BLAST: A blue light-assisted secretion toolkit tunable by reversible protein-protein interactions

11 12 Precise control over protein secretion is essential for programming intercellular communication and coordinating 13 complex physiological responses. However, conventional methods relying on transcriptional regulation or 14 chemical induction often lack the spatiotemporal precision and reversibility required to mimic endogenous 15 signaling dynamics. Here, we present the Blue Light-Assisted Secretion Toolkit (BLAST), a genetically encoded 16 system that orchestrates protein release from the endoplasmic reticulum via light -tunable protein -protein 17 interactions. BLAST comprises two complementary modules utilizing both light -induced iLID/SspB association 18 (a-BLAST) and LOV2/Zdk1 dissociation (d-BLAST). Both modules harness the highly conserved RXR motif to 19 enforce strict ER confinement in the dark state. Most importantly, by utilizing non -destructive steric masking 20 rather than enzymatic cleavage, BLAST achieves unprecedented temporal resolution with strict reversibility. We 21 demonstrate that both systems can be repeatedly toggled ON and OFF, instantaneously arresting cargo release 22 upon light withdrawal to generate highly controlled, pulsatile secretion profiles. Leveraging this dynamic control, 23 we successfully achieved the rapid, robust, and light-triggered secretion of complex therapeutic proteins, including 24 insulin and interleukin-12. By bypassing transcriptional delays and irreversible activation steps, BLAST provides 25 a generalized, plug-and-play platform for the on-demand delivery of therapeutic proteins, significantly expanding 26 the optogenetic toolbox for synthetic biology and cell-based therapies. 27 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint 1. Introduction 28 29 Secreted proteins serve as the primary currency of intercellular communication, orchestrating diverse 30 physiological processes ranging from immune responses to synaptic transmission [1-3]. The dysregulation of these 31 secretory pathways is often linked to severe pathological conditions[4-6], underscoring the need for precise 32 therapeutic interventions. However, natural secretion is a highly dynamic event, often occurring in pulsatile or 33 localized bursts [7-9]. Recapitulating these complex profiles for therapeutic and synthetic biology applications 34 requires tools that allow for the precise spatiotemporal control of protein release, surpassing the capabilities of 35 constitutive secretion. 36 Current strategies for controlling protein secretion predominantly rely on transcriptional regulation [10-17]. While 37 effective for long-term expression, these methods are inherently limited by the slow kinetics of transcription and 38 translation, requiring at least 4 –6 h to reach therapeutic levels [18]. To achieve faster kinetics, post -translational 39 retention systems have been developed. Early approaches, such as the chemically inducible RUSH (retention 40 using selective hooks)[19] or protease-dependent strategies like RELEASE (Retained Endoplasmic Cleavable 41 Secretion)[20], membER/lumER[21], and POSH (protease -mediated post -translational switch) [22] significantly 42 improved response times compared to transcriptional control. However, these methods often depend on exogenous 43 ligands that are difficult to wash out and lack the dynamic reversibility required for strict ON/OFF control. 44 Subsequently, the development of optogenetic tools, such as optoPOSH[22] and optoPASS[23], introduced a new 45 level of spatiotemporal resolution. By employing pMag and nMag dimerization modules to reconstitute a split 46 protease upon illumination [24,25], optoPOSH and optoPASS successfully triggers protein release. However, 47 because this system fundamentally relies on irreversible proteolytic cleavage, it functions as a single-activation 48 switch, preventing the reversible toggling of secretion required to mimic dynamic physiological signals. 49 To overcome these limitations, we developed the Blue Light -Assisted Secretion Toolkit (BLAST), a genetically 50 encoded platform that exerts rapid and reversible control over protein secretion. Distinct from previous cleavage-51 based methods, BLAST strategically harnesses the RXR (Arginine-X-Arginine) motif—a stringent sorting signal 52 utilized by native channels —to enforce strict spatial confinement within the ER [26,27]. By coupling this robust 53 RXR-mediated retention with light -tunable protein-protein interactions (PPIs), BLAST employs non-destructive 54 steric masking rather than enzymatic cleavage. This mechanism not only bypasses the central dogma lag to enable 55 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint the immediate release of pre -synthesized cargo, but crucially allows for the instantaneous cessation of secretion 56 upon light withdrawal. We engineered two complementary systems: an associative module (a -BLAST) based on 57 Light-Inducible Dimer (iLID)/stringent starvation protein B (SspB) dimerization[28,29] and a dissociative module 58 (d-BLAST) based on Light-Oxygen-Voltage sensing domain 2 (LOV2)/Zdk1 dissociation [30]. Here, we 59 demonstrate that BLAST provides a robust, plug -and-play solution for the spatiotemporally precise delivery of 60 various cargo proteins, offering a generalized method to program intercellular signaling with blue light. 61 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint 2 Results 62 63 2.1 Design and characterization of the associative BLAST system 64 To engineer a precision tool for light -regulated protein secretion, we de signed the associative Blue Light -65 Assisted Secretion Toolkit (a -BLAST) (Fig. 1a). This system exploits the 470 nm blue light -dependent 66 heterodimerization between the iLID and its binding partner, SspB [28,29]. The a -BLAST system comprises two 67 modular components: an ER -anchored cargo module and a cytosolic regulator module. The prototype cargo 68 module is designed with an N -terminal protein of interest (POI), followed by a Furin cleavage site, a 69 transmembrane domain (TM), an SspB sequence, and a C-terminal RXR ER-retention motif[27]. In the dark state, 70 the cargo is strictly confined within the ER due to the exposed RXR motif, which is recognized by the COPI 71 retrieval machinery[31]. Upon blue light illumination, the cytosolic iLID regulator binds to the SspB moiety on the 72 cargo. We hypothesized that th is light-induced association would sterically mask the RXR motif [27,32], non-73 destructively overriding the retention signal and permitting the cargo's exit to the trans -Golgi network. 74 Subsequently, endogenous Furin proteases within the Golgi cleave the processing site, resulting in the secretion 75 of the mature POI. 76 To optimize the dynamic range of this prototype, we utilized Secreted Embryonic Alkaline Phosphatase (SEAP) 77 as a reporter system. We compared the wild -type iLID with three mutants (N414L, V416L, and N414L/V416L) 78 known to exhibit distinct binding affinities (Fig.1b, Supplementary Fig. 1a) [29]. Consistent with previous 79 characterizations of iLID variants[29], the double mutant (N414L/V416L) exhibited the highest dynamic range and 80 was selected for all subsequent optimizations (Fig.1c, Supplementary Fig. 1c). 81 Next, to maximize the signal -to-noise ratio, we focused on optimizing the C -terminal topology of the retention 82 module. Since RXR-mediated retention is sensitive to its spatial distance from the membrane and the surrounding 83 steric context [31], we hypothesized that altering the spacer length via SspB concatemerization would critically 84 influence retention efficiency. We generated a library of cargo variants containing 1 × to 9× tandem repeats of 85 SspB (Fig. 1 d). Through a comprehensive screen of SspB valencies and plasmid transfection ratios (Cargo: 86 Regulator), we evaluated the secretory performance of each configuration (Fig. 1e, f). Although the 9 × SspB 87 variant at a 1:3 ratio exhibited the highest apparent fold-change (31.5-fold), its absolute maximum secretion level 88 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint under blue light was relatively diminished. This reduction is likely attributable to the structural burden and 89 reduced translation or folding efficiency imposed by the excessively large tandem repeats. Therefore, we selected 90 the 6× SspB variant coupled with a 1:3 transfection ratio as the optimal configuration . This setup provided the 91 ideal balance between stringent dark-state retention and a robust, high-capacity protein output, achieving a 16.3-92 fold increase in SEAP secretion (Fig. 1 c). These results demonstrate that precisely tuning both the steric 93 environment of the retention motif and the stoichiometry of the system components is critical for achieving 94 accurate secretory control. 95 96 2.2 Benchmarking the reversible retention mechanism against proteolytic release 97 To rigorously benchmark the efficiency of the reversible steric masking strategy of a-BLAST against established 98 irreversible proteolytic cleavage systems, we compared BLAST with the RELEASE system[20]. RELEASE utilizes 99 a split protease to cleave an ER-retention signal (KKMP), a mechanism that typically requires close proximity to 100 the ER membrane for efficient proteolytic processing[31]. To ensure a direct , fair mechanistic comparison 101 independent of the stimulus modality (light v ersus drug), we engineered a chemogenetic analog of a -BLAST 102 (chemo-BLAST) by substituting the iLID/SspB optogenetic pair with the rapamycin -inducible FRB /FKBP 103 dimerization system (Supplementary Fig. 2a)[33]. 104 We first optimized the chemo -BLAST architecture by varying the valency of the FRB -fused retention module 105 (1×, 2×, 4×, and 6×). In contrast to the optogenetic a-BLAST (which preferred a 6× configuration), the chemo-106 BLAST variant exhibited the highest rapamycin -dependent dynamic range with 4× FRB repeats (Supplementary 107 Fig. 2b). Subsequent optimization of the transfection ratio identified that a 1:3 (Cargo: Regulator) ratio yielded a 108 maximal 11.3-fold induction (Supplementary Fig. 2c). Next, we performed a head -to-head comparison between 109 the optimized chemo -BLAST and the RELEASE system (employing both TEVp and TVMVp split proteases). 110 The results unequivocally demonstrated that chemo-BLAST significantly outperformed the RELEASE system in 111 terms of overall secretion efficiency (Supplementary Fig. 2d). Notably, chemo -BLAST exhibited markedly 112 superior kinetics, inducing significant protein secretion within just 4 h of rapamycin treatment, whereas the 113 RELEASE system showed a pronounced temporal delay. These findings strongly suggest that the non-destructive 114 steric masking mechanism of BLAST allows for immediate cargo exit upon inducer binding, effectively bypassing 115 the kinetic bottleneck associated with the proteolytic cleavage step required by systems like RELEASE. 116 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint 117 2.3 Design and characterization of the dissociative BLAST system 118 To provide a complementary mode of regulation characterized by rapid response kinetics, we developed the 119 dissociative BLAST (d -BLAST) system, inspired by the Light-Oxygen-Voltage sensing domain 2 Trap and 120 Release of Protein (LOVTRAP) optogenetic tool (Fig. 2a) [30]. Unlike the associative a-BLAST, this system 121 operates via light -induced dissociation. The d -BLAST architecture consists of a n ER-anchored cargo module 122 (POI-Furin-TM-mCherry-LOV2) and a cytosolic regulator module (Zdk1-RXR) (Fig. 2a). In the dark, the Zdk1 123 domain of the regulator binds tightly to the LOV2 domain on the cargo. This interaction transiently recruits the 124 RXR motif to the cargo, effectively hijacking the COPI retrieval machinery to strictly retain the protein within 125 the ER. Upon blue light irradiation, the Jα helix of LOV2 unfolds, causing the rapid dissociation of the Zdk1 -126 RXR complex. This instantaneous removal of the retention signal permits the cargo to escape the ER and proceed 127 through the secretory pathway. 128 Crucially, because the RXR -mediated retention mechanism requires a specific spatial distance from the 129 membrane to function effectively, we incorporated mCherry as a spacer to optimally position the RXR motif away 130 from the ER membrane [31]. Validating this design, constructs lacking the mCherry spacer exhibited severe dark-131 state leakage (Supplementary Fig. 3), confirming the spacer ’s structural necessity for strict spatial retention. 132 Having established the optimal cargo architecture containing the mCherry spacer, we next optimized the 133 photocycle kinetics to enable efficient secretion while minimizing phototoxicity . We screened LOV2 mutants 134 (V416T, V416I, V416L) with varying dark reversion rates, hypothesizing that slow -reverting variants would 135 maintain the dissociated (ON) state longer under pulsed illumination. Using a 12 h stimulation protocol (2 s ON/58 136 s OFF) (Fig. 2b), we confirmed that the V416L mutant, which possesses the slowest reversion kinetics, yielded 137 the highest fold-change in SEAP secretion (Fig. 2c). 138 However, a persistent challenge with dissociative systems is basal leakage in the dark state due to the incomplete 139 capture of the cargo by cytosolic regulators. To address this, we engineered a membrane -tethered regulator by 140 fusing the Zdk1 -RXR module to the N -terminal signal anchor of cytochrome P450 (P450n) [34,35] (Fig. 2d). We 141 reasoned that immobilizing the regulator on the ER membrane would drastically increase its local effective 142 concentration around the cargo, thereby enforcing stricter retention in the dark. Consistent with our design 143 rationale, the P450-anchored regulator substantially minimized dark-state basal leakage compared to the cytosolic 144 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint version, boosting the optogenetic dynamic range from 7.1 -fold to an impressive 20.6-fold (Fig. 2e). To further 145 maximize protein yield, we optimized the production timeline. Extending the protocol from a 3-day schedule (12 146 h expression/12 h light) to a 4-day schedule (24 h expression/24 h light) resulted in a remarkable 79.1-fold increase 147 in cumulative protein secretion (Fig. 2f). 148 149 2.4 Kinetic profiling and benchmarking of BLAST against proteolytic switches 150 To precisely define the temporal resolution of our toolkit, we profiled the secretion kinetics of both a -BLAST 151 and d-BLAST. HEK293T cells expressing optimized constructs were subjected to blue light or dark conditions 152 for varying durations (0 to 24 h) starting 24 h post -transfection (Fig. 3a, b). Strikingly, d-BLAST induced 153 statistically significant SEAP secretion within just 1 h (3.5-fold), whereas a-BLAST required 4 h of illumination 154 to achieve significant release (5.2-fold) (Fig. 3c). Comparing the two, d-BLAST exhibited superior rapid-response 155 kinetics, achieving a 2.2-fold increase in secretion after only 30 m of illumination. However, this high sensitivity 156 was accompanied by a time -dependent increase in basal secretion in the dark, likely attributable to the 157 hypersensitivity of the dissociative mechanism to ambient light exposure during sample handling. In contrast, a -158 BLAST maintained exceptionally low background levels because the RXR retention motif is inherently 159 incorporated into the cargo module itself, resulting in a robust signal-to-noise ratio over prolonged durations. Thus, 160 the toolkit offers a strategic trade -off: d-BLAST for applications requiring immediate release, and a -BLAST for 161 those demanding minimal background. 162 We further validated the universality of BLAST across diverse mammalian cell lines. Both systems functioned 163 robustly in HeLa, BHK -21, and HepG2 cells (Supplementary Fig. 4a -d), confirming that BLAST relies on the 164 conserved secretory machinery of mammalian cells and is applicable to a wide range of biological contexts. 165 Next, we benchmarked BLAST against optoPASS [23], a recently developed state -of-the-art system that triggers 166 secretion via the light-induced reassembly of a split protease. While optoPASS also utilizes blue light, it relies on 167 an influenza virus -derived Furin cleavage site (FCS: RRRKKR/GL), whereas our BLAST employs a Sindbis 168 virus-derived sequence (SGRSKR/SV). To ensure a rigorous, variable -controlled comparison, we engineered a 169 chimeric variant named NEW optoPASS, in which the FCS and transmembrane (TM) domains were replaced 170 with those from BLAST (Fig. 3d). Under identical conditions, although NEW optoPASS yielded higher absolute 171 secretion levels, it suffered from significant basal leakage in the dark, likely due to the irreversible nature of 172 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint proteolytic escape. Conversely, BLAST demonstrated a significantly superior signal-to-noise ratio (Fig. 3e). This 173

was further supported by reciprocal domain-swapping experiments (Supplementary Fig. 5). 174 To rigorously demonstrate the dynamic, multi -toggling capability of our toolkit —a critical requirement for 175 mimicking pulsatile physiological signaling —we evaluated the reversibility of BLAST in direct comparison to 176 NEW optoPASS. We hypothesized that the non -destructive, reversible masking of the ER retention signal in 177 BLAST would enable the immediate cessation of cargo exit upon light withdrawal, offering superior ON/OFF 178 control. To test this, we tracked cumulative SEAP secretion over a 9 -hour period under an alternating light/dark 179 cycle (3 h ON / 3 h OFF / 3 h ON). Supernatants were sampled at the corresponding time points (0, 3, 6, and 9 h). 180 This schedule allowed us to precisely monitor secretion profiles during illumination windows, while evaluating 181 dark-state retention by comparing cumulative levels before and after the dark period. 182 Strikingly, the resulting profiles revealed a profound mechanistic divergence, highlighting the fundamental 183 difference between irreversible cleavage and reversible masking (Figure 3f). The NEW optoPASS system 184 exhibited a continuous, unchecked increase in cumulative secretion; a substantial accumulation of SEAP was 185 detected at 6 h compared to the 3 h time point (Phase 2), demonstrating that proteolytic escape completely fails 186 to terminate protein release during the dark phase. In stark contrast, both a-BLAST and d-BLAST generated highly 187 controlled "staircase" secretion profiles. While blue light illumination robustly triggered secretion, cumulative 188 SEAP levels firmly plateaued during the intervening dark phase (between 3 h and 6 h), confirming that the re -189 establishment of steric masking instantaneously arrests further cargo release. This strict ON/OFF control is further 190 corroborated by the area under the curve (AUC) analysis (Figure 3g –i), which clearly quantifies the minimal 191 secretion during the dark phase for both BLAST systems compared to the continuous leakage of NEW optoPASS. 192 Collectively, these results confirm that unlike irreversible "one -off" protease-dependent switches, the reversible 193 retention mechanism of BLAST permits strict, repeatable multi -toggling, providing the unprecedented temporal 194 resolution required for advanced synthetic biology applications. 195 196 2.5 Visualization of intercellular signaling and high-precision spatial control 197 To visualize the intercellular transfer of secreted proteins, we engineered a synthetic paracrine signaling circuit 198 utilizing EGFP as a fluorescent cargo. We established a sender -receiver co -culture system: sender cells were 199 transfected with the BLAST-EGFP module, while receiver cells expressed a membrane -tethered GFP-nanobody 200 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint designed to capture secreted EGFP on their surface (Fig. 4a)[36]. Sender and receiver cells were seeded on separate 201 coverslips and then co-cultured in close proximity, with the receiver coverslip inverted over the sender monolayer 202 (Fig. 4b). Following blue light illumination, we analyzed the intercellular transfer via confocal microscopy. In the 203 dark conditions, receiver cells showed no observable green fluorescence, confirming the tight retention of EGFP 204 within the sender cells. In contrast, under blue light stimulation, distinct EGFP signals were detected on the plasma 205 membrane of receiver cells, strongly co-localizing with the mCherry surface marker (Fig. 4c). This result visually 206 confirms that BLAST facilitates the efficient secretion and subsequent intercellular transfer of functional cargo 207 proteins, which can be specifically recognized by neighboring target cells. 208 Furthermore, we demonstrated the capacity of BLAST for spatially precise control using a photomasking 209 strategy. Monolayers of HEK293T cells expressing a -BLAST-EGFP or d -BLAST-EGFP were illuminated 210 through a 2 -mm slit mask (Supplementary Fig. 6a, b). To enable rigorous quantitative analysis, the d -BLAST 211 system utilized its intrinsic mCherry spacer as a reference marker. In contrast, because the a -BLAST cargo lacks 212 this structural component, we co -transfected mCherry to serve as a normalization control. As anticipated, we 213 observed a spatially defined negative photopatterning effect. Specifically, the targeted region exposed to blue light 214 exhibited a pronounced reduction in intracellular EGFP fluorescence, while the signal from the mCherry reference 215 marker remained completely unchanged compared to the surrounding dark areas (Supplementary Fig. 6c, d). This 216 localized depletion of the EGFP cargo, coupled with the stable mCherry signal, clearly indicates that protein 217 secretion was actively triggered only within the illuminated zone without affecting overall cell viability or basal 218 expression, thereby confirming that BLAST enables high-resolution spatiotemporal control of protein release. 219 220 2.6 Light-triggered release of therapeutic proteins: Insulin and Interleukin-12 221 To evaluate the clinical potential and translatability of BLAST, we tested its ability to regulate the secretion of 222 two medically significant proteins: human preproinsulin and the immunomodulatory cytokine interleukin-12 (IL-223 12). First, we engineered a preproinsulin cargo module by flanking the C-peptide with two Furin cleavage sites to 224 ensure its maturation into active insulin via the endogenous secretory pathway (Fig. 5a )[37]. To facilitate efficient 225 processing, we co -expressed Furin protease in HEK293T cells and quantified insulin release by measuring 226 secreted C-peptide levels via enzyme-linked immunosorbent assay (ELISA). Consistent with our SEAP reporter 227 results, both a-BLAST and d-BLAST induced significant insulin secretion within 2 h of illumination, exhibiting 228 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint a clear dependency on the duration of light exposure (Fig. 5b). 229 Next, we extended the application to IL-12, a potent heterodimeric cytokine (comprising p35 and p40 subunits) 230 critical for T-cell activation and cancer immunotherapy (Fig. 5c)[38,39]. Given its complex hetero dimeric structure, 231 the successful secretion of IL-12 serves as a stringent test for the capacity of BLAST to handle multi -subunit 232 proteins. Subsequent ELISA quantification revealed that d-BLAST triggered detectable IL-12 release within 2 h, 233 whereas a-BLAST required 3 h to achieve comparable induction, further confirming the superior kinetic response 234 of the dissociative system (Fig. 5d). 235 To provide an orthogonal validation of IL -12 release beyond ELISA -based quantification, and to specifically 236 verify the biological activity of the optogenetically released cytokine, we employed IL -12 reporter cells (HEK -237 Blue™ IL-12), which express SEAP upon STAT4 activation (Supplementary Fig. 7). These assays demonstrated 238 a robust 12.4 - and 11.3-fold induction in IL -12-mediated signaling for a -BLAST and d -BLAST, respectively, 239 compared to dark-state controls. Taken together, these findings collectively demonstrate that BLAST provides a 240 versatile, on-demand platform for the controlled delivery of diverse therapeutic proteins, ranging from metabolic 241 hormones to complex immunomodulators. 242 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint 3 Discussion 243 244 The demand for precise spatiotemporal control over protein signaling has driven the development of various 245 optogenetic post-translational toolkits in synthetic biology[40-42]. While traditional transcription-based circuits are 246 effective for long -term modulation, their inherent 4 -to 6 h kinetic lag limits their utility for mimicking rapid 247 physiological events such as hormonal surges or neurotransmitter release[7,43-45]. Although recent advances in post-248 translational secretion, such as optoPASS[23] and RELEASE[20], have significantly improved temporal resolution, 249 they largely rely on irreversible proteolytic cleavage to achieve secretion[22,46,47]. 250 Here, we present BLAST, a versatile optogenetic platform that achieves rapid and reversible control of protein 251 secretion by harnessing light -tunable PPIs. The architectural hallmark of BLAST is its ability to override ER 252 retention through steric masking rather than proteolytic cleavage. Crucially, this non -destructive mechanism 253 endows BLAST with true reversibility —a feature fundamentally lacking in cleavage -based systems. As 254 demonstrated by our alternating light/dark multi -toggling assays, while proteolytic switches lose control and 255 continuously leak cargo in the dark following initial activation, BLAST achieves a highly regulated "staircase" 256 secretion profile. The immediate re -establishment of the RXR masking upon light withdrawal completely and 257 instantaneously arrests further cargo release. This capacity for repeated, strict ON/OFF switching provides the 258 unprecedented temporal resolution required to emulate dynamic, pulsatile physiological signaling. 259 By situating the regulator module in the cytoplasm rather than the ER lumen, we bypassed the extensive protein 260 engineering typically required to adapt optogenetic domains to the oxidative ER environment [21,48,49]. This 261 modularity allows BLAST to be easily repurposed with alternative PPI pairs. Furthermore, our strategic selection 262 of the RXR motif over the common KK MP motif provides a significant advantage. While KK MP-mediated 263 retention is highly sensitive to its proximity to the membrane [31,50,51], RXR motif allow s for greater spatial 264 flexibility[26,31]. In the d -BLAST system, this enabled us to incorporate mCherry as a structural spacer , without 265 compromising retention efficiency, thereby facilitating the real-time visualization of cargo trafficking—a feature 266 often challenging to implement in other systems. 267 A key advantage of the BLAST toolkit is the availability of two complementary operating modes, allowing users 268 to tailor the system to specific experimental constraints. The associative a -BLAST, driven by iLID/SspB 269 dimerization, offers exceptional stringency with minimal basal leakage in the dark state. This makes it the optimal 270 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint choice for applications requiring a strict “OFF” switch, such as the regulation of potent cytokines or toxic payloads 271 where background signaling must be minimized[52]. Conversely, the dissociative d-BLAST, based on LOV2/Zdk1, 272 exhibits superior rapid-response kinetics, triggering significant secretion within 1 h of illumination. Although this 273 sensitivity is accompanied by slightly higher basal secretion, d -BLAST is particularly advantageous for 274 mimicking acute physiological bursts, such as insulin spikes or neurotransmitter release, where speed is paramount. 275 Thus, BLAST offers a strategic trade -off: a-BLAST for high signal -to-noise precision, and d -BLAST for rapid 276 temporal resolution. 277 In comparison to other light-triggered release systems, such as the UV-based PhoCl switch[53,54], BLAST offers 278 superior biocompatibility. While single -component UV switches are structurally compact, the high -energy UV 279 radiation required for their activation is often cytotoxic and unsuitable for prolonged live-cell imaging[55]. BLAST 280 utilizes 470 nm blue light, which is not only biocompatible but also permits high -resolution spatial patterning, as 281 demonstrated by our slit -masking experiments. Unlike chemical induction systems like RUSH [19], which are 282 limited by the diffusion and slow washout of ligands, BLAST enables localized, on-demand protein delivery with 283 high spatial precision. 284 Beyond its utility as a basic research tool for studying vesicular transport and protein trafficking, BLAST holds 285 significant therapeutic potential. We successfully demonstrated the light -triggered release of functionally active 286 insulin and IL -12 in HEK293T cells . Given the broad applicability of the BLAST platform across diverse 287 mammalian cell lineages established earlier, this success suggest that BLAST can be used to engineer various 288 non-specialized cells into programmable bio -factories for the localized delivery of therapeutic proteins. We 289 anticipate that BLAST will serve as a foundational tech nique for synthetic biology, enabling the development of 290 next-generation cell-based therapies with unprecedented spatiotemporal control. 291 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint 4 Experimental Section 292 293 4.1 Design and construction of plasmid vectors 294 All plasmids used in this study are detailed in the Supplementary Information. Plasmids were constructed using 295 standard molecular cloning techniques. DNA fragments were obtained from Addgene or synthesized (Bionics, 296 Republic of Korea) and digested with restriction enzymes (New England Biolabs; NEB). Ligation was performed 297 using T4 DNA Ligase (NEB, M0202S), and the resulting products were transformed into DH5α Escherichia coli 298 competent cells (Enzynomics, Republic of Korea). PCR amplifications were performed using KOD Hot Start 299 DNA Polymerase (Novagen) or Platinum SuperFi II PCR Master Mix (Invitrogen) according to the 300 manufacturer’s protocols. All plasmid sequences were verified by Sanger sequencing (Bionics , Republic of 301 Korea). Large-scale plasmid preparation was performed using 200 m L cultures. SnapGene software (Insightful 302 Science) was used for vector design and sequence documentation. 303 304 4.2 Cell culture and plasmid DNA transfection 305 HEK293T, HeLa, HepG2, and BHK-21 cells were obtained from the American Type Culture Collection (ATCC). 306 HEK-Blue™ IL-12 reporter cells were purchased from InvivoGen. All cell lines were confirmed to be 307 mycoplasma-free and maintained in Dulbecco’s Modified Eagle’s Medium (DMEM; WELGENE, Republic of 308 Korea) supplemented with 10% fetal bovine serum (FBS; Gibco, USA) and 1% penicillin -streptomycin (Gibco) 309 at 37°C in a humidified 5% CO₂ atmosphere. For subculturing, cells were dissociated with 0.25% trypsin-EDTA, 310 centrifuged at 123 × g for 1 m, and resuspended in fresh medium. Cells were seeded onto poly -D-lysine 311 (Invitrogen)-coated 24-well plates at a density of 1 × 105 cells per well. Cell viability and counting were assessed 312 using a Countess automated cell counter (Invitrogen). HEK293T cells were transfected 24 h post-seeding using 313 the jetOPTIMUS DNA transfection kit (Polyplus) following the manufacturer’s instructions (0.25 μg DNA and 314 0.25 μl reagent per well). HeLa, HepG2, and BHK-21 cells were transfected 24 h post-seeding using the PEI MAX 315 DNA transfection kit (Polyscience) following the manufacturer’s instructions (0.5 μg DNA and 2 μ g reagent per 316 well). 317 318 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint 4.3 Blue-light illumination 319 For optogenetic blue light irradiation of the cells as described in the figure, a 470 -nm light-emitting diode (LED) 320 panel (Green Energy Star) was installed inside a 37°C incubator maintained at 5% CO2. To prevent overheating 321 via direct contact with the LEDs, sample culture plates were placed on three empty plates (total height, 6cm). The 322 duty cycle of the LED panel was controlled by two serially connected electronic timers (IRT16-D, Han Seung). 323 The light intensity was calibrated to 4.0 mW cm⁻² using a power meter (PM100D, Thorlabs). For dark conditions, 324 plates were strictly shielded from ambient light using aluminum foil. 325 326 4.4 Chemical stimulation 327 Rapamycin (Tocris) was prepared as a 3 mM stock solution in DMSO (Sigma -Aldrich). For chemo -BLAST 328 induction, cells were treated with the indicated concentration of rapamycin or an equivalent volume of DMSO as 329 a vehicle control. 330 331 4.5 SEAP assay 332 For the quantification of SEAP, the cell culture supernatants were collected and centrifug ed at 15,800 × g for 5 333 m. Endogenous alkaline phosphatase was heat-inactivated at 60°C for 1 h. A 40 μl aliquot of the supernatant was 334 mixed with 100 μl of reaction buffer (5.1 M diethanolamine, L-homoarginine, and 10 mM MgCl₂) and incubated 335 at 37°C for 10 m. Subsequently, 60 μl of pNPP substrate (Thermo Fisher Scientific) was added. Absorbance at 336 405 nm was monitored at 30 s intervals for 30 m using an Infinite F50 microplate reader (TECAN). Vmax values 337 were calculated using Magellan software (TECAN). 338 339 4.6 Reversibility assay 340 To evaluate the dynamic, reversible control of protein secretion, HEK293T cells transfected with a -BLAST, d-341 BLAST, or NEW optoPASS constructs were seeded in 6 -well plates. At 48 h post-transfection, the cells were 342 subjected to a 9 h alternating light/dark cycle (3 h ON / 3 h OFF / 3 h ON). Blue light stimulation was delivered 343 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint using a 470 nm LED panel with a pulsed duty cycle (2 s ON / 58 s OFF). During the dark phases, culture plates 344 were strictly shielded from ambient light using aluminum foil. To track cumulative secretion profiles, 200 μL 345 aliquots of culture supernatant were sampled from each well (initial volume: 2 mL) at designated time points (0, 346 3, 6, and 9 h). Immediately following each sampling, an equal volume (200 μL) of pre-warmed fresh medium was 347 replenished to maintain a constant culture volume and minimize cellular stress. To precisely quantify the 348 cumulative SEAP secretion and account for the serial dilution caused by the repeated sampling and replenishment 349 process, the raw SEAP activity ( 𝐶𝑛, measured as Vmax) at each 𝑛-th time point was mathematically corrected. 350 The corrected cumulative concentration (𝐶𝑐𝑜𝑟𝑟𝑒𝑐𝑡𝑒𝑑,𝑛) was calculated using the following equation: 351 𝐶𝑐𝑜𝑟𝑟𝑒𝑐𝑡𝑒𝑑,𝑛 = 𝐶𝑛 + 𝑣 𝑉 ∑ 𝐶𝑖 𝑛−1 𝑖=1 352 where 𝑣 represents the sampled volume (200 μL), 𝑉 represents the total culture volume in the well (2,000 μL), 353 and the summation term accounts for the total mass of SEAP removed during all preceding sampling events. 354 355 4.7 Image processing and visualization 356 To visualize intercellular protein transfer, receiver cells (transfected with the reporter) and sender cells 357 (transfected with BLAST) were prepared on separate coverslips and wells, respectively. The coverslip containing 358 receiver cells was transferred to the sender cell well, placed with the cell-attached side facing downward to ensure 359 proximity. Following 24 h of blue light stimulation (2 s ON / 58 s OFF), the coverslips were transferred to a new 360 plate, fixed with 4% paraformaldehyde (CUREBIO) for 20 m, and washed with PBS. Samples were mounted 361 using Crystal Mount (Biomeda). Confocal images were acquired using an LSM800 microscope (Zeiss) with a 362 40×/0.8 NA objective. For the slit -masking experiment (Supplementary Fig . 6), fluorescence scanning was 363 performed using a Typhoon FLA 9500 (GE Healthcare). 364 365 4.8 Quantification of insulin and IL-12 secretion 366 For insulin quantification, HEK293T cells were co -transfected with the BLAST -preproinsulin constructs and 367 Furin protease to enable maturation. Secreted human C -peptide levels in the supernatant were quantified using a 368 Human C-peptide ELISA Kit (R&D Systems; Catalog #DICP00). Similarly, IL-12 secretion was measured using 369 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint a Human IL-12 p70 ELISA Kit (R&D Systems; Catalog #D1200) according to the manufacturer’s instructions. 370 To utilize HEK -Blue™ IL-12 reporter cells for measuring optogenetically induced IL -12 secretion, a 5 -day 371 experimental protocol was employed (Supplementary Fig. 7) . HEK293T cells and HEK -Blue™ IL-12 reporter 372 cells were seeded into 24 -well plates on Day 1 and Day 3, respectively. Twenty -four h after seeding, HEK293T 373 cells were transfected with BLAST -IL-12 (Day 2). Twenty-four h post-transfection, cells were cultured for 24 h 374 under dark or blue light conditions (Day 3). On Day 4, culture supernatants containing secreted IL -12 were 375 collected and transferred to HEK-Blue™ IL-12 reporter cells seeded on Day 3. After 24 h of incubation to promote 376 signal transduction, SEAP activity was measured on Day 5. 377 378 4.9 Statistical analysis 379 Data are presented as mean values ± S.D . from at least three biologically independent experiments. Statistical 380 comparisons were performed using one-way or two-way ANOVA followed by Tukey’s multiple comparisons test 381 using GraphPad Prism or Microsoft Excel. Adjusted P -values are denoted as: ns (not significant), * p < 0.05, **p 382 < 0.01, ***p < 0.001, and ****p < 0.0001. 383 384 4.10 Manuscript Preparation and AI Usage 385 During the preparation of this manuscript, the authors utilized Gemini (Google) for English language editing, 386 stylistic polishing, and formatting of the text. After using this tool, the authors carefully reviewed and extensively 387 edited the content. The authors take full responsibility for the final content, underlying data, and scientific 388 accuracy of this publication. 389 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint

390 This study was supported by grants from the National Research Foundation of Korea (grant nos. RS -2020-391 NR051270 and RS-2024-00340694 to D.L.). We thank all members of the laboratory for their critical discussions 392 and comments. 393 394 395 Conflicts of Interest 396 The authors declare no conflict of interest. 397 398 399 Author Contributions 400 M.S., S.L., and D.L. developed the concept. M.S., S.L., and M.C. designed and implemented the methodology 401 and performed the experiments. M.S., S. L., M.C., and D.L. performed data curation and visualization. D.L. 402 supervised the project and secured funding. M.S., S.L., and D.L. wrote the manuscript. All authors reviewed and 403 edited the manuscript. 404 405 406 Data Availability Statement 407 The data that support the findings of this study are available from the corresponding author upon reasonable 408 request. 409 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint 410 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint Figure 1 411 Development of the associative Blue Light-Assisted Secretion Toolkit (a -BLAST). (a) Schematic illustration of 412 the a-BLAST mechanism. The a-BLAST system consists of two parts: a cargo module and an associative regulator 413 module. The cargo module includes the POI, a Furin cut site, a TM domain, SspB repeats, and an RXR -type ER 414 retention sequence. The associative regulator comprises iLID, which binds to SspB under blue light. Exposure to 415 blue light triggers SspB -iLID dimerization, which sterically interferes with the RXR -mediated ER retention. 416 Consequently, the cargo is transported to the trans -Golgi network, where it is cleaved by endogenous Furin 417 protease to release the POI. (b) Construct configurations of a -BLAST components: the cargo (upper) and the 418 associative regulator (lower). The cargo features SEAP as a reporter, a TM domain, a Furin cleavage site, 6× SspB, 419 and an RXR motif. The regulator variants include iLID mutants with delayed recovery kinetics. (c) Comparative 420 analysis of blue light-dependent protein secretion based on kinetic modifications of iLID. The iLID double mutant 421 (N414L, V416L) exhibited the highest fold change (16.3 -fold). Blue light (470 nm) was applied for 24 h with a 422 duty cycle (2 s ON / 58 s OFF). (d) Plasmid library demonstrating sequential incrementation of SspB repeats (A: 423 1×, 3×, 6×, and 9 ×) and the dual -mutant iLID (B). (e) Histogram of SEAP activity and fold changes based on 424 varying transfection ratios and SspB valency. The configuration using 6× SspB and a 1:3 transfection ratio yielded 425 the most robust fold change (18.1 -fold). (f) Heatmap summarizing blue light -induced SEAP activity (white -to-426 blue gradient) and fold changes (white -to-gray gradient). Open circles represent individual measurements from 427 three biologically independent samples. Data are presented as means ± S.D. Significance was assessed using one-428 way ANOVA followed by Tukey’s multiple comparisons test. (ns = not significant, *P < 0.05, **P < 0.01, ***P 429 < 0.001, ****P < 0.0001). 430 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint 431 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint Figure 2 432 Development of the dissociative-BLAST (d-BLAST) system. (a) Schematic diagram of the d-BLAST system. d-433 BLAST consists of two components: a car go module and a dissociative regulator module. The car go module 434 contains a protein of interest (POI; e.g., SEAP), a Furin cleavage site, a transmembrane (TM) domain, an mCherry 435 spacer, and a LOV2 domain. The dissociative regulator consists of Zdk1 fused to a C-terminal RXR ER retention 436 motif. In the dark, Zdk1 binds tightly to LOV2, recruiting the RXR motif to the car go and retaining it in the 437 endoplasmic reticulum (ER). Blue light induces the dissociation of Zdk1 -RXR from the cargo, allowing the POI 438 to be transported to the trans-Golgi network (Trans-GA), where it is cleaved by native Furin protease and secreted. 439 (b) Structural maps and experimental timeline. Schematic representation of the car go and dissociative regulator 440 constructs, along with a 3-day experimental schedule. (c) Kinetic optimization using LOV2 mutants. Comparison 441 of different LOV2 domain mutants (WT, V416T, V416I, and V416L) to control protein secretion. The V416L 442 mutant, which has the slowest reversion kinetics, exhibited the highest fold change in SEAP secretion. (d) 443 Schematic of ER membrane-localized regulator. To reduce basal leakage, the dissociative regulator was anchored 444 to the ER membrane by fusing a Cytochrome P450 signal -anchor sequence (P450n) to the N -terminus of Zdk1-445 RXR. (e) Influence of regulator localization on secretion. Comparison of SEAP secretion levels using different 446 Zdk1 variants. Anchoring the regulator (P450n-Zdk1-RXR) to the ER membrane significantly reduced dark-state 447 leakage compared to the cytosolic version, improving the signal -to-noise ratio to 20.6 -fold. The Zdk1 -AAA 448 mutant lacks retention function, resulting in high, light -independent secretion. (f) Optimization of experimental 449 schedules. Comparison of d -BLAST performance between a 3 -day and 4 -day schedule. The 4 -day schedule 450 significantly increased the total level of light -induced protein secretion (79.1 -fold). Open circles represent 451 individual measurements from three biologically independent samples. Data are presented as means ± S.D. 452 Significance was assessed using one -way ANOVA followed by Tukey’s multiple comparisons test (ns = not 453 significant, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). 454 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint 455 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint Figure 3 456 Characterization of a -BLAST and d -BLAST depending on blue light duration and reversibility. (a) Structural 457 diagrams for kinetic characterization. Schematic representation of optimized SEAP-a-BLAST (upper) and SEAP-458 d-BLAST (lower). The a-BLAST system utilizes 6× SspB and the iLID (N414L, V416L) double mutant. The d -459 BLAST system employs the LOV2 domain and the P450n -anchored Zdk1 -RXR regulator. (b) Experimental 460 timeline. Cells were subjected to varying intervals of blue light illumination (0, 0.5, 1, 4, 8, and 24 h) starting 24 461 h post-transfection. (c) Summary graphs showing SEAP secretion levels for a-BLAST (left) and d-BLAST (right) 462 across different illumination times. Both systems exhibit light-dependent secretion, with d-BLAST showing rapid 463 release within 1 h (3.5 -fold) and a 2.2 -fold increase even at 0.5 h. a -BLAST shows robust, significant secretion 464 starting at 4 h (5.2 -fold). (d) Schematic of the NEW optoPASS benchmarking construct. The original Furin 465 cleavage site and TM domain were replaced with those used in BLAST. Blue light triggers pMag -nMag 466 dimerization, reconstituting split protease (TVMVp) activity to excise the KKMP retention motif. (e) Comparison 467 of SEAP secretion and signal -to-noise ratios between a -BLAST, d-BLAST, and NEW optoPASS. While NEW 468 optoPASS shows higher absolute secretion, the BLAST systems demonstrate superior dynamic ranges (20.5-fold 469 for a-BLAST, 29.2-fold for d-BLAST) due to lower dark -state leakage compared to NEW optoPASS (9.8 -fold). 470 (f) Cumulative SEAP secretion profiles demonstrating the reversibility of the systems over a 9 -h alternating 471 light/dark cycle (3 h ON / 3 h OFF / 3 h ON). Both a -BLAST (green line) and d-BLAST (orange line) generate 472 strictly controlled staircase profiles, instantaneously halting secretion during the dark phase. In contrast, NEW 473 optoPASS (magenta line) exhibits continuous, irreversible leakage. (g –i) Area under the curve (AUC) 474 quantification of secretion levels during each 3-h phase for (g) a-BLAST (green bars), (h) d-BLAST (orange bars), 475 and (i) NEW optoPASS (magenta bars). The strictly minimal AUC during Phase 2 (Dark) for both BLAST 476 systems definitively confirms robust ON/OFF control compared to the cleavage -based system . Open circles 477 represent individual measurements from three biologically independent samples . Data are presented as means ± 478 S.D. Statistical significance was assessed using one-way ANOVA followed by Tukey’s multiple comparisons test 479 (ns = not significant, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). 480 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint 481 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint Figure 4 482 Visualization of a -BLAST and d -BLAST with EGFP and GFP nanobody. ( a) Plasmid configurations of the 483 BLAST-EGFP systems and the reporter. Schematic of EGFP -fused a-BLAST (left) and d-BLAST (middle). For 484 visualization, the SEAP reporter was replaced with EGFP. The receiver reporter construct (right) consists of a 485 membrane-tethered GFP-nanobody (GBP) fused to an mCherry transfection marker. (b) Schematic illustrating the 486 sequential co-culture procedure. Sender cells (transfected with BLAST-EGFP) and receiver cells (transfected with 487 the GBP reporter) were prepared on separate coverslips. The receiver coverslip was then transferred onto the 488 sender cell plate. For the light -induced group, blue light stimulation triggers EGFP secretion and subsequent 489 capture by the nanobody on the receiver cells, while dark conditions strictly retain EGFP within the sender cells. 490 (c) Representative confocal images of a-BLAST and d-BLAST signaling. Confocal microscopy images showing 491 receiver cells in dark (upper) and blue light (lower) conditions. The green signal (EGFP) indicates the successful 492 capture of secreted EGFP by the nanobody on the plasma membrane of receiver cells. The red signal (mCherry) 493 confirms the presence and localization of the reporter module on the cell surface. Scale bar: 20 μm. 494 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint 495 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint Figure 5 496 Secretion of therapeutic proteins with a-BLAST and d -BLAST. ( a) The human preproinsulin (preproINS) 497 construct contains a signal peptide, B -chain (yellow), C -peptide (dark gray), and A -chain (yellow), with 498 engineered Furin cleavage sites flanking the C-peptide for maturation. To facilitate efficient processing within the 499 Golgi apparatus, Furin protease was co -transfected with the BLAST modules. (b) Kinetic profiling of light -500 induced insulin secretion. Summary graphs of secreted C -peptide levels, quantified by ELISA, as a proxy for 501 insulin secretion from a-BLAST (left) and d-BLAST (right). Both systems exhibited significant, time -dependent 502 insulin release starting from 2 h of illumination (8.2-fold for a-BLAST, 8.4-fold for d-BLAST), reaching maximal 503 induction at 24 h (13.8-fold for a-BLAST, 19.3-fold for d-BLAST). (c) Plasmid configurations for IL-12 secretion. 504 Schematic of the heterodimeric cytokine IL-12-a-BLAST (left) and d-BLAST-IL-12 (right) constructs. (d) Kinetic 505 profiling of light-induced IL-12 secretion. Summary graphs showing IL-12 secretion levels measured by ELISA. 506 Significant secretion was observed starting at 3 h for a-BLAST (2.5-fold) and 2 h for d-BLAST (2.2-fold). At the 507 24 h time point, d-BLAST (4.7-fold) demonstrated a slightly higher dynamic range compared to a -BLAST (4.5-508 fold). Open circles represent individual measurements from three biologically independent samples. Data are 509 presented as means ± S.D. Statistical significance was assessed using one -way ANOVA followed by Tukey’s 510 multiple comparisons test (ns = not significant, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). 511 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint

512 513 [1] A. M. Benham, Cold Spring Harb Perspect Biol 2012, 4, a012872. 514 https://doi.org/10.1101/cshperspect.a012872 515 [2] C. Viotti, Methods Mol Biol 2016, 1459, 3. https://doi.org/10.1007/978-1-4939-3804-9_1 516 [3] J. M. Gutierrez, A. Feizi, S. Li, T. B. Kallehauge, H. Hefzi, L. M. Grav, D. Ley, D. Baycin Hizal, 517 M. J. Betenbaugh, B. Voldborg, H. Faustrup Kildegaard, G. Min Lee, B. O. Palsson, J. Nielsen, 518 N. E. Lewis, Nat Commun 2020, 11, 68. https://doi.org/10.1038/s41467-019-13867-y 519 [4] J. B. Welsh, L. M. Sapinoso, S. G. Kern, D. A. Brown, T. Liu, A. R. Bauskin, R. L. Ward, N. J. 520 Hawkins, D. I. Quinn, P. J. Russell, R. L. Sutherland, S. N. Breit, C. A. Moskaluk, H. F. Frierson, 521 Jr., G. M. Hampton, Proc Natl Acad Sci U S A 2003, 100, 3410. 522 https://doi.org/10.1073/pnas.0530278100 523 [5] Z. Fu, E. R. Gilbert, D. Liu, Curr Diabetes Rev 2013, 9, 25 524 [6] Y. Liu, H. Zhang, X. Li, T. He, W. Zhang, C. Ji, J. Wang, Mol Biol Rep 2025, 52, 236. 525 https://doi.org/10.1007/s11033-025-10316-6 526 [7] O. Alcazar, P. Buchwald, Front Endocrinol (Lausanne) 2019, 10, 680. 527 https://doi.org/10.3389/fendo.2019.00680 528 [8] G. A. Stamatiades, U. B. Kaiser, Mol Cell Endocrinol 2018, 463, 131. 529 https://doi.org/10.1016/j.mce.2017.10.015 530 [9] S. L. Lightman, M. T. Birnie, B. L. Conway -Campbell, Endocr Rev 2020, 41. 531 https://doi.org/10.1210/endrev/bnaa002 532 [10] A. M. Monteys, A. A. Hundley, P. T. Ranum, L. Tecedor, A. Muehlmatt, E. Lim, D. Lukashev, 533 R. Sivasankaran, B. L. Davidson, Nature 2021, 596, 291. https://doi.org/10.1038/s41586-021-534 03770-2 535 [11] Y. Zhou, D. Kong, X. Wang, G. Yu, X. Wu, N. Guan, W. Weber, H. Ye, Nat Biotechnol 2022, 536 40, 262. https://doi.org/10.1038/s41587-021-01036-w 537 [12] H. S. Li, D. V. Israni, K. A. Gagnon, K. A. Gan, M. H. Raymond, J. D. Sander, K. T. Roybal, J. K. 538 Joung, W. W. Wong, A. S. Khalil, Science 2022, 378, 1227. 539 https://doi.org/10.1126/science.ade0156 540 [13] Y. Pan, S. Yoon, J. Sun, Z. Huang, C. Lee, M. Allen, Y. Wu, Y. J. Chang, M. Sadelain, K. K. 541 Shung, S. Chien, Y. Wang, Proc Natl Acad Sci U S A 2018, 115, 992. 542 https://doi.org/10.1073/pnas.1714900115 543 [14] M. Cui, S. Lee, S. H. Ban, J. R. Ryu, M. Shen, S. H. Yang, J. Y. Kim, S. K. Choi, J. Han, Y. Kim, 544 K. Han, D. Lee, W. Sun, H. B. Kwon, D. Lee, Nat Chem Biol 2024, 20, 353. 545 https://doi.org/10.1038/s41589-023-01480-6 546 [15] A. Bertschi, B. A. Stefanov, S. Xue, G. Charpin-El Hamri, A. P. Teixeira, M. Fussenegger, Nucleic 547 Acids Res 2023, 51, e28. https://doi.org/10.1093/nar/gkac1256 548 [16] B. Z. Stanton, E. J. Chory, G. R. Crabtree, Science 2018, 359. 549 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint https://doi.org/10.1126/science.aao5902 550 [17] M. Xie, H. Ye, H. Wang, G. Charpin -El Hamri, C. Lormeau, P. Saxena, J. Stelling, M. 551 Fussenegger, Science 2016, 354, 1296. https://doi.org/10.1126/science.aaf4006 552 [18] K. Krawczyk, S. Xue, P. Buchmann, G. Charpin -El-Hamri, P. Saxena, M. D. Hussherr, J. Shao, 553 H. Ye, M. Xie, M. Fussenegger, Science 2020, 368, 993. 554 https://doi.org/10.1126/science.aau7187 555 [19] G. Boncompain, S. Divoux, N. Gareil, H. de Forges, A. Lescure, L. Latreche, V. Mercanti, F. 556 Jollivet, G. Raposo, F. Perez, Nat Methods 2012, 9, 493. https://doi.org/10.1038/nmeth.1928 557 [20] A. E. Vlahos, J. Kang, C. A. Aldrete, R. Zhu, L. S. Chong, M. B. Elowitz, X. J. Gao, Nat Commun 558 2022, 13, 912. https://doi.org/10.1038/s41467-022-28623-y 559 [21] A. Praznik, T. Fink, N. Franko, J. Lonzarić, M. Benčina, N. Jerala, T. Plaper, S. Roškar, R. Jerala, 560 Nature Communications 2022, 13, 1323. https://doi.org/10.1038/s41467-022-28971-9 561 [22] M. Mansouri, P. G. Ray, N. Franko, S. Xue, M. Fussenegger, Nucleic Acids Res 2023, 51, e1. 562 https://doi.org/10.1093/nar/gkac916 563 [23] X. Wang, L. Kang, D. Kong, X. Wu, Y. Zhou, G. Yu, D. Dai, H. Ye, Nature Chemical Biology 564 2023. https://doi.org/10.1038/s41589-023-01433-z 565 [24] F. Kawano, H. Suzuki, A. Furuya, M. Sato, Nat Commun 2015, 6, 6256. 566 https://doi.org/10.1038/ncomms7256 567 [25] H. Jung, S. W. Kim, M. Kim, J. Hong, D. Yu, J. H. Kim, Y. Lee, S. Kim, D. Woo, H. S. Shin, B. O. 568 Park, W. D. Heo, Nat Commun 2019, 10, 314. https://doi.org/10.1038/s41467-018-08282-8 569 [26] M. Gassmann, C. Haller, Y. Stoll, S. Abdel Aziz, B. Biermann, J. Mosbacher, K. Kaupmann, B. 570 Bettler, Mol Pharmacol 2005, 68, 137. https://doi.org/10.1124/mol.104.010256 571 [27] K. Michelsen, H. Yuan, B. Schwappach, EMBO Rep 2005, 6, 717. 572 https://doi.org/10.1038/sj.embor.7400480 573 [28] G. Guntas, R. A. Hallett, S. P. Zimmerman, T. Williams, H. Yumerefendi, J. E. Bear, B. Kuhlman, 574 Proc Natl Acad Sci U S A 2015, 112, 112. https://doi.org/10.1073/pnas.1417910112 575 [29] S. P. Zimmerman, R. A. Hallett, A. M. Bourke, J. E. Bear, M. J. Kennedy, B. Kuhlman, 576 Biochemistry 2016, 55, 5264. https://doi.org/10.1021/acs.biochem.6b00529 577 [30] H. Wang, M. Vilela, A. Winkler, M. Tarnawski, I. Schlichting, H. Yumerefendi, B. Kuhlman, R. 578 Liu, G. Danuser, K. M. Hahn, Nat Methods 2016, 13, 755. https://doi.org/10.1038/nmeth.3926 579 [31] S. Shikano, M. Li, Proc Natl Acad Sci U S A 2003, 100, 5783. 580 https://doi.org/10.1073/pnas.1031748100 581 [32] M. Horak, K. Chang, R. J. Wenthold, J Neurosci 2008, 28, 3500. 582 https://doi.org/10.1523/jneurosci.5239-07.2008 583 [33] T. Inobe, N. Nukina, J Biosci Bioeng 2016, 122, 40. 584 https://doi.org/10.1016/j.jbiosc.2015.12.004 585 [34] J. W. Park, J. R. Reed, L. M. Brignac -Huber, W. L. Backes, Biochem J 2014, 464, 241. 586 https://doi.org/10.1042/bj20140787 587 [35] E. P. Neve, M. Ingelman -Sundberg, Anal Bioanal Chem 2008, 392, 1075. 588 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint https://doi.org/10.1007/s00216-008-2200-z 589 [36] M. H. Kubala, O. Kovtun, K. Alexandrov, B. M. Collins, Protein Sci 2010, 19, 2389. 590 https://doi.org/10.1002/pro.519 591 [37] M. Yanagita, H. Hoshino, K. Nakayama, T. Takeuchi, Endocrinology 1993, 133, 639. 592 https://doi.org/10.1210/endo.133.2.8344203 593 [38] Z. Jia, D. Ragoonanan, K. M. Mahadeo, J. Gill, R. Gorlick, E. Shpal, S. Li, Front Immunol 2022, 594 13, 952231. https://doi.org/10.3389/fimmu.2022.952231 595 [39] L. Rindlisbacher, M. N. Navarro, B. Becher, Nat Rev Immunol 2026. 596 https://doi.org/10.1038/s41577-025-01255-1 597 [40] M. Mansouri, T. Strittmatter, M. Fussenegger, Adv Sci (Weinh) 2019, 6, 1800952. 598 https://doi.org/10.1002/advs.201800952 599 [41] H. Huang, M. J. Dunlop, Curr Opin Microbiol 2026, 89, 102702. 600 https://doi.org/10.1016/j.mib.2025.102702 601 [42] S. B. Chew, E. Harjabrata, C. J. H. Goh, Q. Ong, Biotechnol Adv 2026, 87, 108761. 602 https://doi.org/10.1016/j.biotechadv.2025.108761 603 [43] P. Greengard, Science 2001, 294, 1024. https://doi.org/10.1126/science.294.5544.1024 604 [44] X. Deng, D. Peng, Y. Yao, K. Huang, J. Wang, Z. Ma, J. Fu, Y. Xu, J Diabetes 2024, 16, e13557. 605 https://doi.org/10.1111/1753-0407.13557 606 [45] C. Liu, D. Chu, K. Kalantar -Zadeh, J. George, H. A. Young, G. Liu, Adv Sci (Weinh) 2021, 8, 607 e2004433. https://doi.org/10.1002/advs.202004433 608 [46] M. Mahameed, S. Xue, B. A. Stefanov, G. C. Hamri, M. Fussenegger, Adv Sci (Weinh) 2022, 609 9, e2105619. https://doi.org/10.1002/advs.202105619 610 [47] C. A. Aldrete, C. An, C. C. Call, X. J. Gao, A. E. Vlahos, Curr Opin Biomed Eng 2024, 32. 611 https://doi.org/10.1016/j.cobme.2024.100555 612 [48] F. Cesaratto, A. López -Requena, O. R. Burrone, G. Petris, J Biotechnol 2015, 212, 159. 613 https://doi.org/10.1016/j.jbiotec.2015.08.026 614 [49] M. Mahameed, P. Wang, S. Xue, M. Fussenegger, Nat Commun 2022, 13, 7350. 615 https://doi.org/10.1038/s41467-022-35161-0 616 [50] E. C. Gaynor, S. te Heesen, T. R. Graham, M. Aebi, S. D. Emr, J Cell Biol 1994, 127, 653. 617 https://doi.org/10.1083/jcb.127.3.653 618 [51] M. R. Jackson, T. Nilsson, P. A. Peterson, J Cell Biol 1993, 121, 317. 619 https://doi.org/10.1083/jcb.121.2.317 620 [52] A. P. Teixeira, M. Fussenegger, Adv Sci (Weinh) 2024, 11, e2309088. 621 https://doi.org/10.1002/advs.202309088 622 [53] W. Zhang, A. W. Lohman, Y. Zhuravlova, X. Lu, M. D. Wiens, H. Hoi, S. Yaganoglu, M. A. 623 Mohr, E. N. Kitova, J. S. Klassen, P. Pantazis, R. J. Thompson, R. E. Campbell, Nat Methods 624 2017, 14, 391. https://doi.org/10.1038/nmeth.4222 625 [54] P. Kashyap, S. Bertelli, F. Cao, Y. Kostritskaia, F. Blank, N. A. Srikanth, C. Schlack -Leigers, R. 626 Saleppico, D. Bierhuizen, X. Lu, W. Nickel, R. E. Campbell, A. J. R. Plested, T. Stauber, M. J. 627 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint Taylor, H. Ewers, Nat Methods 2024, 21, 666. https://doi.org/10.1038/s41592-024-02204-x 628 [55] Z. Ma, Z. Fu, B. Hou, X. Zeng, J. Wang, Y. Tan, Y. Jiang, N. Xu, C. Tan, J Am Chem Soc 2025, 629 147, 37005. https://doi.org/10.1021/jacs.5c14442 630 631 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint BLAST: A blue light-assisted secretion toolkit tunable by reversible protein-632 protein interactions 633 Meiying Shen1, 2, Seunghwan Lee1, 2, Kyuye Song1, Mingguang Cui1, Dongmin Lee1, * 634 635 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint 636 Supplementary figure 1 637 Optimization of the a -BLAST prototype using iLID mutants . (a) Schematic of the prototype a -BLAST 638 components. The cargo module (upper) consists of SEAP, a Furin cleavage site, a transmembrane (TM) domain, 639 a single SspB domain, and an RXR retention motif. The regulator module (lower) expresses iLID variants (e.g., 640 N414L) derived from iLID, engineered to modulate dark -state recovery kinetics. (b) Experimental timeline for 641 SEAP secretion assay. Schematic of the 4 -day protocol. Cells were seeded (Day 1), transfected (Day 2), and 642 incubated for 24 h. Blue light stimulation (Day 3) was applied for 24 h using a pulsed duty cycle (2 s ON / 58 s 643 OFF) before supernatant analysis (Day 4). (c) Screening of iLID affinity mutants. Comparison of light -induced 644 SEAP secretion using different iLID variants (N414L, V416L, and N414L/V416L) paired with the prototype 645 cargo (1× SspB). While single mutants exhibited high basal leakage, the double mutant (N414L, V416L) showed 646 a statistically significant fold change (2.0 -fold), identifying it as the optimal candidate for further engineering. 647 Open circles represent individual measurements from three biologically independent samples. Data are presented 648 as means ± S.D. Statistical significance was assessed using one -way ANOVA followed by Tukey’s multiple 649 comparisons test (ns = not significant, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). 650 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint 651 Supplementary Figure 2 652 Benchmarking of chemogenetic BLAST against the RELEASE system . (a) Design of the chemogenetic BLAST 653 (chemo-BLAST) and experimental timeline. Schematic of the plasmid configurations (upper) and the 4 -day 654 experimental schedule (lower). To enable a direct mechanistic comparison independent of light stimulation, the 655 optogenetic modules ( iLID/SspB) were replaced with the chemically inducible rapamycin -binding domains 656 (FKBP/FRB). The cargo (A) contains varying repeats of FRB (1×, 2×, 4×, and 6×), and the regulator (B) consists 657 of FKBP. Rapamycin (1 μM) was used as the inducer. (b) Optimization of FRB valency. Summary graph of SEAP 658 secretion levels with varying FRB repeats. The construct containing 4 × FRB repeats exhibited the most effective 659 switching performance (4.2-fold induction). (c) Optimization of transfection ratios. Using the 4 × FRB construct, 660 SEAP levels were measured across different cargo (A) to regulator (B) ratios. The 1:3 ratio (62.5 ng : 187.5 ng) 661 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint yielded the highest dynamic range (11.3 -fold). (d) Head-to-head kinetic comparison with the RELEASE system. 662 The optimized chemo -BLAST was compared against the RELEASE system, which employs split proteases 663 (TEVp or TVMVp) to cleave a KKMP ER retention motif. Chemo-BLAST induced significant protein secretion 664 starting at 2 h and achieved a higher fold change (7.4 -fold at 24 h) compared to the RELEASE systems (3.2 -fold 665 for TEVp, 3.9-fold for TVMVp), demonstrating superior kinetics and signal-to-noise ratio. Open circles represent 666 individual measurements from three biologically independent samples. Data are presented as means ± S.D. 667 Statistical significance was assessed using one -way ANOVA followed by Tukey’s multiple comparisons test (ns 668 = not significant, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). 669 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint 670 Supplementary Figure 3 671 Impact of the mCherry spacer on d -BLAST basal retention. To evaluate the structural necessity of the mCherry 672 spacer, HEK293T cells were co -transfected with the ER -anchored regulator (P450n -Zdk1-RXR or the non -673 functional Zdk1 -AAA control) and d -BLAST cargo variants either lacking (SEAP -Furin-TM-LOV2) or 674 containing (SEAP-Furin-TM-mCherry-LOV2) the mCherry spacer. The Zdk1-AAA mutant, lacking the retention 675 motif, served as a negative control for constitutive secretion. Notably, the construct lacking mCherry exhibited 676 higher basal secretion in the dark compared to the mCherry -containing variant. This suggests that the mCherry 677 spacer improves retention efficiency by optimizing the spatial distance of the LOV2 binding domain from the ER 678 membrane. Open circles represent individual measurements from three biologically independent samples. Data 679 are presented as means ± S.D. Statistical significance was assessed using one-way ANOVA followed by Tukey’s 680 multiple comparisons test (ns = not significant, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). 681 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint 682 Supplementary Figure 4 683 Validation of BLAST functionality across diverse mammalian cell lineages . Quantification of SEAP secretion 684 levels in the supernatants of various mammalian cell lines following transient transfection with a -BLAST or d-685 BLAST. The assay was performed in (a) HepG2 (human liver cancer), (b) BHK -21 (baby hamster kidney), (c) 686 HeLa (human cervical cancer), and (d) HEK293T (human embryonic kidney) cells. Both a-BLAST and d-BLAST 687 exhibited robust, light -dependent protein secretion with high dynamic ranges across all tested cell types, 688 confirming the broad applicability of the toolkit. Open circles represent individual measurements from three 689 biologically independent samples. Data are presented as means ± S.D. Statistical significance between dark and 690 blue light conditions was assessed using one-way ANOVA followed by Tukey’s multiple comparisons test (****P 691 < 0.0001). 692 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint 693 Supplementary Figure 5 694 Benchmarking of NEW BLAST variants containing optoPASS -derived domains. (a) Design of chimeric NEW 695 BLAST constructs. Schematic representation of the engineered NEW a -BLAST-SEAP and NEW d -BLAST-696 SEAP systems alongside SEAP-optoPASS. To strictly control domain-specific effects, the native Furin cleavage 697 site (FCS) and transmembrane (TM) domain of the original BLAST systems were replaced with the influenza 698 virus-derived sequence (RRRKKR/GL) and the CD4 anchoring domain, respectively, matching the exact 699 configuration used in optoPASS. (b) Summary graph of SEAP secretion levels. While optoPASS showed higher 700 absolute secretion levels, it suffered from high basal leakage in the dark. Consequently, the NEW a-BLAST (10.1-701 fold) and NEW d-BLAST (7.6-fold) variants exhibited superior dynamic ranges compared to optoPASS (3.9-fold). 702 This confirms that the superior signal -to-noise ratio of BLAST is intrinsic to its reversible retention mechanism, 703 rather than dependent on specific structural domains. Open circles represent individual measurements from three 704 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint biologically independent samples. Data are presented as means ± S.D. Statistical significance between dark and 705 blue light conditions was assessed using one-way ANOVA followed by Tukey’s multiple comparisons test (ns = 706 not significant, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). 707 708 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint 709 Supplementary Figure 6 710 High-resolution spatial control of protein secretion via slit -masking. ( a) Schematic of EGFP -fused BLAST 711 constructs. Diagrams of the a-BLAST-EGFP and d-BLAST-EGFP vectors used for spatial patterning experiments. 712 EGFP serves as the fluorescent cargo to visualize intracellular depletion upon secretion. (b) Experimental 713 workflow for slit-guided photopatterning. Schematic illustrating the procedure: Transfected cell monolayers were 714 covered with a black mask containing a 2 mm slit and exposed to blue light (470 nm) for 24 h. (c) and (d) 715 Visualization of spatially restricted secretion. Representative fluorescence images (upper) and corresponding 716 intensity profile plots (lower) for a-BLAST (c) and d-BLAST (d). Left (EGFP): Shows the intracellular cargo. A 717 distinct dark band (fluorescence depletion) is visible in the slit region, indicating light-triggered secretion. Middle 718 (mCherry): Shows the co-transfected mCherry cell marker, which serves as a non -secreted reference. The signal 719 remains constant across the slit, confirming uniform cell density. Right (Merge): Merged image showing the 720 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint spatial contrast between the secreted cargo (green) and the retained marker (red). Scale bar: 2 mm. 721 722 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint 723 Supplementary Figure 7 724 Functional validation of optogenetically secreted IL -12 using a reporter cell assay . ( a) Schematic of the 725 experimental workflow for bioactivity validation. The protocol spans 5 days. HEK293T cells transfected with 726 BLAST-IL-12 were subjected to dark or blue light conditions for 24 h (Day 3). On Day 4, the conditioned media 727 containing secreted IL-12 was harvested and transferred to HEK -Blue™ IL-12 reporter cells (seeded on Day 3). 728 After a 24 h incubation to allow for signal transduction, SEAP activity was quantified (Day 5). (b) Plasmid 729 configurations. Schematics of the a -BLAST-IL-12 (upper) and d -BLAST-IL-12 (lower) constructs used for the 730 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint assay. (c) Illustration of the JAK-STAT signaling pathway in the reporter cells. Binding of secreted IL -12 to the 731 IL-12 receptor complex activates Tyk2/J AK2, leading to STAT4 phosphorylation. Phosphorylated STAT4 732 dimerizes and translocates to the nucleus to induce SEAP expression. (d) Summary graph of SEAP activity 733 induced by the conditioned media. The results confirm that both systems secrete biologically active IL -12 upon 734 blue light stimulation, exhibiting robust fold changes (12.4 -fold for a -BLAST and 11.3 -fold for d -BLAST) 735 compared to the dark control. Data are presented as means ± S.D. Statistical significance was assessed using one-736 way ANOVA followed by Tukey’s multiple comparisons test (****P < 0.0001). 737 738 .CC-BY-NC-ND 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted March 31, 2026. ; https://doi.org/10.64898/2026.03.30.715452doi: bioRxiv preprint