nonraft domains. We then performed fluorescence recovery after the photobleaching (FRAP) experiment to assess the fluidity change of different domains (Fig. 2B) (51). The nanoheater-lighted region was first photobleached, and the recovery signals were recorded to achieve domain-selective fluidity testing (Fig. 2C). As shown in Fig. 2D, the recovery kinetics of Alexa647 fluorophore sped up after laser illumination. For the raft-anchored nanoheater treated cells, the calculated FRAP ratio increased from 45 to 62%, with the diffusion coefficient increasing from 0.11 μm² s⁻¹ to 0.13 μm² s⁻¹ (Fig. 2 D, Top, Inset), while for the non-raft-anchored nanoheater treated cells, the calculated FRAP ratio also increased from 52 to 70%, and the calculated diffusion coefficient increased from 0.14 μm² s⁻¹ to 0.16 μm² s⁻¹ (Fig. 2 D, Bottom, Inset). It’s worth noting that the fluorescence of origami was still visible on the cell membrane after laser illumination, indicating the binding of aptamer remains effective under heating conditions. To further clarify the heat-induced fluidity change, we performed a control experiment to study the FRAP signal of origami (without Au NRs)-treated cells under 808 nm laser illumination; no obvious diffusion coefficient changes were observed (SI Appendix, Fig. S12). Overall, the FRAP results suggested that raft and nonraft regions possess different fluidity changes in response to heat stimuli, which further solidified the domain-dependent heating strategy.

Domain-Dependent Cell Migration Changes after Targeted Heating Treatment.

Membrane fluidity changes alter the motility and exposure of membrane structures, such as antigens, receptors, and adhesive molecules, thereby it has been considered as a key physical property that affects cell adhesion (52, 53), although the precise molecular mechanisms remain unclear. After the treatments with the nanoheater, we observed a change in cell migration behavior by plotting migration trajectories using time-lapse imaging (Fig. 3 A–E).

Fig. 3.

The targeted heating treatment leads to a domain-dependent cell migration change. (A) The migration trajectories of randomly selected cells (n = 10) after various treatments for 24 h. Ten representative cells were selected at time zero and tracked by time-lapse imaging at 20-min intervals for 24 h by the positions of cell nuclei. The correlated quantitative analysis of travel speed (B), spreading area (C), persistence (D), and persistence index (E) of randomly selected migrating cells for each incubation condition. Statistical analysis was done with two-tailed unpaired t test (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). The directivity of the cell movement was described by the persistence index, defined as the ratio of the vector distance (the distance between the start and end of the movement) to the total path length, and the persistence, defined as the time it takes the cell to change its initial direction by 90°. (F) Representative images of MCF-7 cells with different nanoheater treatments after 24 h of cell culture. Cell nuclei were stained by Hoechst staining, F-actin was stained by ActinRed 555, and FAK was stained by Alexa647-labeled antibody. (Scale bar: 10 μm, enlarged: 5 μm.)

From the migration analysis (54), we found that when the nanoheater was anchored at raft domain, cell migration was significantly lowered even in the absence NIR laser illumination. The travel speed dropped from 0.150 ± 0.022 μm/min to 0.091 ± 0.011 μm/min, the spreading area decreased from 995.5 ± 250.2 μm² to 531.2 ± 102.6 μm², while the persistence and persistence index also decreased from 282 ± 73 min to 146 ± 45.3 min, 1.67 ± 0.48 to 1.27 ± 0.29, respectively. The phenomenon of weakened cell migration upon anchoring at lipid rafts is similar to our previous work (55), which was explained as raft-anchored DNA nanostructures blocked interactions between cell surface adhesion receptor, CD 44, and extracellular matrix. Upon heating by NIR, the migration of raft-anchored cells was recovered to the level of untreated cells.

On the contrary, the non-raft-targeted nanoheater-treated cells revealed significantly enhanced cell migration after NIR treatments, the travel speed, and spreading area raised to 0.210 ± 0.035 μm/min and 1572.2 ± 347.1 μm², respectively, and the persistence and persistence index increased to 436 ± 54.8 min and 2.67 ± 1.20, respectively. Compared to the control group of (−) NIR laser illumination, the nonraft nanoheater-anchored cells obtained an almost 1.5 times improved migration ability. In addition to single cell migration analysis, this domain-dependent cell migration changes at the ensemble level were also verified by cells scratch tests (SI Appendix, Fig. S13). Notably, cell migration was only slightly inhibited when treated only with CTxB (SI Appendix, Fig. S14). The migration regulation was weakly dependent on nanoheaters concentration, and 5 nM of nanoheaters was proved to be the optimized concentration (SI Appendix, Fig. S15). Moreover, the targeted heating strategy was also applied on other cell lines (HepG2 and Hela), and similar migration change behaviors were observed (SI Appendix, Fig. S16), suggesting the spatial-resolved heat treatment might have a certain level of spectrum on regulating cell migration behavior.

Cell migration is controlled by a variety of activities combining protrusive and contractile forces, normally generated by integrin and actin (56, 57). Focal adhesion kinase (FAK), a nonreceptor tyrosine kinase, is an adaptor protein that primarily regulates adhesion signaling and cell migration (58), we thereby compared the cell morphologies, expression and distribution of FAK protein after different treatments. After the cells were fixed, ActinRed 555 was used to stain F-actin, and FAK was stained by Alexa647-labeled antibody. As shown in Fig. 3F, in control cells without any treatment, blurred filopodia protrusions were observed at the edge of the cell, a typical phenotype of malignant tumor cells, and FAK protein was expressed in the cytoplasm. After NIR laser illumination, the filopodia protrusions could be more clearly visible in the nonraft nanoheaters-anchored cells, and FAK protein was colocalized in the filopodia protrusions, while only smooth surfaces were observed in the raft-anchored nanoheaters treated cells. According to the western blot results, the expression of FAK protein was decreased in raft (−NIR) group while increased in nonraft (+NIR) group (SI Appendix, Fig. S17). The morphologies of cells after different treatments agreed with the tendency of migration trajectories, which further solidified the conclusion that heating at nonraft domain enhances cell migration, while at raft domain has no significant influences.

Noteworthily, cell migration behavior is closely related to the process of epithelial-mesenchymal transition (EMT) (59, 60). We examined the expression of EMT markers, E-cadherin, and vimentin (SI Appendix, Fig. S18). The raft (−NIR) group showed an increased expression of E-cadherin and a decreased vimentin expression, implying an inhibition of EMT, while the protein expression in the nonraft (+NIR) group indicated a promotion of EMT, which also agreed with our cell migration results.

The differences between heating the raft and nonraft regions in the plasma membrane are quite interesting since it indicates that there might be a different mechanism influencing the assembly of focal adhesion other than heating itself. Our results showed that domain-dependent heating by itself increases membrane fluidity, which possibly elevates the chance of membrane protein oligomerization, such as integrin, through diffusion. On the other hand, it has well-established that focal adhesion assembly (61), especially integrin activation is closely related to lipid rafts (62, 63). Since our assay directly puts effects on rafts in a well-defined spatial resolution, these results trigger our curiosity: What might be the effects of our nanoheaters on membrane heterogeneity?

Activation of Nanoheaters Changed Lipid Packing Behavior.

To answer such question, we first applied a classical polarity-sensitive membrane indicator LAURDAN (64, 65) on cell plasma membrane, and the calculated generalized polarization (GP) values (SI Appendix, Fig. S19) showed no significant changes after NIR laser illumination. This is not surprising since the spatial resolution of confocal microscopy is not sufficient to justify raft and nonraft regions, the obtained GP value was only an ensemble average that was unable to reflect the phase change of raft or nonraft region.

Previously, we have built a great variety of polymer-tethered lipid bilayers to mimic the biophysical properties of lipid rafts. Additionally, through reconstituting transmembrane integrin to raft-mimicking bilayers, we showed raft-sequestration of integrin, either in its activated or nonactivated form, is closely related to the hydrophobic mismatch and line tension (66). These results indicated that in vitro reconstitution systems show advantages in elucidating potential biophysical mechanisms.

Therefore, to further clarify the influence of nanoheaters on lipid organization, we then built an artificial lipid bilayer model as a mimic of the cell membrane to interrogate the domain-dependent heating-induced phase transition and migration regulation. Briefly, the lipid bilayer was built on a glass substrate with DOPC/DPPC/cholesterol mixtures at a 1:1:1 ratio with a well-defined raft-mimicking liquid-order phase (L_o) and corresponding liquid disorder phase (L_d) which has almost the highest line tension at room temperature (67). In this experiment, we used a unique total internal reflection fluorescence microscopy (TIRFM) imaging and confocal fluorescence correlation spectroscopy (FCS) analysis merged in one microscope to explore the phase separation and diffusive of interested domains upon heating treatment. Hereby, we used NBD-PE that is known to enrich in L_o phase as phase marker (68), GM1 and TfR were reconstituted to the lipid bilayers, CTxB, raft- and non-raft-targeted nanoheaters were added following GM1 and TfR reconstitution and imaged using TIRFM. Texas red DHPE were added to the bilayer at 0.01% and used as a marker for membrane diffusivity during FCS analysis; a strong overlapping between NBD-PE and Texas red DHPE was observed under our conditions.

As shown in Fig. 4A, a clear cutting-edge border was observed before the addition of CTxB and raft-targeting nanoheaters, suggesting the successful establishment of a phase separation model. After the addition of CTxB and Alexa647-labeled raft-targeted nanoheater, a strong colocalization between nanoheater and L_o phase was observed (Fig. 4 B, Top), on the contrary, Alexa647-labeled non-raft-targeted nanoheater showed weak colocalization to L_o phase (Fig. 4 B, Bottom). However, after 105 s of NIR illumination, the size of L_o phase was greatly decreased (Fig. 4 C, Top) with the average diameter drop from 14.9 ± 4.4 μm to 7.0 ± 0.6 μm (SI Appendix, Fig. S20). In contrast, when applying non-raft-targeted nanoheaters, there was no obvious change in the size of L_o phases (Fig. 4 C, Bottom).

Fig. 4.

The nanoheaters treatment induces a change of lipid packing behavior in the raft-mimicking lipid mixtures. (A) Images of the lipid bilayer incorporated with a raft-mimicking L_o phase and L_d phase. (Scale bar: 10 μm [the same as below].) (B) Colocalization of raft-targeted nanoheater with L_o phase (Top) and non-raft-targeted nanoheater with L_d phase (Bottom). (C) Representative images of L_o phase before and after L_o-targeted heating (Top) and L_d-targeted heating (Bottom). (D) FCS analysis of L_o and L_d phases before and after L_o-targeted heating. (E) FCS analysis of L_o and L_d phases before and after L_d-targeted heating. (F) Comparison of diffusion coefficients of L_o and L_d phases before and after L_o-targeted heating and L_d-targeted heating. Data were collected from at least three different bilayers with the same protocols. Statistical analysis was done with a two-way ANOVA examination.

Apparent diffusion coefficients at specific regions are closely related to membrane packing properties, whereas the differences between different phases reflect the line tension in between. The diffusion coefficient of Texas Red DHPE was obtained with FCS (Fig. 4D). When applying heating at L_o, FCS curves showed that although the overall lipid diffusivity increased upon NIR laser illumination, the diffusion coefficient of L_o phase increased more significantly and even higher than that of L_d phase (Fig. 4F), indicating the line tension between the two phases declined upon L_o-targeted nanoheater function.

In other words, the nanoheaters functioned on L_o phases homogenize the bilayers that decrease the lipid heterogeneity. When heating the membrane, protein assembly accelerates that provides higher chance for integrin oligomer clustering upon diffusion, moderately increases cellular migration rate. On the other side, the raft-targeted nanoheaters also melt the raft, which reduces the stability of activated integrin clusters. Therefore, the recovered migration of raft-anchored cells was estimated to stem from the synergy of the above two factors (69). When applying nanoheaters target to L_d phases, the lipid diffusion of L_d phases increased. However, it is quite interesting that lipid diffusion in L_o phase greatly decreased, introducing the greatest diffusion barrier between the two domains (Fig. 4 E and F). It is yet to be determined on the potential mechanism, but the elevated line tension could be a great contributor to observed larger focal adhesion and also higher cellular migration speed.

These results, combined with previous in vitro reconstitution studies, help to explain our findings on cell migration behavior (70). When applying the nanoheaters at nonraft regions, the line tension between the raft and nonraft regions increases, which helps to stabilize lipid rafts and possibly stabilize the assembled focal adhesions; therefore, cells obtain a larger spreading area and higher migration rate. On the other hand, when applying nanoheaters at raft regions, the differences between the raft and nonraft regions become smaller, it is therefore hard to form a stable environment that allows the assembly of activated integrin, as indicated by smaller focal adhesions, smaller cell spreading area, and slower cell migration time comparing to the heating at L_d phase.

Membrane Heterogeneity Manipulation Caused Significant Differences in Cellular Signaling.

Our results indicated that the nanoheater application may fundamentally change membrane protein assemblies through the alternation of membrane heterogeneity, we next performed direct data-independent acquisition (DIA) proteomics to further explore the correlated signal transductions. From the migration behavior analysis, we particularly focused on raft (−NIR) and nonraft (+NIR) groups. The distribution of different proteins in each group after different treatments is shown in Fig. 5 A, B, E, and F. For the raft (−NIR) treatment group, compared with the control group (without any treatment), 49 proteins were significantly up-regulated and 90 proteins were down-regulated. While in the nonraft (+NIR) treatment group, 281 and 543 proteins were up- and down-regulated, respectively. Gene Ontology (GO) biological process analysis demonstrated raft (−NIR) and nonraft (+NIR) groups alter biological process (BP), molecular function (MF), and cellular component (CC), including cellular process, protein-containing complex and molecular function regulator activity (Fig. 5 D and H). It is worth noting that in the functional analysis of BP, MF, and CC, the number of differential proteins in non-raft (+NIR)-treated cells was much greater than that in the raft (−NIR) treated cells.

Fig. 5.

Proteomics analysis of MCF-7 cells after the domain-dependent heating treatment. (A) Volcano plot visually showing the number of up-regulated (red dots) and down-regulated (blue dots) proteins after raft (–NIR) treatment with adjusted P < 0.05 and log2(fold changes) > 0. (B) The number of up-regulated (red) and down-regulated (blue) proteins after raft (–NIR) treatment. (C) KEGG pathway analysis of significantly altered proteins in raft (–NIR) group compared to the control group. The size of the circle corresponds to the protein counts (from the reference pathway), and the color corresponds to the adjusted P-value. (D) GO analysis indicates the biological functions of differently expressed proteins, including biological process, molecular function, and cellular components in the raft (–NIR) group. (E) Volcano plot visually showing the number of up-regulated (red dots) and down-regulated (blue dots) proteins after nonraft (+NIR) treatment with adjusted P < 0.05 and log2(fold changes) > 0. (F) The number of up-regulated (red) and down-regulated (blue) proteins after nonraft (+NIR) treatment. (G) KEGG pathway analysis of significantly altered proteins in nonraft (+NIR) group compared to the control group. The size of the circle corresponds to the protein counts (from the reference pathway), and the color corresponds to the adjusted P-value. (H) GO analysis indicates the biological functions of differently expressed proteins, including biological process, molecular function, and cellular components in the nonraft (+NIR) group. (I) A heat map of related cell migration proteins among control, raft (–NIR), and nonraft (+NIR) groups.

Moreover, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis demonstrated that both raft (−NIR) (Fig. 5C) and nonraft (+NIR) (Fig. 5G) groups were related to signal transduction in cell migration. For the raft (−NIR) group, many signaling pathways, including Peroxisome proliferation-activated receptors (PPARs), regulation of actin cytoskeleton, and cell adhesion molecules, were regulated. Among them, the regulation of actin cytoskeleton and cell adhesion molecules were directly involved in cell migration. Particularly, PPARs are members of the proliferation-activated receptor family that are involved in cell cycle regulation, inhibition of cell growth after activation, and other pathways in the occurrence and development of tumors (71–73). For the nonraft (+NIR) group, oxidative phosphorylation, focal adhesion, and extracellular matrix (ECM) receptor interaction pathways were mainly affected. Moreover, the cell migration-related proteins were significantly altered after raft (−NIR) or nonraft (+NIR) treatment (Fig. 5I). For example, CARMIL-2, which blocks the activity of the F-actin heterodimer capping protein (CP) in the front edge of migrating cells (74), was decreased after raft (−NIR) treatment, while increased after nonraft (+NIR) treatment, indicating the heterogeneity alteration could indeed regulate cell migration behavior. We also examined proteomic results of raft (−NIR) versus raft (+NIR) and nonraft (−NIR) versus nonraft (+NIR) (SI Appendix, Fig. S21), and the results demonstrated similar pathway changes, which further support our conclusion.

We next harness the ability to regulate cell migration with remote NIR laser to exemplify in vivo therapeutic applications. The enhanced cell migration by heating at the nonraft domain was used to promote wound healing in vivo (75, 76). A mouse wound model was established (Fig. 6A), non-raft-anchored nanoheater was cast at the wound, and an 808 nm NIR laser was then introduced to illuminate the wound for 4 min.

Fig. 6.

In vivo wound healing performance of the nanoheaters treatments. (A) Schematic illustration of the mouse wound healing model treated by the non-raft-targeted nanoheater. (B) Images of wound healing in mice within 8 d after treatments. (Scale bar: 5 mm.) (C) Closure rate of wounds in 8 d after treatments. The variation was represented by the SD of three independent replicates in all graphs. (D) Weight changes for each group of mice. The variation is represented by the SD of three independent replicates in all graphs.

The wound closure percentages and body weights were measured every 2 d after treatments. As shown in Fig. 6B, the nonraft (+NIR) group revealed the highest healing ability with a 92.3% wound closure percentage after 8 d, while the wound closure percentage in control (−NIR), nonraft (−NIR), and control (+NIR) group were only 61.2%, 62.2%, and 62.1%, respectively (Fig. 6C). Moreover, only a small fluctuation in body weight was observed for the non-raft-targeted nanoheater-treated mice (Fig. 6D), indicating the excellent biosafety of the nanomolecular heater.

Discussion

In summary, we developed a nanoheater system that specifically heats interested domains of the cell membrane to improve its fluidity. We found the membrane fluidity changes of different domains further interfere phase separation of membranes.

The fundamental of these nanoheater systems is to regulate lipid heterogeneity in the plasma membrane, which is not only a key regulator for cell migration but also varies in signal pathways. Since different types of cells have different dominant signal pathways, certain outcomes other than cell migration are out of the scope of this research. However, this nanoheater system might have a broader spectrum of applications that we are currently unaware of. Moreover, considering the heating is a result of photothermal treatment, besides the domain-dependent spatial resolution, it also possesses a higher temporal resolution. Therefore, a remote regulation of cellular behaviors at different cell cycles is also anticipated.

The ability of domain-dependent regulating cell membrane heterogeneity enables us to explore the underestimated biological function of lipid rafts, providing a possibility to set a direct correlation of physical (size, lifetime, and so on) and compositional features of lipid raft with physiological functions (differentiation, programmed death, and metabolism) on both cells and potentially tissue of cells.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.