Light-activated macromolecular phase separation modulates transcription by reconfiguring chromatin interactions

Design of LAMPS

Optogenetic approaches have recently been leveraged to dissect the biophysical properties of biomolecular condensates by fusing the photolyase homology region (PHR) domain of the Arabidopsis thaliana cryptochrome 2 (CRY2) (19) to various IDR domains, enabling inducible and reversible condensation in living cells (14–18). When fused to dCas9, light-induced condensates targeted to telomeres preferentially formed across low-density genomic regions, mechanically extruding chromatin into spatially segregated territories akin to chromosomal compartments (17). The nonuniform condensation of genomic regions suggests that biomolecular condensation could generate the biophysical force to shape chromatin architecture such as enhancer-promoter looping for gene activation; however, this hypothesis has not been substantiated by the direct measurement of condensation-mediated effects on gene transcription and chromatin configuration (17, 20). Thus, there remains an unmet need for experimental means to simultaneously induce targeted condensation, assess functional consequences of condensate formation, and dissect their biomolecular composition at native chromatin.

In designing LAMPS, we sought to leverage light-induced condensate formation, dCas9 and single guide RNA (sgRNA)–mediated targeting to a single endogenous genomic locus, and affinity purification to characterize nuclear condensates in situ (Fig. 1A). We first devised a multicistronic construct containing an N-terminal IDR, an mCherry fluorescent reporter, a CRY2-PHR fusion protein, a dCas9 flanked by nuclear localization signals (NLS), and a C-terminal biotin acceptor (BioTAP) recognized by endogenous biotin ligases (Fig. 1B) (21, 22). Upon coexpression of sequence-specific sgRNAs and in vivo biotinylation of dCas9-CRY2-mCherry-IDR fusion proteins in living cells, the genomic locus–associated macromolecules are induced to form condensates by blue light illumination and subsequently isolated by biotin-streptavidin–based high-affinity purification. The purified protein-DNA complexes are identified and characterized by proteomics and chromatin conformation capture (3C) (23) for study of condensation-associated proteins and DNA looping, respectively (Fig. 1A) (21, 22, 24). Hence, LAMPS offers the unique capacity for spatiotemporal control of nuclear condensation with parallel analysis of gene transcription and chromatin regulation, enabling the dissection of the molecular links between nuclear condensation and transcriptional function at endogenous genomic loci.

Activation of endogenous gene transcription by LAMPS-induced condensation

The efficiency of light-activated macromolecule clustering is a critical factor in the design of inducible condensation. We first optimized the blue light–responsive CRY2 module in the LAMPS design. Several CRY2 variants have been described with differing light-activated oligomerization propensities, including a Glu-to-Gly substitution at position 490 of CRY2 (CRY2^E490G) (25), a E490R substitution (CRY2^E490R) (26), and a variant modified with a C-terminal 9-mer peptide (CRY2^Clust) (Fig. 1B) (27). We tested constructs containing wild-type (WT) or variant CRY2 fused to mCherry and dCas9 (fig. S1A) and observed their oligomerization properties. Upon transient expression in HeLa cells followed by flow cytometry sorting of cells based on comparable expression of both mCherry and zsGreen from dCas9-CRY2 and sgRNA constructs, respectively (fig. S1, B to D), all dCas9-CRY2 constructs exhibited dim and diffuse signals distributed within the nucleoplasm under the nonilluminated (dark) condition. After light illumination for 5 min at 488 nm, mCherry puncta were readily detected (Fig. 1C). Similar puncta formation was observed in cells with or without nontargeting (sgGal4) or sequence-specific sgRNAs (sgHS2), indicating that sgRNA expression has no effect on dCas9-CRY2–mediated condensation. Moreover, dCas9-CRY2^Clust consistently produced the largest and brightest puncta, indicating the most robust oligomerization upon light activation (Fig. 1C).

We next determined whether LAMPS-mediated condensate formation at dCas9-targeted genomic loci would affect endogenous gene transcription. In proof-of-principle studies, we chose to target the HS2 enhancer of the β-globin locus control region (LCR) and the BCL11A promoter as two testbed cis-regulatory elements (CREs) (Fig. 1B). The HS2 enhancer is required for the transcriptional activation of β-like globin genes (e.g., HBB) (21), whereas BCL11A is a TF that controls developmental hemoglobin switching (28). We reasoned these well-characterized CREs, and the low basal expression of their target genes in HeLa cells would provide ideal testbeds to study gene activation. Last, we considered the incorporation of self-associating IDRs into LAMPS, as these domains are the major determinants of biomolecular condensation through multivalent interactions (1, 20). Two prototypical domains, the N-terminal IDR of the FET (FUS, EWS, and TAF15) family protein FUS (FUS^N) and the C-terminal IDRs of the bromodomain protein BRD4 (BRD4∆N), are well-established to undergo condensation (4, 7). We therefore examined the effect of incorporating IDRs into LAMPS-induced condensates on target gene expression.

We engineered constructs with or without FUS^N or BRD4∆N and measured gene expression before and after light-induced clustering in cells coexpressing dCas9-CRY2 with or without HS2 enhancer– or BCL11A promoter–targeting sgRNAs (sgHS2 and sgBCL11A), respectively (fig. S1, A and B). While light-induced clustering of dCas9-CRY2 variants without IDRs had variable or no effects on HBB and BCL11A expression, incorporation of IDRs to dCas9-CRY2^Clust consistently produced the most notable gene activation despite comparable expression levels of all dCas9-CRY2 variants (Fig. 1D and fig. S1, C and D). LAMPS-induced gene activation was dependent on the coexpression of enhancer- or promoter-targeting sgRNAs, illustrating an dCas9/sgRNA-mediated on-target effect. Moreover, the IDR-fused dCas9-CRY2^Clust consistently displayed larger sizes, higher intensities, and faster kinetics of puncta formation relative to other variants following a single pulse of blue light illumination (fig. S2, A to C). These results demonstrate that light-induced clustering of IDR-containing protein complexes activates endogenous gene transcription at native chromatin. Moreover, CRY2^Clust outperforms other variants tested for optogenetic control of nuclear condensation and was selected for the finalized LAMPS design in subsequent studies.

Biophysical properties of LAMPS-activated nuclear condensates

Having established the role of the IDR-containing dCas9-CRY2^Clust in gene activation, we next assessed the kinetics and biophysical properties of LAMPS-induced condensation by stably expressing IDR-fused dCas9-CRY2^Clust with the CRE-targeting sgRNAs in HeLa cells. Compared to control cells (no IDR) or cells under the dark condition, blue light illumination for 5 min markedly increased HBB and BCL11A mRNA (11- to 23-fold for HBB and 21- to 35-fold for BCL11A, respectively; Fig. 2A), consistent with the results of the initial transient overexpression studies (Fig. 1D). By sampling cells at various time points after a single 5-min pulse of illumination, we noted that HBB or BCL11A mRNA and primary transcripts were progressively and significantly increased over time and peaked at 30 to 45 min after illumination (fig. S3, A and B). Furthermore, we confirmed that LAMPS condensation could activate target gene expression at other loci including the MYOD promoter and IL1RN enhancer upon blue light illumination (fig. S3C), indicating that LAMPS-mediated gene activation is generalizable to multiple genomic loci.

To assess the underlying biophysical properties of LAMPS-induced condensation, we performed a series of imaging studies. We noted progressively increased intensities of LAMPS-induced puncta after blue light illumination with a half-life of puncta formation (t_1/2) of 97.6 ± 18.8 s and 82.8 ± 21.6 s for dCas9-CRY2^Clust-FUS^N and dCas9-CRY2^Clust-BRD4∆N, respectively (Fig. 2B). We next sought evidence to validate whether LAMPS-activated condensates have material properties consistent with membraneless structures in living cells. First, we observed fusion events between distinct puncta upon LAMPS-induced condensation, resulting in their progressive aggregation into larger structures (Fig. 2C), consistent with liquid-like behavior. Second, we performed fluorescence recovery after photobleaching (FRAP) experiments to determine whether LAMPS-induced puncta are dynamic structures that continually exchange constituent molecules with surrounding partners at steady state, a known property of biomolecular condensates (1–4). Puncta formed by dCas9-CRY2^Clust-FUS^N and dCas9-CRY2^Clust-BRD4∆N quickly recovered after localized photobleaching with a half-life of recovery at 7.5 ± 1.1 s and 5.5 ± 3.2 s, respectively (Fig. 2D). Notably, the puncta formed by dCas9-CRY2^Clust without IDR fusion (no IDR) recovered at a slower recovery rate than LAMPS with IDRs (FUS^N or BRD4∆N) after photobleaching (Fig. 2E), indicating that the presence of IDR potentiates condensate dynamics. Together, these results strongly suggest that LAMPS-activated macromolecule clustering exhibits material properties consistent with biomolecular condensates in living cells.

Molecular determinants of LAMPS-mediated nuclear condensation in living cells

Our tool-building experiments revealed that only the combination of the CRY2^Clust variant and IDRs induced puncta formation with notable gene activation (Figs. 1D and 2, A to E), suggesting that IDRs are indispensable for LAMPS-mediated effects on transcription. These findings raised an important question about the molecular determinants of LAMPS-mediated nuclear condensation in living cells. We reasoned that CRY2^Clust oligomerization confers LAMPS with sufficient nucleating energy to seed condensates but alone lacks the biophysical and/or biochemical property conferred by IDRs, which is needed to selectively compartmentalize molecules into condensates for gene activation. Consequently, CRY2^Clust in this model can potentiate the nucleation of IDR-mediated condensates more than the other CRY2 variants to induce IDR-dependent gene activation.

To test this hypothesis, we leveraged two recently established mutants, one impairing CRY2^Clust homo-oligomerization by a Leu-7-Lys (L7K) CRY2^Clust mutation (27) and the other abrogating FUS^N-IDR condensation by 15 Tyr-to-Ser substitutions (29). We first tested the CRY2^Clust L7K mutant (hereafter Clust^mut; Fig. 3A) that alters the hydrophobicity of the C-terminal Clust 9-mer peptide (ARDPPDLDN) required for its enhanced homo-clustering compared to WT CRY2 (27). We observed that FUS^N-fused LAMPS condensates harboring Clust^mut had reduced kinetics of puncta formation compared to those with native Clust upon blue light illumination (Fig. 3B). Clust^mut-induced puncta were also less dynamic, exhibiting slower recovery kinetics after localized photobleaching with a half-life of recovery 10.7 ± 2.7 s compared to 4.6 ± 0.9 s for native Clust (Fig. 3C). These findings suggest that the CRY2 moiety primarily seeds condensate formation, as impairing Clust-driven clustering propensity by point mutation while retaining FUS^N may still endow nascent Clust^mut condensates with an IDR-dependent composition that promotes gene activation, although with reduced efficiency due to impaired CRY2 seeding. To test the role of IDRs in condensation dynamics, we incorporated the FUS^N IDR mutant, which contains 15 Tyr-to-Ser substitutions in a critical tract of FUS^N IDR known to impair condensation (hereafter FUS^Nmut; Fig. 3A) (29). As expected, FUS^Nmut fused to CRY2^Clust both impaired light-induced condensate formation and reduced its recovery rate after photobleaching to a greater degree than Clust^mut, whereas double mutation of Clust and FUS^N (hereafter FUS^Nmut-Clust^mut) impaired light-activated condensation to levels comparable to native Clust without IDRs (Figs. 2E and 3, B and C). The dominant effect of FUS^N mutation on condensate dynamics suggests that IDRs may confer permeability to condensates such that constituent molecules are dynamically incorporated.

To test whether and how the modulation of LAMPS condensation by altering clustering efficiency or IDR sequence affects target gene transcription, we measured gene expression in HeLa cells coexpressing WT or mutant LAMPS with HS2 enhancer-targeting sgRNAs by transient transfection and in stable cell lines. Notably, Clust^mut alone reduced the target gene (HBB) expression to 68.9 and 65.8% of the levels induced by native CRY2^Clust in both transient and stable expression cells (Fig. 3, D and E). FUS^Nmut alone reduced HBB expression to 26.9 and 21.1% of the levels induced by WT FUS^N, whereas the FUS^Nmut-Clust^mut double mutant abolished gene activation (Fig. 3, D and E), suggesting a dominant role of IDRs in LAMPS-mediated gene activation. Together, these findings not only provide insight into distinct functions of LAMPS components (CRY2^Clust versus IDR) in controlling condensation-mediated gene activation but also showcase how the modularity of LAMPS components could be broadly leveraged to fine-tune gene activity and dissect the biophysical basis of condensation-dependent gene regulation using programmable condensates.

Targeting of LAMPS-induced condensation to native genomic loci

Transcription-dependent biomolecular condensation emerges as a key feature of gene regulation (9–12, 30, 31), yet the precise mechanisms by which these condensates modulate gene transcription at native chromatin remain elusive. A unique feature of LAMPS is the ability to capture de novo–formed condensates in situ for biochemical and mechanistic studies (Fig. 1A). This is achieved by streptavidin-based affinity purification of biotinylated dCas9-tethered protein-DNA complexes at endogenous genomic loci (21, 22).

To validate the LAMPS approach for identifying the macromolecular composition of locus-specific nuclear condensates in living cells, we first sought evidence for on-target association of LAMPS complexes at the sgRNA-targeted HS2 enhancer. Upon stable expression of dCas9-CRY2^Clust-FUS^N or dCas9-CRY2^Clust-BRD4∆N with or without sgHS2 in HeLa cells, the dCas9-tethered chromatin was cross-linked, fragmented, and affinity-purified using streptavidin magnetic beads, followed by analysis of the captured DNA using the LAMPS–chromatin immunoprecipitation (ChIP) assay (Fig. 4A) (21, 22, 24). We observed significant dCas9 enrichment at the HS2 locus (852- to 1687-fold relative to nontargeting sgRNA control; Fig. 4B), indicating efficient recruitment of LAMPS complexes to native chromatin. Moreover, on-target dCas9 signal further increased upon light induction likely because of LAMPS-mediated clustering. To evaluate specificity, we performed deep sequencing of dCas9-bound genomic DNA. Notably, the recruitment of dCas9 by sgHS2 resulted in enrichment of HS2 with few nonspecific dCas9 binding regions, including predicted off-targets, at genome scale (Fig. 4C and fig. S4, A and B). These chromatin assays provide evidence that LAMPS complexes are highly enriched at the dCas9/sgRNA-targeted genomic loci.

Altered chromatin accessibility and DNA looping by LAMPS-activated condensation

We next sought to address how LAMPS-induced condensation mechanistically activates gene transcription. We hypothesized that de novo condensation creates a local chromatin environment permissive for recruitment of TFs and coactivators through IDR-mediated multivalent interactions. To test this, we performed assay for transposase-accessible chromatin using sequencing (ATAC-seq) to assess chromatin accessibility at LAMPS-targeted HS2 before and after blue light illumination. Despite strong enrichment of sgRNA-mediated dCas9 binding at HS2 (Fig. 4, B and C), no detectable ATAC-seq signal was observed before illumination (fig. S5A), suggesting that dCas9 binding alone has little effect on accessibility. By contrast, light-activated condensation increased chromatin accessibility at HS2 for both dCas9-CRY2^Clust-FUS^N and dCas9-CRY2^Clust-BRD4∆N (Fig. 4, D and E, and fig. S5A). Notably, cells expressing the mutant FUS^Nmut also displayed increased chromatin accessibility upon blue light illumination (Fig. 4E), suggesting that LAMPS clustering is sufficient to modulate chromatin accessibility at the target loci but the IDR-mediated composition of condensates is required for subsequent gene activation. These results support the model by which LAMPS-induced nuclear condensation increases chromatin accessibility to facilitate the recruitment of TFs, enabling the assembly of transcriptional complexes for gene activation.

Chromatin organization maintained by long-range DNA interactions such as enhancer-promoter looping is critical for transcriptional regulation (32). Given the unique feature of LAMPS for unbiased analysis of locus-regulating chromatin interactions (21, 22), we examined HS2-associated chromatin looping before and after light-activated condensation. Upon coexpression of dCas9-CRY2^Clust-FUS^N or dCas9-CRY2^Clust-BRD4∆N with sgHS2, long-range DNA contacts associated with dCas9-targeted HS2 were cross-linked, followed by Dpn II digestion and proximity ligation of interacting DNA fragments. Locus-specific interactions were captured by streptavidin-based purification of biotinylated dCas9 and analyzed by paired-end sequencing using the LAMPS–3C sequencing (3C-seq) assay (Fig. 4D).

Using this approach, we first noted sparse and spurious interactions at HS2 before illumination, consistent with low basal enhancer activity in HeLa cells (Fig. 4E and fig. S5B). By contrast, we detected increased numbers and distances of long-range DNA interactions for both dCas9-CRY2^Clust-FUS^N and dCas9-CRY2^Clust-BRD4∆N after blue light illumination. The identification of DNA loops between the HS2 enhancer, HBB and HBD promoters, and HBBP1 intergenic regions is consistent with their roles in β-globin transcription (21, 33, 34). We observed significant up-regulation of the HBB, HBD, and HBBP1 genes, which were associated with more consistent enhancer-promoter loop formation, but variable or no change in the expression of the more distal and silent olfactory genes (Fig. 4F). We performed parallel experiments with FUS^Nmut LAMPS and observed reduced long-range chromatin interactions, including known enhancer-promoter loops (Fig. 4E and fig. S5B), consistent with a critical role for IDRs in LAMPS-mediated condensation and gene activation (Fig. 3). Together, these results demonstrate that LAMPS-mediated condensation induces de novo formation of chromatin loops, providing direct evidence for the emergent model that the condensation of macromolecules may represent the biophysical force to structure the three-dimensional (3D) genome (17).

Notably, LAMPS-activated HeLa cells contained fewer long-range interactions compared to HS2-associated interactions in embryonic/fetal-stage human umbilical cord blood-derived erythroid progenitor cells (HUDEP1) or adult-stage (HUDEP2) erythroblasts that endogenously express β-globin genes (Fig. 4E) (35, 36). Hence, while LAMPS-mediated condensation is sufficient to initiate chromatin looping for gene activation, the stabilization and/or propagation of these loops likely require additional mechanisms such as the action of lineage-specific TFs, the recruitment of specific chromatin regulators, and/or the epigenetic state of local chromatin.

Reconfiguration of proteomic composition for gene activation by LAMPS-activated condensation

Biomolecular condensation is recognized as a fundamental mechanism underlying diverse intracellular structures, yet little is known of their regulatory composition under physiological conditions where their functional consequences have been established. Having demonstrated the biophysical properties of LAMPS-activated nuclear condensates and the determinants for their effects on gene transcription and chromatin configuration, we next sought to elucidate the molecular composition and dynamics of light-induced condensates in living cells.

To this end, we leveraged the biotinylated dCas9 moiety and performed proteomic analysis of purified LAMPS-induced condensates (Fig. 1A). Specifically, HeLa cells stably expressing dCas9-CRY2^Clust-FUS^N (or with FUS^Nmut) and sgHS2 were illuminated with blue light and immediately cross-linked to stabilize condensate-chromatin interactions, while nonilluminated cells were processed in parallel as a negative control. Stringent urea washes were performed to remove cytoplasmic and nuclear soluble fractions and to enrich chromatin-bound complexes; this step is also intended to deplete nonchromatin bound condensates nucleated by CRY2 clustering to further enrich for on-target condensates (21). Condensate-associated proteins cross-linked to chromatin were then isolated and copurified with dCas9 using streptavidin-conjugated magnetic beads, followed by on-bead trypsin digestion and mass spectrometry-based proteomic profiling (Fig. 5A).

We identified 258 and 184 significantly enriched or depleted nuclear proteins, respectively, in light-activated condensates relative to nonilluminated controls (log₂ fold change ≥ 1 or ≤ −1, P < 0.01 from three replicate experiments; table S1). Notably, RNA polymerase II (Pol II) subunits (POLR2B, POLR2G, and POLR2I), general TFs (TAF1, TAF2, and GTF2F1), transcriptional coactivators (EP300, PAF1, and CCNT1), mediators (MED1, MED23, and MED24), and cohesin complexes (RAD21, SMC1A, SMC1B, and SMC3) were among the top enriched proteins. By contrast, epigenetic corepressors including human silencing hub (HUSH) subunits PPHLN1, FAM208A, and MPHOSPH8 (37, 38), RNA splicing factors, and nuclear lamins (LMNA and LMNB1) were the top depleted proteins in LAMPS-induced condensates (Fig. 5B). To identify candidate cellular pathways relevant to LAMPS-induced condensation, we performed gene ontology and pathway analyses of the enriched or depleted proteins. Chromatin organization, regulation of gene transcription, and nucleic acid binding were among the top enriched biological processes or molecular functions, whereas RNA splicing, DNA packaging, and nuclear organization were associated with the depleted proteins (Fig. 5C and table S2). The enrichment and depletion of distinct protein complexes supports the notion that nuclear condensation provides a mechanism to selectively partition proteins into compartmentalized, higher-order assemblies to regulate gene transcription (39). To validate candidate condensate-associated proteins identified from proteomics, we examined the chromatin occupancy of RNA Pol II (RPB1) and cohesin subunits (RAD21 and SMC1). We detected enrichment of Pol II and cohesin after light illumination at HS2 but not at nontargeted genomic loci (fig. S6A), consistent with their recruitment to LAMPS-targeted condensates.

By parallel analysis of the proteomic composition of LAMPS-activated condensates with the mutant (FUS^Nmut) IDR, we observed a nearly complete loss of enriched coactivators and chromatin structure factors identified from native IDR-mediated condensates (Fig. 5, B and D, and fig. S6B). IDR mutation also abrogated the selective depletion of nuclear lamina proteins and corepressors associated with the native IDR. These results, together with the analyses of condensate properties (Fig. 3, A to C), gene transcription (Fig. 3, D and E), and long-range chromatin looping (Fig. 4E), provide strong support for a dominant role of IDRs in mediating the selective compartmentalization of nuclear factors into LAMPS-seeded condensates for gene activation.

Together, by profiling the proteome composition and 3D conformation of light-activated nuclear condensates in living cells, we provide evidence that inducible condensate formation at an endogenous enhancer element facilitates the selective recruitment of TFs, chromatin regulators, and RNA binding proteins, with concomitant depletion of repressor complexes, to induce chromatin remodeling and gene activation (Fig. 6). Therefore, by engineering LAMPS-activated condensation in living cells, our findings support a model whereby targeted formation of macromolecular condensates induces reconfiguration of chromatin architecture and accessibility to modulate gene transcription.

Cell culture

HeLa cells were obtained from the American Type Culture Collection (ATCC, catalog no. CCL-2) and validated by short tandem repeat profiling analysis by the vendor and were free of mycoplasma contamination. Cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% fetal bovine serum (FBS) and 1% penicillin/streptomycin. For lentivirus production, human embryonic kidney (HEK) 293T cells were cultured in DMEM with 1× GlutaMAX and 1× sodium pyruvate at 37°C with 5% CO₂.

Cloning of LAMPS vectors

sgRNA cloning

sgRNAs for site-specific targeting of genomic regions were designed to minimize off-target cleavage based on publicly available filtering tools (http://crispor.tefor.net/). To minimize potential interference between the chromatin binding of dCas9 and trans-acting factors, sgRNAs were designed to target the proximity of CREs as previously described (21, 22). Two sgRNAs were used per target site. The oligonucleotides for sgHS2 or sgBCL11A (table S3) were phosphorylated and annealed in the following reaction: 10 μM guide sequence oligo, 10 μM reverse complement oligo, T4 ligation buffer (1×), and 5 U of T4 polynucleotide kinase (NEB) with the cycling parameters of 37°C for 30 min followed by 95°C for 5 min and then ramp down to 25°C at 5°C/min. The annealed oligos were cloned into the sgRNA vectors using a Golden Gate Assembly strategy including the following: 100 ng of circular sgRNA vector plasmid, 0.2 μM annealed oligos, NEBuffer 2.1 (1×), 20 U of Bbs I restriction enzyme (NEB), 0.2 mM adenosine triphosphate, bovine serum albumin (0.1 mg/ml), and 750 U of T4 DNA ligase (NEB) with the cycling parameters of 20 cycles at 37°C for 5 min and 20°C for 5 min, followed by 80°C for 20 min.

Transient transfection

HeLa cells were seeded into six-well or 35-mm plates at 50% confluent cell density overnight and then transfected with LAMPS vectors with or without sgRNAs (two independent sgRNAs per target mixed at 1:1 ratio) using Lipofectamine 3000 following the manufacturer’s protocol (Invitrogen, L3000075). Briefly, solution A was prepared with 125 μl of Opti-MEM I reduced serum medium (Gibco) and 3.75 μl of Lipofectamine 3000. Solution B was prepared with 125 μl of Opti-MEM, 1.5 μg of LAMPS and 1 μg of sgRNA vector DNA (2.5 μg of DNA total), and 5 μl of P3000 enhancer. Solutions A and B were vortexed briefly, pooled, and incubated for 5 to 10 min at room temperature. The mixture was added to the cells dropwise. Cells were harvested for imaging or gene expression analyses 24 to 48 hours posttransfection.

Lentiviral transduction and generation of stable cell lines

Lentiviruses containing sgRNAs (mixed at 1:1 ratio for two independent sgRNAs) or LAMPS constructs were packaged in HEK293T cells using Lipofectamine 3000 with the following modifications: 6.5 μg of psPAX2 (Addgene, #12260), 3.5 μg of VSV-G (Addgene, #8454), and 10 μg of pLenti expression vector (dCas9 or sgRNAs) were cotransfected using Lipofectamine 3000 into HEK293T cells on a 10-cm petri dish with DMEM at 37°C with 5% CO₂. After incubation for 6 hours, transfected DMEM was changed with 6 ml of prewarmed lentivirus packaging media (Opti-MEM I reduced serum medium with GlutaMAX, supplemental with final concentrations of 5% FBS and 0.2 mM sodium pyruvate). Lentiviral particles were then harvested from the supernatant at 48 and 72 hours posttransfection, pooled, and centrifuged briefly to clear cell debris. The viral supernatant was precipitated using the PEG-it Virus Precipitation Solution (Systems Biosciences) following the manufacturer’s instructions. To transduced cells, concentrated viral suspensions were added dropwise to target cells at ≥95% confluent cell density in growth media containing polybrene (10 μg/ml), incubated for 24 hours before changing to fresh media, and then recovered for 2 to 3 days at 37°C followed by fluorescence-activated cell sorting (FACS) for high dCas9/mCherry-expressing cells. Sorted cells were expanded to be transduced with sgRNA-expressing lentiviruses. To obtain cells with a high sgRNA expression, the top 5% of zsGreen-positive cells were FACS-sorted 48 hours posttransfection and selected with puromycin (1 μg/μl). The sequences for sgRNAs are listed in table S3.

Gene expression analysis

Fluorescence imaging

HeLa cells were plated on glass-bottom dishes (Cellvis, catalog no. D35-20-0-TOP) and transfected with LAMPS constructs using Lipofectamine 3000. Fluorescence imaging was performed 24 hours posttransfection on a Nikon Ti2 inverted epifluorescence microscope with a Yokogawa W1 dual spinning disk scan head equipped with six solid-state lasers (405, 445, 488, 514, and 640 nm) for imaging and one additional laser for stimulation (473 nm). Fluorescence images were captured using a Photometrics Prime BSI and back-side illuminated scientific complementary metal-oxide-semiconductor (sCMOS) camera. For time-lapse imaging, a 60× oil lens was used for high-resolution images and conducted with an Okolab Stage Top incubator, which provides incubation with full environmental control including heating, humidity, and CO₂ regulation. Blue light illumination was performed with the built-in 488-nm laser source (5% input). The fluorescence intensities of light-induced puncta were calculated by the NIS-Elements AR microscope imaging software (Nikon, NIS-element AR version 4.0). To quantify the area of puncta, puncta were selected as the region of interest (ROI); the area of ROIs was then analyzed by the NIS-Elements software. The changes in puncta intensity (F − F₀) were calculated by measuring the local intensity changes of puncta, where F is the fluorescence at time t and F₀ is the fluorescence before blue light illumination.

LAMPS–ChIP sequencing

LAMPS–3C-seq

LAMPS-proteomics

Assay for transposase-accessible chromatin using sequencing

A total of 5 × 10⁴ cells were washed twice in 1× PBS and resuspended in 500 μl of cell lysis buffer [10 mM tris-HCl (pH 7.4), 10 mM NaCl, 3 mM MgCl₂, and 0.1% NP-40]. Cell lysates were centrifuged at 500g for 10 min at 4°C. Isolated nuclei were resuspended in 50 μl of tagmentation mix [10 mM TAPS {[tris(hydroxymethyl)methylamino]propanesulfonic acid} (pH 8.0) (Sigma-Aldrich), 5 mM MgCl₂, and 2.5 μl of Tn5 transposase] and incubated at 37°C for 30 min. Tagmentation reaction was quenched by adding 10 μl of 0.2% SDS followed by incubation at room temperature for 2 min and 55°C for 7 min. Tn5 transposase–tagged DNA was purified using the QIAquick MinElute PCR Purification kit (QIAGEN), amplified using the KAPA HiFi Hotstart PCR Kit (KAPA), and sequenced on an Illumina NextSeq 500 system. ATAC-seq raw reads were trimmed to remove the adaptor sequence using Cutadapt (55) and aligned to hg38 genome assembly using Bowtie1 (60) with the parameters --best --strata -k 1 -m 1. Only tags that uniquely mapped to the genome were used for analysis.

Statistics and reproducibility

Statistical details including N, mean, and statistical significance values are indicated in the text and figure legends. Error bars in the experiments represent SEM or SD from either independent experiments or independent samples. All statistical analyses were performed using GraphPad Prism, and the detailed information about statistical methods is specified in figure legends or Materials and Methods. The numbers of independent experiments or biological replicate samples and P values (*P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001; n.s., not significant) are provided in individual figures. P < 0.05 was considered statistically significant.

This PDF file includes:

Other Supplementary Material for this manuscript includes the following: