AIE-enabled transfection-free identification and isolation of viable cell subpopulations differing in the level of autophagy

ABSTRACT Elevated macroautophagy/autophagy, typically characterized by increased autophagosome accumulation, occurs in a wide variety of physiological and pathophysiological processes, but the current methodology for studying autophagy aberration in native non-transfected cells is rather limited. Here we show that LKT, an engineered molecular probe composed of a cell-penetrating peptide, an LC3-interacting motif and the aggregation-inducedemission (AIE) luminogen tetraphenylethylene, achieved robust identification and isolation of viable autophagy-varying cell subpopulations without the need of foreign reporter gene expression. Non-fluorescent in water, LKT fluorescence is activated upon interaction with liposomes in an AIE-dependent fashion, and the presence of LC3 on the liposome membrane dramatically boosted LKT fluorescence enhancement. In LKT-treated GFP-LC3 HeLa cells, induction of autophagy with rapamycin or trehalose, and blockade of autophagy with chloroquine, both produced bright GFP-LC3-colocalizing LKT puncta, leading to an increase in LKT fluorescence that facilitated efficient separation of cells based on the level of autophagosome accumulation. Using fluorescence-activated cell sorting, we were able to isolate cell subpopulations varying in the level of basal autophagy from a variety of cultured cell lines and primary cells. In a proof-of-concept study, we isolated autophagy-high and autophagy-low subpopulations from differentiated THP-1 cells and revealed that the autophagy-high THP-1 cells, compared to their autophagy-low counterparts, exhibited a higher level of NLRP3 protein expression and a stronger NLRP3 inflammasome activation following nigericin challenge. Our work demonstrated the unique power of the AIE technology and LKT, filling a void, should prove valuable for autophagy research. Abbreviations 3-MA, 3-methyladenine; AIE, aggregation-induced emission; AIEgens, aggregation-induced emission luminogens; ATG5, autophagy related 5; BMDM, bone marrow-derived macrophage; CQ, chloroquine; DiD, 1,1’-dioctadecyl-3,3,3’,3’-tetramethylindodicarbocyanine perchlorate; DiO, 3,3’-dioctadecyloxacarbocyanine perchlorate; DMSO, dimethyl sulfoxide; d-THP-1, differentiated THP-1; FACS, fluorescence activated cell sorting; FBS, fetal bovine serum; FCCP, carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone; GABARAP, GABA type A receptor-associated protein; GFP, green fluorescent protein; HBSS, Hanks’ balanced salt solution; HPLC, high-performance liquid chromatography; HRP, horseradish peroxidase; IL1B, interleukin 1 beta; KT, an AIE probe composed of a cell-penetrating peptide and an AIEgen tetraphenyl ethylene; LC3-II, lipidated LC3; LDH, lactate dehydrogenase; LIR, LC3-interacting region; LKR, engineered molecular probe composed of an LC3-interacting peptide, a cell-penetrating peptide and a non-AIE fluorescent molecule rhodamine; LKT, engineered molecular probe composed of an LC3-interacting peptide, a cell-penetrating peptide and an AIEgen tetraphenyl ethylene; LPS, lipopolysaccharide; MAP1LC3/LC3, microtubule associated protein 1 light chain 3; MEF, mouse embryonic fibroblast; mRFP, monomeric red fluorescent protein; NHS, N-hydroxysuccinimide; NLRP3, NLR family pyrin domain containing 3; PBS, phosphate-buffered saline; PCC, pearson’s correlation coefficient; PL, photoluminescence; PMA, phorbol 12-myristate 13-acetate; RAP, rapamycin; RIM, restriction of intramolecular motions; s.e.m., standard error of the mean; SPR, surface plasmon resonance; SQSTM1/p62, sequestosome 1; TAX1BP1, Tax1 binding protein 1; TPE, tetraphenylethylene; TPE-yne, 1-(4-ethynylphenyl)-1,2,2-triphenylethene; Tre, trehalose; u-THP-1: undifferentiated THP-1; UV-Vis, ultraviolet visible


Introduction
Macroautophagy/autophagy is a dynamic and conserved degradation pathway, characterized by a series of steps including the emergence of cup-shaped phagophore, cargo sequestration, formation of double-membraned autophagosomes, fusion between autophagosomes with lysosomes to form autolysosomes, and cargo degradation [1][2][3].A critical yet complex role of autophagy has been amply demonstrated in the development and therapy of a great variety of human diseases, such as cancer, neurodegenerative diseases, diabetes, autoimmune and cardiovascular disease [4][5][6].It is generally perceived that basal autophagy, operating at a low level, exists in essentially every mammalian cell to maintain the homeostasis and survival of the cell and plays critically-important roles in many physiological processes [7][8][9][10][11][12][13].However, circumstantial evidence indicates that, for a given cell type, the level of basal autophagy varies considerably among individual cells.Exemplary works are the reports revealing that approximately one-third of aged hematopoietic stem cells exhibit high autophagy levels to maintain a low metabolic state with robust long-term regeneration potential [14], and that autophagy variation within a cell population determines cell fate through selective degradation of PTPN13/Fap-1 (tyrosine-protein phosphatase nonreceptor type 13/Fas-associated phosphatase 1) [15].Conversely, cellular autophagy is frequently elevated to high levels in response to physical, chemical and biological stress, and this enhanced autophagy has been closely linked to many physiological aberrations and pathological conditions [16][17][18][19].Identification and isolation of viable and fully-functional cell subpopulations differing in the autophagy level would greatly benefit the research that concerns either basal or induced autophagy.However, this line of work has been hampered by a lack of convenient and reliable methodology.Various acidotropic dyes, such as acridine orange, monodansylcadaverine and Cyto-ID, have shown certain capability in monitoring autophagy [20,21], and Cyto-ID in particular has been used for autophagy-based cell sorting [22], but the specificity of these agents is questionable, as how they interact with the autophagy machinery remains largely unknown.A similar limitation applies to DAL-green and DAP-green, two fluorescent molecules that have the capability of being incorporated into autophagosome membrane and are useful for visualizing newly-formed autophagosomes [23].A more specific approach is to exploit one of the autophagy related proteins, with the MAP1LC3 (microtubule associated protein 1 light chain 3) and GABARAP (GABA type A receptor-associated protein) family of proteins (collectively called LC3 hereafter) being the most widely used, as lipidated LC3 (LC3-II) remains the only protein marker that is reliably associated with completed autophagosomes [24].Indeed, LC3 conversion by western blotting, measuring the conversion of cytosolic free LC3 to membrane-conjugated LC3, is perhaps the most widely used assay for assessing autophagic level.As LC3 lacks intrinsic fluorescence and cannot easily be tracked, exogenous expression of LC3 fused to a fluorescent protein such as green fluorescent protein (GFP) is generally deployed.The fluorescence of GFP-LC3 is quenched in the acidic environment of autolysosomes, thus cells with high autophagy flux would exhibit diminished GFP fluorescence and be separated from autophagy-low cell subpopulations via fluorescence activated cell sorting (FACS) [25,26], and the methodology based on this principle has been successfully applied to enable the exemplary studies mentioned above [14,15].In addition to LC3, selective autophagy receptors specifically degraded during autophagy, such as SQSTM1/p62 (sequestosome 1) and NBR1 (NBR1 autophagy cargo receptor), can also be used for flow cytometric analyses of autophagy after fusion to GFP or HaloTag as a reporter [27,28].The requirement for a foreign reporter in all of the above cases, however, severely restricts its applicability for difficult-to-transfect cell lines and primary cells, in addition to the undesirable complications arising from the over-expression of the foreign reporter.Furthermore, while FACS sorting based on GFP-LC3 fluorescence change enables separation and isolation of cells with a high autophagic flux (as occurred under autophagy induction), it does not distinguish viable cells in a state of autophagic blockade.Isolation of fully viable and functionally-competent cell subpopulations differing in the autophagy level, regardless of the state of autophagy and without the need of foreign reporter expression, is highly desirable but currently unattainable.
A powerful emerging technology in fluorescent materials is aggregation-induced emission (AIE), which refers to a unique phenomenon that some fluorogens having both a twisted conformation and molecular rotators (or vibrators) are weakly emissive when molecularly dissolved in solution but highly fluorescent as aggregates [29][30][31].Restriction of intramolecular motions (RIM), including restriction of intramolecular rotations and restriction of intramolecular vibrations, has been proposed to be the primary mechanism underlying the enhanced fluorescence of AIE luminogens (AIEgens) in an aggregated state [32,33].The rapid development of AIE research in the recent years has greatly promoted the application of AIEgens in the biomedical field, especially for sensing and imaging [34,35].Notably, lysosome-targeting and mitochondria-targeting AIEgens have been utilized to visualize autophagy and mitophagy, respectively, but these probes are not specific to the autophagic process and have not been used for autophagy-based cell isolation [36,37].Given that covalent conjugation of cytoplasmic LC3 protein to autophagosome membrane is the characteristic hallmark of autophagy, we hypothesized that an AIE probe with the ability to specifically interact with LC3 might display sufficient difference in fluorescence intensity upon binding to membrane-conjugated versus free LC3 and thus facilitates sorting of individual cells differing in the autophagy level.In this report we demonstrate the feasibility of this approach and also reveal, in a variety of cell types, prevalent heterogeneity in the level of basal autophagy.

Synthesis and characterization of LKT
In our design, the molecular probe intended for autophagybased cell identification and sorting would encompass three properties: cell internalization, LC3-specific interaction and AIE characteristics.LKT, containing a cell-penetrating poly-K motif (KKKKKKKKKK) [38], an LC3-interacting region (LIR, with the sequence DDDWTHL) from the SQSTM1 [39] and the propeller-shaped AIEgen tetraphenylethylene (TPE) [40], fulfilled these requirements (Figure 1A and S1A).In addition to promoting cell uptake, the poly-K motif also served to increase the solubility of LKT in aqueous solution.LKT was synthesized via a click reaction between the azide-modified peptide LK and 1-(4-ethynylphenyl)-1,2,2-triphenylethene (TPE-yne), followed by purification with high performance liquid chromatography (HPLC) (Fig. S1B and S1C).The identity of LKT was verified by mass spectrometry and ultraviolet visible (UV-Vis) spectral analyses (Fig. S1D and Figure 1B, respectively).Similar to the peptide LK, LKT did not form any sizable particles in water, suggesting a single-molecule existing state, while TPE was aggregated with a size of close to 1,000 nm, as revealed by dynamic light scattering (Figure 1C).In agreement with the expected AIE characteristics, LKT emitted low fluorescence in water but became highly fluorescent when deposited on a film (Figures 1D,E), while the aggregated TPE was already highly fluorescent in water (Figure 1D).LKT also exhibited excellent photostability after internalization into cells, displaying minimally-diminished fluorescence after repeated excitations (Fig. S1E), a typical and useful feature of AIE molecules.As expected, LKT interacted with purified LC3 protein, with a Kd of approximately 0.312 ± 0.051 μM determined by surface plasmon resonance (SPR; Figure 1F).This interaction was dependent on the LIR motif, as KT, a poly-K-conjugated TPE probe but without LIR (Fig. S1F-J), did not bind to LC3 (Figure 1F).In agreement with these results, pre-coated LC3 effectively "pulled down" LKT molecules in solution, while pre-coated bovine serum albumin (BSA) was minimally active (Figure 1G).LC3 also "pulled down" some KT molecules, presumably due to the "sticky" nature of LC3 [41], but the extent of KT pull-down was much less than that of LKT (Figure 1G, right panel).These results verified that LKT is an AIE probe with a relatively high affinity to interact with LC3.

LC3-promoted LKT fluorescence enhancement upon binding to liposomes
The extent of LC3 lipidation, resulting from covalent conjugation of the cytosolic LC3 protein to the phospholipid phosphatidylethanolamine on both the inner and outer autophagosome membranes, is a reliable measure of the autophagy level [24].To provide a preliminary indication that LKT might be able to differentiate cells with varying degree of LC3 lipidation, we chemically conjugated LC3 protein to liposomes (Fig. S2A).The presence of LC3 on liposomes was confirmed by western blotting (Fig. S2B).LKT emitted low level of fluorescence in water, but increased LKT fluorescence was detected when unconjugated liposome was added (Figure 2A).In contrast, LKR, which differed from LKT by having a non-AIE fluorescent molecule rhodamine in place of TPE (Fig. S2C) and also existed as single molecules in aqueous solution (Fig. S2D), displayed no change in fluorescence upon the addition of liposome (Fig. S2E), indicating that the AIE property was necessary for the observed fluorescence enhancement of LKT.Importantly, the addition of LC3-conjugated liposome resulted in a dramatic and statistically-significant further increase in LKT fluorescence over that achieved by the addition of unconjugated liposome (Figures 2A,B).These results suggested that a nonspecific interaction of the hydrophobic AIE molecule TPE with the liposome membrane, through an AIE mechanism, facilitated increased fluorescence emission of LKT, while the presence of LC3 on the liposome membrane, through a high-affinity interaction with the LIR motif of LKT, further enhanced TPEmembrane interaction, possibly by more efficiently bringing LKT molecules to the proximity of liposome membrane.Consistent with this hypothesis, enhanced LKT fluorescence in solution was visible by naked eye after the addition of either unconjugated or LC3-conjugated liposome, with the LC3-conjugated liposome eliciting a stronger enhancement effect, but nearly all of the fluorescence was observed at the bottom of the Eppendorf tube after centrifugation, which sedimented liposomes but not unbound LKT (Figure 2C), demonstrating that increased fluorescence occurred for liposome-bound LKT but not for LKT in solution.In further support, addition of increasing concentrations of free soluble LC3 protein elicited minimal change in LKT fluorescence (Figure 2D).Conversely, an increase in LKT concentration did significantly enhance LKT fluorescence (Figure 2E), but the magnitude of this fluorescence change was much less than the LKT fluorescence increase elicited by the addition of LC3-conjugated liposomes (Figure 2A).KT, an AIE probe that lacked the LIR motif and did not interact with LC3 (Figure 1F), also exhibited increased fluorescence in the presence of liposome, but the conjugation of LC3 to the liposome diminished, rather than enhanced, this response (Figure 2F and S2F), suggesting that the presence of LC3 on the liposome membrane decreased the space available for the nonspecific KTliposome interaction.Collectively, the above results revealed an AIE-dependent fluorescence enhancement upon LKT-liposome interaction, an effect that was dramatically boosted by the presence of LC3 on the liposome membrane.

Formation of GFP-LC3-cololalizing LKT puncta upon autophagosome accumulation
LC3 conjugation to autophagosome membranes, commonly observed as LC3 puncta, is a reliable measure of autophagosome accumulation, the level of which reflects a net change in autophagosome formation and turnover.Indeed, both rapamycin (RAP) and trehalose (Tre), known to induce more autophagosome formation than turnover, caused prominent green puncta formation in GFP-LC3 HeLa, the HeLa cells that stably expressed GFP-LC3 (Figure 2G).Robust GFP-LC3 puncta was also observed after the treatment of chloroquine (CQ), an agent that blocked autophagosome turnover [42], among many other pathways it affected.Co-treatment of these cells with LKT produced blue punctate dots, which significantly colocalized with GFP-LC3 puncta as revealed by Pearson's correlation coefficient (PCC) values of 0.47, 0.66 and 0.69 for RAP, Tre and CQ treatments, respectively (Figure 2G).In CQ-treated HeLa cells, we also observed many LKT puncta, most of which co-localized with the endogenous LC3 puncta as visualized by immunofluorescence, indicating that LKT could also detect the endogenous LC3-II (Fig. S2G).Notably, the majority of phosphate-buffered saline (PBS)treated HeLa cells displayed little LKT fluorescence, but a proportion of cells did exhibit significant GFP-LC3colocalizing LKT puncta, suggesting heterogeneity in the basal autophagy level (Fig. S2H).Conversely, treatment with CQ, which increased the number of autophagosomes, led to extensive LKT positive puncta, and correspondingly enhanced LKT fluorescence, in the majority of cells (Fig. S2I).Knockdown of ATG5 (autophagy related 5) largely abolished both GFP-LC3 and LKT puncta formation elicited by either Tre or CQ, while having little effect on the low occurrence of KT puncta (Fig. S2J and S2M), strongly supporting that LKT puncta resulted from an interaction with LC3-II on the autophagosomes.Conversely, knockdown of selective autophagy receptors SQSTM1 and TAX1BP1 (Tax1 binding protein 1) did not affect LKT puncta formation elicited by either Tre or CQ (Fig. S2K-M), suggesting that LKT puncta were not due to LC3 incorporation into aggresomes facilitated by selective autophagy receptors.
To assess how autophagosome-lysosome fusion affects LKT fluorescence, we examined HeLa cells expressing monomeric red fluorescent protein (mRFP)-GFP-LC3B after treatment with CQ or Tre in the presence of LKT.Tre, which triggered autophagy with normal flux (resulting in decreased SQSTM1; Fig. S2N), showed many red (mRFP) puncta but fewer colocalizing green (GFP) puncta, resulting in mostly red dots (autolysosomes) in the merged image, consistent with GFP quenching and degradation in the acidic autolysosomes.Conversely, CQ, which blocked autophagosome-lysosome fusion (as evidenced by dramatically increased SQSTM1, a common substrate of autophagy; Fig. S2N), induced as many red puncta as the green puncta, resulting in mostly yellow dots (autophagosomes) in the merged image, a result to be expected as GFP fluorescence would not be decreased in this situation (Figure 2H, with the quantified results shown in Fig. S2O).Importantly, the strong blue puncta under both Treinduced and CQ-blocked autophagy suggested that LKT would exhibit similarly-enhanced fluorescence in autolysosomes as in autophagosomes.As expected, LKR also produced red punctate dots that colocalized with GFP-LC3 puncta upon autophagy induction by Tre (Fig. S2P).Moreover, the time course of LKT puncta formation coincided well with that of GFP-LC3 puncta formation after CQ treatment (Figure 2I).These results indicated that LKT has the potential to replace GFP-LC3 as a much more convenient transfection-free reporter for monitoring the occurrence of autophagy.It should be pointed out that the LIR motif of LKT, as derived from the SQSTM1 protein, can interact with different members of the LC3 family [24] including LC3A, LC3B2, LC3C, GABARAP, GABARAPL1 and GABARAPL2, in addition to LC3B which was the form of LC3 in GFP-LC3.This may explain why only approximately 50% of LKT-positive puncta were also GFP-LC3 positive (Fig. S2Q; quantified from Figure 2G), as LC3 family members other than LC3B were known to participate in autophagosome formation.Conversely, LKT could offer a broader coverage than GFP-LC3 in visualizing autophagosomes.Treatment of cells with the detergent saponin can induce non-autophagic GFP-LC3 aggregation [43].In agreement, we observed rapid GFP-LC3 dot formation in saponin-treated GFP-LC3 HeLa cells (Fig. S2R).However, saponin failed to induce LKT puncta, suggesting that LKT interaction with LC3 was unable to turn on LKT fluorescence even when LC3 is aggregated (Fig. S2R).

Enhanced LKT fluorescence in the cells with increased autophagosome accumulation
The ability to emit enhanced fluorescence upon binding to liposome-conjugated LC3 suggests that LKT would show a higher fluorescence in the cells containing more autophagosomes (thus more LC3-II), as is the case under blocked autophagy and many of the induced-autophagy situations.Indeed, HeLa cells co-treated with LKT and Tre, an autophagy inducer that elicited more autophagosome formation than turnover, while exhibiting increased autophagosome accumulation as evidenced by LC3 conversion (Figure 3A and S3A), also displayed higher LKT fluorescence under flow cytometric analyses with excitation at 355 nm (Figure 3B and S3B).Conversely, 3-methyladenine (3-MA), an inhibitor of autophagosome formation, significantly reduced both LC3 conversion (Figure 3A) and LKT fluorescence enhancement (Figure 3B and S3B) elicited by Tre.Similar LKT fluorescence change under treatment of Tre and 3-MA was observed in the mouse embryonic fibroblasts (MEF) and the undifferentiated THP-1 (u-THP-1) cells (Fig. S3C-F).Notably, Tre-enhanced cellular fluorescence was observed for LKT but not the non-AIE probe LKR in GFP-LC3 HeLa cells (Figure 3C), demonstrating that the AIE property was essential for the probe to respond to autophagy induction.Largely in agreement with the reported results [25], GFP-LC3 displayed diminished fluorescence upon autophagy induction with Tre (Figure 3C), although only a small change was observed in our case, possibly because of heterogeneity in the GFP-LC3 expression level for the GFP-LC3 HeLa cells maintained in our laboratory.Conversely, no GFP-LC3 fluorescence change was observed in GFP-LC3 HeLa cells treated with CQ (Figure 3D), consistent with the notion that CQ acts mostly by blocking autophagosome-lysosome fusion [42], thus unable to elicit sufficient GFP-LC3 quenching.In contrast, significant LKT fluorescence increase was observed following treatment of HeLa cells with either RAP or CQ, demonstrating that LKT, unlike GFP-LC3, was capable of detecting both induced and blocked autophagy (Figure 3E and S3G).Notably, a more profound LKT fluorescence enhancement was seen in CQ-treated cells as compared to RAP-treated cells (Figure 3E, right panel), suggesting that LKT may be able to distinguish between autophagy-induced and autophagy-blocked cell subpopulations.To further demonstrate this, we labeled Tre-and CQtreated HeLa cells with green (3,3'dioctadecyloxacarbocyanine perchlorate, DiO) and red (1,1'-dioctadecyl-3,3,3',3'-tetramethylindodicarbocyanine perchlorate, DiD) dyes, respectively, and mixed them before conducting flow cytometric analyses.LKT fluorescence partially separated the green and red cells into two, largely overlapping peaks (Figure 3F).A similar separation between RAP-and CQ-treated cells was also observed (Fig. S3H).However, it should be pointed out that certain autophagy flux-elevating conditions may elicit a higher rate of autophagosome turnover than the rate of autophagosome formation, leading to decreased autophagosome accumulation and diminished LKT fluorescence.An example is starvation under certain conditions.Similar to Tre or RAP treatment, HeLa cells treated with Hanks' balanced salt solution (HBSS) for 6 h exhibited elevated autophagic flux as evidenced by increased SQSTM1 degradation (Fig. S3I), but in contrary to Tre treatment, starvation resulted in decreased autophagosome accumulation (Fig. S3I) and lower LKT fluorescence, with the latter being vividly demonstrated by flow cytometric analyses of dye-coded cells (Fig. S3J).These results were consistent with published reports [44].Thus, high LKT fluorescence identifies cells with increased autophagosome accumulation but not necessarily those with increased autophagic flux.
To assess the impact of autofluorescence and potential LKT fluorescence enhancement from nonspecific interactions with cellular membranes, we analyzed cell fluorescence change in response to Tre treatment, which triggered robust LKT puncta formation (Figure 2G).Upon excitation at 355 nm, HeLa cells displayed a weak autofluorescence, which did not change by Tre treatment (Figure 3G).KT treatment led to a significant increase in fluorescence, the extent of which was relatively constant for a given cell type and unaltered by the presence of Tre, suggesting fluorescence enhancement from nonspecific TPE interaction with intracellular membranes.Compared to KT, LKT displayed higher fluorescence in the basal (PBS-treated) state, consistent with the notion that some of the HeLa cells had relatively high basal autophagy level and would show increased fluorescence in response to LKT but not KT.Importantly, Tre treatment led to a further, statistically-significant increase in LKT fluorescence over PBS-treated cells, a response not observed under KT treatment (Figure 3G).This LIR-dependent fluorescence enhancement forms the basis for LKT to measure the level of autophagy.
To rule out the possibility that the enhanced LKT fluorescence was due to increased cellular internalization of LKT rather than elevated autophagy level, we added CQ to LKTpretreated HeLa cells after removing uninternalized LKT in the medium.No more LKT cellular uptake was possible in this situation, but we still observed a significant LKT fluorescence increase at 6 h post CQ addition (Figure 3H), similar to that seen under LKT and CQ co-treatment, strongly arguing that the enhanced LKT fluorescence reflected increased LC3 conversion rather than LKT cellular internalization.
Additional experiments further validated the usefulness of LKT as an autophagy probe.LKT exhibited minimal toxicity and autophagy-modulating activity in all of the cell types we examined, including HeLa, MEF and u-THP-1 cells (Fig. S3K-O).Furthermore, LKT had minimal effect on the level of autophagosome accumulation under Tre or CQ treatment (Fig. S3P and S3Q), starvation-elicited SQSTM1 degradation (Fig. S3I), and mitophagy induced by the mitochondria-damaging agent carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone (FCCP) (Fig. S3R).LKT exhibited minimal effect on the level of Tremanifested autophagosome-lysosome fusion as revealed by the same pattern of strong red but weak green signals in mRFP-GFP-LC3B-expressing HeLa cells in the absence and presence of LKT, and the corresponding colocalization analyses, including a pixel intensity analyses of red (RFP) and green (GFP) from a line, PCC analyses and Mander's coefficient analyses (Fig. S3S and Table S1).These results strongly indicated that LKT did not affect the proceeding of autophagy including cargo degradation.Finally, similar LKT fluorescence enhancement in response to Tre was also observed when probe excitation was performed at 405 nm, which is available on most of the flow cytometers, rendering the probe highly convenient for use (Fig. S3T).

LKT-enabled sorting of cell subpopulations differing in the autophagy level
To assess the capability of LKT in isolating cells based on autophagy level, we mixed HeLa cells treated with LKT alone and those treated with LKT plus CQ.Flow cytometry displayed a single peak of LKT fluorescence, but fluorescence-activated sorting revealed that the 25% cell fraction with the highest LKT fluorescence (F-high) exhibited a much higher level of LC3-II than the 25% cell fraction with the lowest LKT fluorescence (F-low) (Figure 4A), indicating that LKT effectively segregated cells differing in the level of blocked autophagy.The same sorting results were obtained by mixing cells treated with LKT alone and those treated with LKT plus Tre (Fig. S4A), demonstrating the capability of LKT in sorting cells differing in the level of induced autophagy.As heterogeneity in the level of basal autophagy has been reported for cultured tumor cell lines as well as for hematopoietic stem cells [14,15], we conducted LKT fluorescence-based sorting of cultured HeLa cells after incubation with LKT for 6 h.Compared to the F-low cells, the F-high cells had more LC3-II (Figure 4B), indicating a higher level of basal autophagy in these cell subpopulations.In support, increased number of autophagic vacuoles were observed under electron microscopy in F-high cells as compared to F-low cells (Figure 4C, with quantified results shown in Fig. S4B).Moreover, prominent LKT blue puncta, co-localizing with the green GFP-LC3 puncta, were observed in F-high, but not F-low, cell subpopulations isolated from GFP-LC3 HeLa cells (Figure 4D, with quantified results shown in Fig. S4C).
The F-high and F-low cell subpopulations exhibited little difference in the level of cell death, determined to be around 7% by ANXA5/annexin V apoptotic analyses (Fig. S4D).Treatment with CQ dramatically increased LC3-II in both F-high and F-low cells, suggesting normal autophagic flux in these cell subpopulations (Figure 4E).Using the same sorting scheme (Fig. S4E), we revealed a higher level of LC3-II in F-high cells than in F-low cells, for all of the four other cell types we have examined, namely MCF-7, u-THP-1, MEF and mouse bone marrow-derived macrophage (BMDM) (Figure 4F).
To further assess whether a smaller difference in the autophagy level can be detected by LKT, we conducted sorting of untreated HeLa cells (thus representing basal state of autophagy) and isolated two cell subpopulations with autophagy level in the adjacent low-medium range: subpopulation 1 (10%-30% of the LKT fluorescence distribution) and subpopulation 2 (30%-50% of the fluorescence distribution).The difference in autophagy level between these two cell subpopulations was very low, as would be expected, but nevertheless western blotting still detected a difference, visible by naked eye, in LC3-conversion (Fig. S4F), demonstrating the robustness of the LKT probe in distinguishing and separating cells with minimal difference in the autophagy level.
To assess how long the probe requires to label LC3B in cells under the basal autophagy state, we performed a flow cytometric analyses after incubation of LKT for differing length of time in HeLa cells.The results indicated that significant LKT fluorescence enhancement was observed at 2 h, with near-maximum fluorescence achieved at 6 h, following LKT treatment (Fig. S4G).Conversely, for HeLa cells pre-treated with CQ, a halfhour incubation with LKT was sufficient to detect LC3 puncta, but optimal detection required at least 2 h, as revealed by fluorescent microscopy (Fig. S4H).Thus, a 2 to 6 h incubation with LKT is generally recommended for analyzing and sorting unstimulated cells, but it may be reduced to as short as half an hour for cells with high autophagy levels.
Collectively, the above results demonstrated that heterogeneity in the basal autophagy level was prevalent in various cell lines and primary cells, and that in all cases LKT reliably sorted cell subpopulations differing in the autophagy level.

A preliminary functional study of autophagy-differing cells sorted by LKT
For LKT to be useful for autophagy research, a critical requirement would be that the isolated cells can be recultured and amenable for functional studies.As a proofof-concept, we have chosen THP-1, a cell line useful for inflammasome research but notoriously difficult to transfect.We first verified that LKT had no inflammasomeactivating activity and did not affect nigericin-induced NLRP3 (NLR family pyrin domain containing 3) inflammasome activation in phorbol 12-myristate 13-acetate (PMA)differentiated THP-1 (d-THP-1) cells (Fig. S5A).Next, we conducted flow cytometric cell sorting of d-THP-1 cells after a 4 h incubation with LKT and isolated F-high and F-low cell subpopulations (Figure 5A, left panel).F-high subpopulation of d-THP-1 cells displayed a much higher basal autophagy level than F-low cells, as revealed by elevated LC3-II (Figure 5A, right panel, with the quantified result in Fig. S5B) and more abundant autophagic vacuoles in the cytosol under a transmission electron microscope (Figure 5B, with the quantified result in Fig. S5C).Both of the F-high and F-low cell subpopulations were fully viable, exhibiting less than 3% cell death (Fig. S5D).Interestingly, F-high d-THP-1 cells expressed a higher level of NLRP3 (Figure 5C) than F-low cells, while little difference was observed for other components of NLRP3 inflammasome, including CASP1 (caspase 1) precursor, CASP1 p20 subunit, IL1B (interleukin 1 beta), PYCARD/ ASC (PYD and CARD domain containing) and GSDMD (gasdermin D) (Fig. S5E).In agreement with higher NLRP3 expression, F-high cells exhibited stronger nigericininduced NLRP3 activation, as revealed by a threefold increase in IL1B secretion (Figure 5D and S5E) and 2.8 times as much in LDH (lactate dehydrogenase) release, an indicator of pyroptotic cell death (Figure 5E), respectively, as compared to the F-low cells.Thus, high basal autophagy in the F-high d-THP-1 cells was associated with both increased NLRP3 expression and enhanced NLRP3 activation in response to NLRP3 stimulus, but why this is the case remained largely unclear.Compared to autophagy-low subpopulation, the autophagy-high cells also displayed a higher level of SQSTM1 (Figure 5F), suggesting blocked autophagic degradation, which would explain increased NLRP3 inflammasome activation.However, CQ treatment further increased LC3-II in F-high as well as in F-low cells (Fig. S5F), suggesting mostly normal autophagic flux, although we could not rule out the possibility of a partial blockade of autophagic flux in the F-high cells.Further investigation is needed to clarify the issue.Notably, the autophagy level of different subpopulations of THP-1 was stable, with the same relative level (high, median or low) observed after a 24 h period of culture post sorting (Fig. S5G).This result differed from that of BJAB lymphoma cells, which reportedly showed a tendency to revert to a common autophagy level for different cell subpopulations after sorting [15], suggesting that the stability of autophagy level is cell type-dependent.Collectively, the above results demonstrated the ability of LKT to isolate viable autophagy-differing cell subpopulations amenable for functional studies.
Based on the above findings, we propose a working model for LKT (Figure 5G).After cellular internalization, presumably aided by the poly-K motif, LKT finds and interacts with cytosolic LC3 through the LIR motif, but this interaction does not lead to LKT fluorescence change.Upon the occurrence of autophagy, LKT-bound LC3 is recruited to the autophagosomes.We propose that the hydrophobic AIE molecule TPE, upon arriving at the surface of autophagosomes, would intercalate into the lipid bilayer, resulting in enhanced fluorescence due to RIM.Alternatively, the internalized LKT may find and interact with LC3 that has already conjugated to autophagosome membranes, again leading to TPE intercalation into lipid bilayer and LKT fluorescence enhancement.In short, the enhanced LKT fluorescence originates from the interaction of LKT with lipid membranes, while LC3 boosts this interaction by efficiently bringing LKT to the proximity of autophagosome membranes.

Discussion
In this work, addressing the unmet need in autophagy research, we have created an LC3-targeting molecular probe LKT, which can be used to visualize autophagosomes, quantify the level of autophagy, and isolate viable cell subpopulations differing in the autophagy level.In all of these cases, only a short incubation of LKT with cells is required, thus offering a highly convenient methodology.The three motifs contained in LKT work in tandem to accomplish autophagy detection: poly-K promotes cellular entry of LKT, LIR enables specific binding of LKT to LC3, and TPE signals the presence of autophagosomes through AIE-dependent fluorescence enhancement.A nonspecific interaction of LKT with intracellular membranes exists, resulting in some fluorescence increase in cells without elevated autophagy.However, for a given cell type, the extent of this nonspecific LKT fluorescence enhancement is relatively constant, giving a stable background fluorescence reading.Importantly, the specific interaction of LKT with an LC3-conjugated membrane, but not with free LC3 protein, resulted in a dramatic statistically-significant further increase in LKT fluorescence, as demonstrated by both the in vitro LC3-conjugated liposome assay and the cellular assays using HeLa cells treated with various autophagy modulators.This relative and additional increase of LKT fluorescence forms the basis for the detection and separation of cells with elevated autophagy.Notably, DALgreen and DAP-green, two fluorescent molecules for visualizing newly-formed autophagosomes, also exhibit increased fluorescence upon incorporating into autophagosome membranes [23].However, unlike LKT, these molecules do not have an autophagosome-targeting motif, and how they recognize autophagosomes is mostly unknown.
The LIR motif of LKT contains the amino acid sequence DDDWTHL, corresponding to residue 337 to 343 of the SQSTM1 protein.It is known that amino acids D337 and D338 of SQSTM1 interact with the basic residues R10 and R11 in the N-terminal arm of LC3B, while amino acids W340 and L343 bind to hydrophobic pockets in LC3B [45].Conversely, during autophagy the carboxyl terminus of the pro-form of LC3B is cleaved by ATG4 (autophagy related 4 cysteine peptidase) to expose a carboxyl terminal glycine, which then becomes conjugated to phosphatidylethanolamine on autophagosomes [46].Thus, LKT binding and autophagosome conjugation probably occurs at the different ends of LC3, and we can reasonably conclude that LKT-bound LC3 most likely can still conjugate to autophagosomes, and that LKT most likely can still bind to autophagosome-conjugated LC3.In principle, LKT may bind to the cytosolic LC3 first, followed by LC3 conjugation to the emerging phagophores or autophagosomes.In this case, LKT would presumably appear on either the outer or the inner membrane of autophagosomes.Alternatively, LKT may bind to LC3 already conjugated on pre-formed (presealed) autophagosomes, and in this case LKT would, in principle, be able to access and bind to LC3 on the outer, but not the inner, membrane of the autophagosome.Importantly, LKT fluorescence is relatively insensitive to changes in the level of free LC3 protein as demonstrated by the in vitro solution assay, implying that cells may express more or less LC3 protein, but the impact of this variation on LKT fluorescence would be minimal as long as LC3 protein is in the cytosol, as is the case in the absence of autophagy.A change in probe concentration does affect LKT fluorescence significantly, but the magnitude of this fluorescence change is much less than the fluorescence increase elicited by the presence of LC3-conjugated liposomes.This result implies that the level of LKT internalization, which may vary in different cells, would have a relatively small impact on the overall LKT cellular fluorescence in cells with significant autophagy occurrence.These characteristics of LKT ensure robust autophagybased separation and sorting of cells that may exhibit significant difference in either LC3 expression or LKT uptake, or both.
In stark contrast to LKT, the non-AIE probe LKR exhibited similar fluorescence upon binding to either unconjugated or LC3-conjugated liposomes, demonstrating the unique capability of the AIE technology.A critical question is why LKT displayed increased fluorescence when bound to membrane-conjugated but not free LC3.We propose that the hydrophobic TPE, the AIE molecule in LKT, intercalated into the phospholipid of the autophagosome membrane, leading to RIM and enhanced fluorescence emission for TPE, while the high-affinity interaction between the LIR motif and LC3 served to efficiently bring LKT to the proximity of autophagosome membranes.Interestingly, saponin treatment, which caused LC3 aggregation, did not induce LKT fluorescence increase or puncta.Our SPR study clearly showed the ability of LKT to bind to non-lipidated LC3, indicating that LIR-mediated interaction does not require LC3 lipidation.Therefore, we concluded that LKT bound to LC3 aggregates but existed in a non-aggregated state that was unable to turn on AIE fluorescence.This lent further support to our proposal that the interaction of TPE with the lipid bilayer of liposome or autophagosome membranes, rather than with the hydrophobic pockets of the LC3 protein, was responsible for LKT fluorescence enhancement.Additional research is needed to verify this proposition.
Another important issue is whether the binding of LKT would impede the function of LC3 during autophagy.Using various assays, we demonstrated that LKT has minimal effect on autophagosome formation, autophagosomelysosome fusion, and cargo degradation elicited by various autophagy modulators.We offer two possible explanations for these findings.One is that the binding of LKT may not compromise the function of LC3 during autophagy.Conversely, at any given time, perhaps only a percentage of cellular LC3 protein molecules are engaged in LKT binding.Thus, even if the LKT-bound LC3 does become dysfunctional, the rest of unbound LC3 molecules may still be sufficient to fulfill LC3's role in autophagy.
Several other points regarding LKT are also noteworthy.Firstly, LKT measures autophagosomes (and probably also autolysosomes) but not necessarily autophagy flux.In principle, increased autophagy flux (by definition, accelerated autophagy process) may lead to three different outcomes for LKT fluorescence: increased (the accelerated process preceding autophagosome formation is faster than the process after autophagosome formation, resulting in a net increase of autophagosome), decreased (the accelerated process preceding autophagosome formation is slower than the process after autophagosome formation, resulting in a net decrease of autophagosome), and no change (the accelerated process preceding autophagosome formation proceeds at the same rate as the process after autophagosome formation, resulting in no change in the number of autophagosome).In agreement, we observed that Tre and starvation, both of which increased autophagy flux in HeLa cells, showed opposite patterns on LKT fluorescence: an increase in LKT fluorescence after Tre treatment, reflecting more autophagosome formation than autophagosome turnover, and conversely a decrease in LKT fluorescence after starvation treatment, reflecting more autophagosome turnover than autophagosome formation.Secondly, a clear difference exists between the autophagybased cell sorting mediated by LKT and that by GFP-LC3.LKT segregates cells based on LC3 lipidation, which reflects the level of autophagosomes and may result from either autophagy induction or blockade, while GFP-LC3 distinguishes cells differing in autophagic flux, which is increased only under autophagy induction.Moreover, in theory, LKT is capable of detecting autophagosome conjugation for all different members of the LC3 family, while GFP-LC3 only reflects the behavior of LC3B.These unique features of LKT represent a big advantage over GFP-LC3, rendering LKT more broadly applicable to study autophagy aberrations under different physiological or pathological contexts.Thirdly, LKT cannot differentiate between LC3 conjugation to autophagosome and conjugation to other membranes, as both lead to LKT fluorescence enhancement as suggested by the liposome study.Notably, in LC3-associated endocytosis (LANDO) and LC3-associated phagocytosis (LAP), LC3 has been shown to conjugate to endosomes and phagosomes, respectively [47,48], thus LKT fluorescence would presumably be elevated as well in these circumstances, potentially complicating autophagy analyses.Conversely, LKT might facilitate identification and sorting of cells undergoing LANDO or LAP.Further work is warranted to address this possibility.
Finally, it is possible, through molecular engineering, to develop improved versions of the probe.The TPE molecule has a relatively high autofluorescence and emits blue light, both of which are unsatisfactory features for a probe.Replacing TPE with an AIE molecule that has higher excitation and emission wavelengths would likely make significant improvements.A key requirement for our designed probe is to have increased fluorescence upon binding to membraneconjugated LC3 but not free LC3.LKT accomplished this through two different mechanisms, namely LKT proximity to autophagosomes aided by the specific LIR-LC3 interaction and TPE intercalation into autophagosome membrane.With the current LKT probe, these two parts are separate and independent.Thus, LKT can gain proximity to nonautophagosome intracellular membranes (albeit at low efficiency) and facilitate TPE intercalation and fluorescence enhancement, leading to increased fluorescence background.An ideal probe would be one that specifically recognize LC3, turning on fluorescence only when LC3 is associated with autophagosomes.
An interesting and potentially significant finding of this work is the prevalent heterogeneity in the basal autophagy level.Essentially all of the cultured cell lines and primary cells we examined displayed considerable cell-to-cell variation in the level of basal autophagy, indicating that autophagy heterogeneity may be more profound than we have perceived.Our preliminary proof-of-concept study, using LKT to isolate viable and functionally-competent autophagy-high and autophagy-low cell subpopulations in THP-1, a difficult-to-transfect cell line, revealed a significant impact of basal autophagy level on NLRP3 inflammasome activation.Autophagy, by actively removing inflammasome activators and even NLRP3 inflammasome components, mostly plays a suppressive role for NLRP3 inflammasome activation [49].But in contrary, our preliminary results revealed a higher capacity of NLRP3 inflammasome activation following nigericin challenge for the autophagy-high THP-1 cell subpopulation.The fact that the SQSTM1 level for this subpopulation of cells is elevated suggested possible blockade in autophagic degradation, offering a possible explanation.Further research is needed to clarify the role of autophagy for NLRP3 inflammasome activation in THP-1 cells.
In conclusion, LKT fills a void and should prove valuable for autophagy research.

Animal studies
C57BL/6 mice were originally purchased from Jackson Laboratories.All animals received care in compliance with the guidelines outlined in the Guide for the Care and Use of Laboratory Animals and maintained at the SPF animal facility.

Isolation and culture of BMDMs
The mice aged about 8 weeks were sacrificed by cervical dislocation.The hind legs were cut off, the muscles were removed, the leg bones were clamped with forceps, and the joints at both ends were cut off.DMEM medium was absorbed with a syringe, and then inserted into the leg bones to remove the marrow cell, centrifuged at 400 g (centrifuge: CTL600, Cenlee) for 5 min, collected the cell pellet, added 1 mL red cell lysis solution (Biosharp, BL503B) for 2 min, and added 3 mL DMEM medium, centrifuged at 400 g (centrifuge: CTL600, Cenlee) for 5 min.Primary BMDMs were cultured for 4 days in DMEM supplemented with 10% FBS, 25% supernatant of L929 cells (filtered with 0.22-μm membrane filter), and 1% penicillin-streptomycin solution (HyClone, SV30010).

Measurements
UV−vis absorption spectra were taken on a Shimadzu UV-2600 spectrophotometer (Kyoto, Japan).Photoluminescence (PL) spectra were recorded on a Horiba Fluoromax-4 spectrofluorometer (Kyoto, Japan).Fluorescent images were acquired by camera in the dark room, with illumination under an UV lamp (QiWei, China).HPLC analyses were performed by LC3000 binary high pressure liquid chromatograph (TongHeng, China).The electrospray ionization mass was measured on Waters ZQ2000 Mass spectrometer (Waters, USA).Dynamic light scattering analyses were performed using a particle size analyzer (Litesizer 500, Austria).

Synthesis of KT
TPE-yne (8.8 mg, 24.9 μmol) and poly-K peptide (10 mg, 8.3 μmol) were dissolved in 0.9 mL of a DMF and water mixture (v:v = 9:1).Click reaction was initiated by the addition of 0.1 mL aqueous solution containing CuSO4 (1 mg, 4.15 μmol) and sodium ascorbate (1.6 mg, 8.3 μmol).This mixture was under reaction for 24 h at room temperature in the dark.Yield for KT was 43.2% (5.6 mg) after HPLC and freeze-drying under vacuum.MS (ESI) m/z for [M + H] 3+ was calculated as 519.10, and found at 519.10.The stock solution preparation of KT is the same as that of LKT.

LKT photostability assay
Photo-stability of LKT in GFP-LC3 HeLa cells was measured under excitation at 405 nm with 5% laser power for 50 cycles, after incubation of cells with LKT (5 μM) for 6 h.

Spr
The GE BIAcore 8K instrument was operated at a constant temperature of 25°C and the CM5 sensor chip (Cytiva, 29-1049-88) was used in this study.Each CM5 sensor chip consists of 8 identical experimental channels and each channel was divided into two flow-cells.In our experimental setup, flow-cell 1 (Fc1) was always kept blank as a reference, while flow-cell 2 (Fc2) was functionalized with the LC3 for interaction studies with LKT.Specifically, the system was first equilibrated with PBS-T buffer (20 mM Na-phosphate, 150 mM NaCl, and 0.05% Tween 20 [Sangon Biotech, A600560], pH 7.4).The sensor chip was activated for 6 min with a mixture of 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide (EDC, 0.2 M; Cytiva, BR-1000-50) and N-hydroxysuccinimide (NHS, 0.05 M; Cytiva, BR-1000-50).In Fc2, this was followed by a 7-min injection of 40 μM LC3 in 10 mM acetate buffer (pH 5.5), in parallel, PBS-T buffer was injected in Fc1.Finally, 1 M ethanolamine-HCl solution was injected onto both Fc1 and Fc2 to block remained NHS-ester groups.Sensorgrams were monitored online to ensure the successful immobilization of LC3 on Fc2.

Microwell binding assay
Corning 96-well polystyrene high-binding enzyme-linked immunosorbent assay (ELISA) microplates (Corning, 42592) were used for this experiment.Purified LC3 protein in PBS was added to the wells and placed overnight at 4°C.After removing the protein solution, each well was washed with PBS containing BSA (w:v, 0.05%) for 5 times.LKT (10 μM) or KT (10 μM) was added into LC3-coated well and incubated at 37°C for 1 h.After removing unbound probe, fluorescence images were observed and captured by a Nikon Ti-E Microscopy.A quantitative comparison was analyzed by ImageJ software.

Liposome preparation
Liposomes and LC3-conjugated liposomes were ordered from SunLipo NanoTech (Shanghai, China).Carboxyl-terminated liposomes were prepared from egg yolk phosphatidylcholine, cholesterol and carboxyl-terminated 1, 2-distearoyl-sn-glycero -3-phosphoethanolamine-poly(ethylene glycol) (DSPE-PEG) (w:w:w, 4:3:3), while LC3-conjugated liposomes were prepared by conjugation of carboxyl-terminated liposomes with LC3 via a condensation reaction between the carboxyl residue of DSPE-PEG and the amine group of LC3 protein.In brief, the lipid materials were dissolved in 2 mL chloroform, and then dried using a rotary evaporator at 37°C in vacuum for 4 h.which was further dried under nitrogen for 2 h to remove trace chloroform.Subsequently, dried thin film was hydrated with PBS (pH 7.4) and then sonicated intermittently by a probe sonicator at 100 W for 15 min, the liposome dispersion was serially passed through 400, 200, 80 and finally 50 nm pore-sized membranes.For the preparation of LC3 conjugated liposomes, carboxyl-terminated liposomes were preactivated to its succinimide by using EDC and NHS, and then reacted with NH2-LC3 protein.The resulting liposome and LC3 conjugated liposome were respectively sized at ~228.1 nm and ~251.6 nm by dynamic light scattering analyses, with a zeta potential of −8.3 and −6.8 mV, respectively.

Liposome-binding assay
After the incubations in the test tubes for 4 h at room temperature, PL spectra were obtained using a RF-6000 fluorescence spectrophotometer (Shimadzu, Japan), with excitation wavelength at 350 nm.Test-tube images were captured by camera under Tanon® UV-2000.

Confocal microscopy
Cells were seeded in confocal glass bottom dishes with 1 × 10 5 cells per dish and cultured overnight.After different treatments, Cells were washed with PBS (pH 7.4) three times and resuspended in 1 mL of culture media before imaging.Fluorescence images were observed and captured by Nikon Ni-E-A1 Microscopy.The fluorescence of LKT, GFP, RFP and LKR were excited with 405, 488, 561 and 561 nm laser, respectively.Colocalization analyses were performed with Nikon NIS-elements viewer software or Fiji ImageJ software.

Autofluorescence assessment
Autofluorescence of HeLa cells was measured by flow cytometric analyses at the 450/50 nm channel on the UV (355 nm) laser octagon.

Delayed CQ treatment experiment
HeLa cells were labeled with LKT (5 μM) for 6 h, and the culture medium containing LKT was removed.The cells were subsequently treated with PBS or CQ (10 μM) for 6 h, followed by flow cytometric analyses using the 530/30 nm channel.

GFP-LC3 and LKT puncta formation
GFP-LC3 HeLa cells were observed under confocal microscopy after treatment.The fluorescence of LKT and GFP was excited with a 405 and 488 nm laser, respectively.GFP-LC3 dot formation was quantified by counting at least 100 cells and expressed as the ratio between the number of cells with at least five GFP-LC3 dots and the number of cells with green fluorescence (essentially 100% for our cells stably expressing GFP-LC3).LKT dot formation was quantified by counting at least 100 cells and expressed as the percentage of cells with at least five blue dots.

siRNAs and RNA interference
The ATG5, SQSTM1, and TAX1BP1 siRNAs used were purchased from Guangzhou RiboBio Co., Ltd.GFP-LC3 HeLa cells were grown in 24-well plates and transfected with above siRNAs by virtue of Lipofectamine 3000 as the transfection reagent, with reference to the manufacturer's protocol.PBS, and control siRNA were used as negative control.48 h after the transfection, the cells were collected and subjected to western blot analyses for determination of the expression level of targeted proteins.Target sequences (5'-3') ATG5: GTCCATCTAAGGATGCAAT. Target sequences (5'-3') SQSTM1: TGAGGAAGATCGCCTTGGA. Target sequences (5'-3') TAX1BP1: CAGTGATGCTGTCAACGTA.

Mitochondria staining
GFP-LC3 HeLa cells were treated with FCCP (10 μM) for 3 h in the absence or presence of LKT (5 μM) for 24 h.Then, Mitochondria staining was performed using MitoTracker Red CMXRos.Briefly, GFP-LC3 HeLa cells were incubated with MitoTracker Red (100 nM) for 30 min at a cell incubator, then washed with PBS.Finally, the cells were observed under the confocal microscope (LSM 800 with Airyscan).Captured fluorescence images were analyzed by Zen Blue software (v2.5, Zen blue edition, Zeiss).

Cell viability determination
Cells were grown in 96-well plates at a density of about 10,000 cells per well.After designated treatments, thiazolyl blue tetrazolium bromide was added to the growing cultures at a final concentration of 0.5 mg/mL and incubated for another 4 h at 37°C.After dissolving the pellets by DMSO, Biotek Cytation 5 was used to detect the signals at 490 nm.

Flow cytometric analyses
LKT fluorescence intensity was detected in the 450/50 nm channel on the UV (355 nm) laser trigon or 450/40 nm channel on the Violet (405 nm) laser trigon.GFP and LKR were detected in the 530/30 nm channel on the blue (488 nm) laser octagon and the 610/20 nm channel on the yellow green (561 nm) laser trigon, respectively.Flow cytometric data were acquired using BD LSRFortessa flow cytometer (BD Biosciences) and analyzed with the FlowJo software (Tree Star, Ashland, OR, USA).

Flow cytometric cell sorting
Cells were sorted by BD FACSAria SORP flow cytometer (BD Biosciences).LKT fluorescence intensity was detected in the 450/50 nm channel on the UV laser trigon.DMEM was used for sorting HeLa, GFP-LC3 HeLa, MEF, BMDM and MCF-7 cells while RPMI 1640 was used for sorting THP-1.All of the medium used for sorting contained 10% FBS.

LKT sorting of dye-labeled cells
HeLa cells were treated with Tre (0.1 M), RAP (1 μM) or chloroquine (10 μM) for 24 h in the presence of LKT (5 μM).After collection, the cells treated with Tre or RAP were stained with DiO (10 μM), while the CQ-treated cells were stained with DiD (10 μM), for twenty min.Following staining, an equal number of Tre-treated (or RAP-treated) cells and CQ-treated cells were mixed and then subject to flow cytometric analyses.The 450/ 50 nm channel on the UV laser trigon, 530/30 nm channel on the blue (488 nm) laser octagon, and 670/14 nm channel on the red (640 nm) laser trigon were used to quantify the fluorescence of LKT, DiO and DiD, respectively.

Starvation
For nutrient deprivation, HeLa cells were treated with HBSS buffer added after removing DMEM supplemented with 10% FBS and washing with PBS buffer.Immunoblotting analyses of HeLa cells treated with HBSS buffer for 6 h in the presence and absence of LKT (5 μM) for 24 h was used to determine LKT's influence on autophagic protein markers.Conversely, HeLa cells were treated with Tre (0.1 M) for 24 h, or starvation for 6 h in the presence of LKT (5 μM) for 24 h.After collection, the cells treated with PBS were stained with DiO (10 μM), while the Tre or HBSS-treated cells were stained with DiD (10 μM), for 20 min.Following staining, an equal number of PBS-treated cell and Tre-treated cells or starved cells were mixed and then subject to flow cytometric analyses (with reference to the method "LKT sorting of dye-labeled cells").

Western blotting
Cell lysates were loaded and separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis/SDS-PAGE and then the protein samples were electro-transferred to nitrocellulose membranes.After blocking with 5% nonfat milk for 1 h and washing with Tris-buffered saline with Tween 20 (TBST; 0.05 M Tris, pH 7.4-buffered saline [NaCl, 138 mM; KCl, 2.7 mM] and 0.1% Tween 20), the membranes were incubated with different primary antibodies at 4°C overnight.Subsequently, the membranes were incubated with HRPconjugated secondary antibodies at room temperature for 1 h.Finally, the protein bands were displayed with HRP substrate under GE Amersham Imager 300.

Transmission electron microscopy
Cells were harvested and fixed at 4°C overnight in 0.1 M Naphosphate buffer (pH 7.4) containing 2.5% glutaraldehyde.After washing with PBS for three times, cells received post-fixation in 1% OsO 4 at room temperature for 1 h.Then, using graded ethanol, cells were dehydrated and embedded in epoxy resin.Next, the samples were cut by ultramicrotome, stained with uranyl acetate and lead citrate, and observed under Tecnai G2 Spirit Transmission Electron Microscopy (FEI, Czech Republic).To quantify autophagic structures, digital images were acquired and the number of autophagy vacuoles was determined per 7 μm 2 field by ImageJ.

Apoptosis assay
The cells were treated with ANXA5-FITC and propidium iodide (PI), according to the manufacturer's instructions for the annexin V-FITC Apoptosis Detection Kit (C1062L, Beyotime, China), followed by flow cytometer analyses.

IL1B assay
PMA-differentiated THP-1 cells (5 × 10 5 /mL) were incubated with 5 μM LKT for 6 h.25% cell fractions with the highest (F-high) and lowest (F-low) LKT fluorescence were isolated by LKT-based flow cytometric sorting and each cell fractions were plated in 24-well plate and cultured for 3 h, respectively, then the medium was changed to Opti-MEM and primed with 100 ng/mL ultra-pure LPS for 3 h, followed by challenge with 0.5 µM nigericin for 30 min.The supernatants were collected, and the concentrations of IL1B were measured using human IL-1B ELISA kits (R&D Systems), according to the manufacturer's instruction.

LDH release assay
F-high and F-low cells sorted from PMA-differentiated THP-1 cells were primed with LPS and challenged with nigericin as described above.The supernatants were collected and levels of LDH released into the culture medium were analyzed using an LDH Cytotoxicity Assay Kit (Beyotime, C0016) according to the manufacturer's instructions.Biotek Cytation 5 was used to detect the signals at 490 nm.

Statistical analyses
GraphPad Prism 8.4.1 software was used for data analyses.Statistical significance was determined by unpaired two-tailed t-test.Unless otherwise noted, the mean was used to represent central tendency, and error bars represent the standard error of the mean (s.e.m.).Unless otherwise noted, every experiment was done with at least three biologically independent replicates.*P < 0.05, **P < 0.01, ***P < 0.001.

Figure 1 .
Figure 1.Design and characterization of LKT.(A) Schematic presentation of LKT.(B to C) UV-Vis spectra (B) and dynamic light scattering analyses (C) of TPEyne, LKT and LK in a mixture of DMSO and water (v:v, 1:199).(D) Photoluminescence (PL) spectra of TPE-yne and LKT in a mixture of DMSO and water (v:v, 1:199).Inset: Photographs taken under the illumination of a UV lamp.(E) PL spectrum of LKT in DMSO and water (v:v, 1:199) solution and in the solid state (film).Inset: Photographs of LKT in solution and in the film state taken under a UV lamp.(F) Representative SPR sensor gram of the LC3 interaction with LKT (black) and with KT (Red).The X-axis is injection time and the Y-axis is response units.(G) Representative fluorescent images of microplate wells pre-coated with LC3 or BSA and followed by incubation with PBS, LKT or KT overnight.Scale bar: 50 μm.Right panel, quantified results by counting blue dots with ImageJ software.Mean ± s.e.m., n = 4; **p < 0.01.Student's t-test.

Figure 4 .
Figure 4. LKT-enabled sorting of cell subpopulations differing in the autophagy level.(A) LKT-based flow cytometric sorting after mixing equal number of HeLa cells treated with LKT alone and those treated with LKT + CQ for 24 h.F-high and F-low represented 25% cell fraction with the highest and lowest LKT fluorescence, respectively.Concentrations: LKT, 5 μM; CQ, 10 μM.Lower right, LC3 immunoblotting of F-high and F-low cells.(B) LKT-based flow cytometric sorting of HeLa cells after incubation with LKT (5 μM) for 6 h.F-high and F-low represented 25% cell fraction with the highest and lowest LKT fluorescence, respectively.Lower right, LC3 immunoblotting of F-high and F-low cells.(C) Electron microscopic images of F-high and F-low cells from (B).Blue arrows indicated autophagic vacuoles, among which, double membrane structure of autophagosomes was marked by green asterisk.Scale bar: 500 nm.(D) Fluorescent microscopic images of F-high and F-low cells isolated by LKT-based flow cytometric sorting of GFP-LC3 HeLa cells.Scale bar: 10 μm.(E) Western blotting analyses of the sorted F-high and F-low HeLa cells further treated with PBS or CQ (10 μM) for 6 h.(F) Western blotting of F-high and F-low cells, obtained through LKT-based flow cytometric sorting of MCF-7, MEF, undifferentiated THP-1 and BMDM cells.The cells were incubated with LKT (5 μM) for 6 h before sorting.

Figure 5 .
Figure 5. Proof-of-concept study using LKT.(A) LKT-based flow cytometric sorting of PMA-differentiated THP-1 cells after incubation with LKT (5 μM) for 6 h.F-high and F-low represented 25% cell fraction with the highest and lowest LKT fluorescence, respectively.The right panel showed LC3 immunoblotting of F-high and F-low cells.(B) Electron microscopic images of sorted F-high and F-low cells from (A).Blue arrows indicated autophagic vacuoles, among which, double membrane structure of autophagosomes was marked by green asterisk.Scale bar: 500 nm.(C) NLRP3 immunoblotting of F-high and F-low PMA-differentiated THP-1 cells, sorted as in (A) and followed by treatment with PBS or LPS (100 ng/mL) for 3 h.(D and E) IL1B secretion (D) and LDH release (E) in the supernatants of sorted F-high and F-low PMA-differentiated THP-1 cells that were treated with PBS (PBS), LPS-primed for 3 h (LPS), or LPS-primed for 3 h followed by nigericin challenge for 30 min (LPS + NIG).Mean ± s.e.m., n = 3; ***p < 0.001.Student's t-test.(F) Western blotting analyses of SQSTM1 in sorted F-high and F-low PMA-differentiated THP-1 cells.(G) a working model.LC3 lipidation "delivers" LKT to autophagosome membrane, where the TPE motif of LKT intercalates into the lipid bilayer, resulting in RIM and fluorescence enhancement for LKT.LIR, LC3-interacting region; GGG, the linker peptide motif.