Intra Q-body: an antibody-based fluorogenic probe for intracellular proteins that allows live cell imaging and sorting

Although intracellular biomarkers can be imaged with fluorescent dye(s)-labeled antibodies, the use of such probes for precise imaging of intracellular biomarkers in living cells remains challenging due to background noise from unbound probes. Herein, we describe the development of a conditionally active Fab-type Quenchbody (Q-body) probe derived from a monoclonal antibody (DO-1) with the ability to both target and spatiotemporally visualize intracellular p53 in living cells with low background signal. p53 is a key tumor suppressor and validated biomarker for cancer diagnostics and therapeutics. The Q-body displayed up to 27-fold p53 level-dependent fluorescence enhancement in vitro with a limit of detection of 0.72 nM. In fixed and live cells, 8.3- and 8.4-fold enhancement was respectively observed. Furthermore, we demonstrate live-cell sorting based on p53 expression. This study provides the first evidence of the feasibility and applicability of Q-body probes for the live-cell imaging of intrinsically intracellular proteins and opens a novel avenue for research and diagnostic applications on intracellular target-based live-cell sorting.


Introduction
Visualization of intracellular proteins in living cells is valuable for understanding the biological principles of cellular homeostasis, dysfunction, and protein dynamics. Fluorescent labeling of the proteins of interest (POIs) is crucial. Over the past decades, intrinsic probes such as genetically encoded uorescent proteins (FPs) and self-labeling enzyme tags, such as SNAPtag 1 and HaloTag, 2 have been widely used for the localization of POIs in living cells. However, fusion of bulky FPs or enzyme tags can result in protein misfolding, mistargeting, imprecise localization, or other artifacts. Currently, small epitope tags (such as split-GFP 3 and ultraviolet-photoswitchable protein-uorophore tags 4 ) and modied amino acids (such as unnatural amino acid-mediated bioorthogonal click reactions 5 ) for specic labeling have been developed for precise visualization of endogenous proteins. Nevertheless, the difficulty of controlling overexpression and time-consuming genetic manipulations have limited their biological applications.
Alternatively, extrinsic probes, such as uorescent dye(s)conjugated antibody fragments, which include antigenbinding fragment (Fab), single-chain variable fragment (scFv), and nanobody (Nb), are prominent tools for tracing intrinsic POIs owing to their high specicities and small sizes (12-55 kDa). 6 Numerous high-quality antibody-based probes have been successfully developed and extensively applied to xed and permeabilized cells, along with the cell surface imaging of live cells for many years. 7,8 However, their broad applicability for imaging intracellular analytes in living cells remains limited for two reasons: (i) uorescent probes are always "on", resulting in potentially high background signals, and (ii) these probes are not intrinsically cell membrane permeable. In recent decades, intracellular delivery approaches, such as laser-induced photoporation, 8 cell-squeezing permeabilization, 7,9 electroporation, 10 lipid nanoparticles, 11 and the latest phase-separating peptides, 12 have been investigated to overcome this obstacle. 7,13,14 Several uorescent dye-labeled nanobodies, 6 such as anti-fascin nanobodies, 8 have been developed and delivered to living cells to detect target antigens. In proof-of-concept experiments, these probes were capable of detecting highly expressed targets, albeit with low signal-to-background (S/B) ratios. However, their sensitivity for the detection of less abundant clinically relevant targets is compromised because of an intracellular excess of unbound or non-specic probes. This excess can reduce the S/B ratio and confound the identication of target proteins, since the probes are always "on".
The Quenchbody (Q-body) is a newly developed uorescent immunosensor 15 with the potential to solve this issue. The Qbody is a site-specic uorescent dye(s)-labeled antibody fragment that exhibits antigen-dependent uorescence signal enhancement. The labeled dye(s) are quenched by intrinsic tryptophan residues of the antibody fragment through photoinduced electron transfer (PET) and/or dye-dye interactions. They light up aer a conformational change triggered by antigen-binding. Therefore, unlike conventional antibody probes, Q-bodies have the advantage of antigen-dependent signal generation. To date, more than ten Q-bodies made of Nb, scFv, or Fab against antigens ranging from small molecules (such as narcotics, 16 chemotherapy drugs, 17 and pesticides 18 ) to proteins (such as bone gla protein, 15 claudins, 19 and HER2 20 ) have been developed. Q-body and related technologies have successfully been employed to develop immunosensors for the detection of various antigens in solutions or on the cell surface of living cells. 21 However, to date, this technology has not been validated for intracellular imaging of targets in living cells.
The p53 tumor suppressor is a key transcription factor that is crucial in DNA repair, cell cycle arrest, and apoptosis under cellular stress. 22,23 Normally, p53 is maintained at low levels by mouse double minute 2 homolog (MDM2), an E3 ubiquitinprotein ligase that targets p53 for proteasomal degradation. Cellular stress signals result in p53 stabilization by posttranslational modications, such as phosphorylation of Ser-20 24 and/or inhibition of the MDM2-p53 interaction. 25 Overexpression of mutant and/or wild-type (WT) p53 is oen detected in human cancers and has been used as an important biomarker for cancer diagnostics/therapeutics. [26][27][28][29][30] Several genetically encoded biosensors have been developed for imaging p53. 31,32 However, difficulties in controlling their in situ expression levels, and the need for in situ genetic manipulation limit many of their applications. Therefore, robust visualization of endogenous p53, both WT and mutant in living cells using an extrinsic probe would contribute to fundamental cell biology studies, clinical diagnosis, and cancer therapeutics.
Here, we constructed an anti-p53 Q-body and transferred it into live cells by electroporation to investigate the applicability of Q-body technology in the imaging of intracellular POIs in living cells. A monoclonal antibody (mAb) designated DO-1, 33 which binds to the N-terminal linear epitope ( 20 SDLWKL 25 ) 34 in the transactivation domain of human p53 was used to construct Q-bodies. This mAb has broad applicability, as its epitope is located in a conserved region that is present in both WT and mutant p53, 35 and it binds competitively with MDM2. 25 We expect that Q-bodies delivered intracellularly by electroporation will reside in a default "off" status unless triggered to the "on" status in the presence of p53 (Scheme 1) while the reduction of p53 levels should result in the Q-bodies being turned "off". As expected, in this "proof-of-concept" study, we proved that the Qbody is antigen-dependent uorescence switchable (Off # On) in the native intracellular environment. This enables spatiotemporal visualization of the dynamics of the target antigen in live cells and can be used in intracellular antigen-specic livecell sorting.

Results and discussion
Improvement of anti-p53 scFv (DO-1) secretive expression Although DO-1 scFv showed good expression in mammalian cells, 33 its productivity in Escherichia coli (E. coli) was too low to be used for Q-body preparation. To obtain sufficient functional scFv DO-1 protein using an E. coli expression system, we constructed a combinatorial consensus mutagenesis library 36 and performed phage display selection to screen variants with both antigen-binding ability and high secretory productivity. A combinatorial consensus mutagenesis library containing seven mutation sites (Table S1 †) was constructed for phage display biopanning. 37 Aer three rounds of biopanning, seven variants were obtained (Table S2 †). These variants were cultured independently in E. coli strain TG1, and their medium supernatants were assayed for secreted antibodies using ELISA (enzymelinked immunosorbent assay). As shown in Fig. S1a, † the yields of the seven variants increased 6-7-fold in comparison to the WT, with the C11 variant showing the highest signal. Therefore, the top mutant (C11) and WT were used for further studies.
Molecular dynamics (MD) simulations are widely applied to understand protein stability, having been used for in silico design of recombinant proteins. 38,39 To understand the structural and/or stability changes for the higher ELISA signal of the mutant C11, MD simulations of the G3-WT_ and G3-C11_scFv were performed [G3 represents GGGTG amino acids employed for transpeptidase (sortase A)-mediated uorescence dye labeling]. The structures of scFvs were predicted using Alpha-Fold2 Colab (Fig. S1b †). MD simulations were performed using GROMACS for 22 ns to reach a relatively stable phase for rootmean-square uctuation (RMSF) calculation. The root-meansquare deviation (RMSD) stabilized aer the 10 ns simulation ( Fig. S2 †), and the 10-20 ns simulation data were used for the RMSF analysis ( Fig. S1c †). Fluctuation of several framework regions of the C11 mutant decreased compared to that of the WT_scFv, indicating potentially improved folding and stability of the C11 mutant.

Preparation of Q-bodies and their performance in PBST buffer
WT_scFv or C11_scFv DO-1 with a GGGTG-tag (G3-tag) at the Nterminus of the heavy chain was expressed in the periplasm of E. coli strain BL21(DE3). To prepare Q-bodies, puried scFvs were labeled with TAMRA dye using a mutant of transpeptidase sortase A (SrtA 7+). 40 This transpeptidase enables formation of a new peptide bond between TAMRA-LPET and G3-C11_scFv. Aer labeling, anti-FLAG antibody-coated beads were used to remove free dyes and obtain puried single-labeled Q-bodies (Scheme 1). SDS-PAGE was performed to analyze FLAG-tag puried proteins and Q-bodies. The images of CBB-staining and uorescence of SDS-PAGE indicated that these Q-bodies have been successfully labeled with dyes, while free dyes efficiently removed (Fig. S3b, and S3c †).
Aer obtaining puried WT_and C11_scFv Q-bodies, doseresponse assays were performed in the presence of a human p53 peptide containing the human p53 epitope to determine their maximum response, EC 50 , and limit of detection (LOD). As presented in Fig. 1a and c, C11_scFv (2.0-fold) and WT_scFv (1.9-fold) Q-bodies showed similar antigen-dependent responses. The EC 50 of WT_ and C11_scFv Q-bodies were calculated as 0.78 nM and 1.6 nM, respectively. The LODs for the human p53 peptide were determined to be 0.076 nM for the WT_scFv Q-body and 0.028 nM for the C11_scFv Q-body. These data indicated that the introduction of mutations in C11 did not affect the performance of the corresponding Q-body.
Several previous studies 16,41 have shown that double-labeled Fabs demonstrate a more intensive turn-on response compared to the corresponding single-labeled scFv Q-bodies, as both hydrophobic dye-dye and dye-antibody interactions can favor deep quenching of the dyes. Therefore, we prepared a puried TAMRA double-labeled Fab Q-body using the C11 mutant, hereaer called C11_Fab Q-body (Scheme 1 and Fig. S1e †). To investigate the inuence of dye labeling on antigen-binding activity, the binding kinetics of C11_Fab and its Q-body were evaluated by bio-layer interferometry assay with immobilized biotinylated human p53 peptide. As shown in  C11_Fab Q-body (3.3 nM) were not signicantly different. This indicated minimal perturbation of the antigen-binding affinity due to the dye labeling of C11_Fab.
To investigate the sensitivity, specicity, and antigendependent uorescence response of the C11_Fab Q-body, uorescence changes were measured in the presence of various concentrations of the human p53 peptide or murine p53 peptide (synthesized as shown in Fig. S5 †). The mouse p53 peptide was added as a negative control to investigate the specicity of the C11_Fab Q-body, as a single amino acid change in the mouse epitope abrogates the binding of DO-1. 33 As shown in Fig. 1b, a remarkable dose-dependent increase up to 27-fold in uorescence intensity upon the addition of the human p53 peptide was observed, while negligible uorescence improvement was observed upon adding the mouse p53 peptide. The EC 50 and LOD of the Fab Q-body for the human p53 peptide were calculated as 60 nM and 0.72 nM, respectively (Fig. 1c). Compared with the single-labeled scFv Q-body, the doublelabeled C11_Fab Q-body showed a remarkable improvement in the maximum response, while retaining sub-nanomolar sensitivity (LOD ¼ 0.72 nM). The uorescence spectra of C11_scFv and C11_Fab Q-bodies showed that the doublelabeled Q-body exhibited a higher quenching efficiency in comparison to the single-labeled Q-body ( Fig. S6a and b †). These results indicate that a sensitive and target-specic Q-body with a high antigen-dependent turn-on signal was developed.
To understand the quenching mechanism of the C11_Fab Qbody, the absorbance spectra of self-quenched and antigenactivated forms of Q-bodies were measured (Fig. S7a †). Compared with the free uorophore (TAMRA-LPETGG), the selfquenched Q-body showed two peaks with maximum absorbance at 520 nm (H-dimer 42 ) and 555 nm (monomer), respectively. The addition of the p53 peptide led to a 520 nm peak shi and the maximum absorbance at 555 nm increased (Table S5 †). The results suggested that apart from Trp residues, 15 H-dimer formation also contributed to the uorescent quenching in the C11_Fab Q-body. Based on the absorbance spectra of the C11_Fab Q-body (Fig. S7a †), the uorescent dye to protein ratio (F/P) was calculated as 190% for the self-quenched form of Qbody and 218% for the activated form of Q-body (Table S5 †). This indicated that C11_Fab was efficiently labeled with TAMRA. The quantum yields of self-quenched and activated forms of Q-bodies with excitation at 500 nm were determined as 0.051 and 0.45, respectively ( Fig. S7a and S7b, S14 and Table S6 †). The quantum yields increased 8.8-fold aer adding antigen. This is slightly higher than the uorescence changes (7.3-fold) shown in Fig. S7b. † The difference might result from the error during quantum yield detection.
One-step imaging of p53 in xed human cancer cells Aer obtaining a high-performance C11_Fab Q-body, we rst applied it to the visualization of p53 in xed cells as a direct immunouorescence (IF) probe in IF imaging in a more timesaving manner. We expected that the Q-body would be in an "on" status only in the presence of p53. To verify this, all IF assays were performed without blocking (before probe treatment to prevent non-specic binding of probes) and washing (aer probe treatment to remove unbound probes). A human colon cancer cell line, HCT116 p53 +/+ (expressing WT p53) and its subtype HCT116 p53 À/À 43-45 were used to verify our assumption. Nutlin-3a, an MDM2 inhibitor, was used to increase the p53 levels. 46 An outline of this experiment is shown in Fig. 2a. As indicated in Fig. 2b and c, negligible signals were observed in HCT116 p53 À/À cells stained with C11_Fab Q-body irrespective of nutlin-3a treatment. Notably, the uorescence signals in HCT116 p53 +/+ cells were signicantly higher than those in p53 À/À cells (8.3-fold) upon nutlin-3a treatment. Fluorescence signals were observed in the nucleus, where p53 predominantly resides. However, when the cells were stained with C11_scFv-TAMRA (a representative of a traditional IF probe that shows no signal switching function because the TAMRA label was at the C-terminus of the light chain far from the antigen-binding site), a high background signal was observed even in p53 À/À cells. The difference in the mean F.I. of nuclei between p53 +/+ and p53 À/À cells stained with C11_scFv-TAMRA (3.8-fold) was lower than that of the C11_Fab Q-body (8.3-fold). These results verify that the C11_Fab Q-body displays an antigen-dependent signal on xed human colon cancer cells and shows a higher S/B ratio than the traditional IF probe (C11_scFv-TAMRA).
We further investigated the efficacy of the C11_Fab Q-body in the detection of mutant p53 in xed samples. SK-BR-3 human breast cancer cells (p53, pR175H), and WiDr human colon adenocarcinoma cells (p53, pR273H) were used in the IF assay. As shown in Fig. S8, † in SK-BR-3 and WiDr cells that were treated with nutlin-3a, the observed uorescence signal showed only a slight increase in comparison to the cells without nutlin-3a treatment. These results are consistent with previous publications stating that nutlin-3a does not affects p53 levels in cells harboring mutant p53 47 and are compatible with the western blot results (Fig. S9 †). These results demonstrated that the C11_Fab Q-body, a new generation of IF probes, could be successfully applied to specically visualize both human WT and mutant p53 in a timesaving manner.

Visualization of p53 dynamics in living cells at nanomolar concentrations
Currently, almost all the extrinsic antibody-based probes are always in an "on" state, which could result in low S/B ratio in the visualization of intracellular targets. This is due to excess and/or non-specically bound probes that generate uorescence irrespective of target engagement. Furthermore, it is extremely difficult to remove these probes by washing, which is normally used to remove excess probes in IF staining assays because most probes are impermeable to the cell membrane. Having established that the C11_Fab Q-body displays antigen-dependent signal turn-on in xed cell imaging, we next evaluated its applicability to live-cell imaging.
As Q-bodies are membrane-impermeable, electroporation was employed to deliver them intracellularly. HCT116 (p53, WT or null), SK-BR-3 (p53, pR175H), and WiDr (p53, pR273H) cells were incubated for 3-4 h to recover aer electroporation with 200 nM C11_Fab Q-body. The expected mechanism of this Qbody in living cells is shown in Fig. 3a. Imaging under washfree conditions showed that uorescence signals colocalizing with nuclear staining (Hoechst) were only observed in p53positive cell lines HCT116 p53 +/+ , SK-BR-3, and WiDr. Conversely, negligible signals were observed in p53-negative (HCT116 p53 À/À ) cells (Fig. 3b). The mean F.I. of the nucleus in p53-positive cells was 8.4-fold higher than that in p53negative cells (Fig. 3c). Co-transfection of Q-body and p53 peptide into HCT116 p53 À/À cells resulted in signals spread over whole cells (Fig. S10 †). The result proved on-target engagement of peptides in p53 À/À cells. However, the signal was spread out as the peptide does not localize to any particular region. Notably, traditional probes can give a similar phenotype in the absence of the target (i.e., false positive). These results indicate that Q-body shows antigen-dependent uorescence enhancement in the complex intracellular environment of live cells.
To evaluate whether the C11_Fab Q-body is stable for a longer period and shows p53 level-dependent uorescence changes in living cells, time-lapse confocal microscopy was performed. To generate HCT116 p53 +/+ cells with different timedependent levels of p53, the cells were treated with nutlin-3a rst for 16 h to increase the p53 levels followed by treatment with cisplatin for 9 h to reduce p53 levels. The uctuations of p53 levels under these treatment conditions were conrmed by western blot assay. As shown in Fig. S11, † compared with the non-treated cells, the p53 levels increased in the cells treated Fig. 2 Wash-free visualization of p53 in fixed human cancer cells using C11_Fab Q-body. HCT116 p53 À/À or p53 +/+ are human colon cancer cell lines. A representative of a traditional direct immunofluorescence probe (C11_scFv-TAMRA) was applied as a control to compare the performance of the C11_Fab Q-body. Nutlin-3a, an MDM2 inhibitor, was used to stabilize the p53 protein and improve p53 levels. (a) Schematic of wash-free fixed cell staining assay by C11_Fab Q-body and C11_ scFv-TAMRA in a time-saving manner. (b) Representative images showing 40 nM C11_Fab Q-body or C11_scFv-TAMRA staining of p53 in HCT116 p53 À/À or p53 +/+ cells after being treated with 12 mM nutlin-3a or 0.06% ethanol for 16 h. TAMRA (red), C11_Fab Q-body or C11_scFv-TAMRA; Hoechst (blue), 1 mg mL À1 Hoechst 33 342; Merge, overlapped TAMRA with Hoechst; DIC, differential interference contrast. Scale bar, 10 mm. (c) Box plot of mean TAMRA intensities in the nucleus subtracted to minimum fluorescence intensities (F.I.) of TAMRA channel. The median F.I. was used to calculate their fluorescence changes between groups. The first four boxes: one-way ANOVA test; the last two boxes: Welch's t-test. ****p < 0.0001; n.s., not significant; from left to right, n ¼ 39, 36, 78, 53, 83, 79 cells. For the box plot, the white line indicates the median, the box indicates 25-75% range, whiskers indicate 1.5 interquartile range, and the black dot indicates outliers.
with nutlin-3a for 25 h. p53 levels decreased in the cells which were rst exposed to nutlin-3a for 16 h then treated with cisplatin for 9 h. Aerwards, time-lapse imaging assay was performed to observe the dynamics of p53 using the C11_Fab Qbody. We hypothesized that nutlin-3a will increase p53 levels to turn on C11_Fab Q-body uorescence, while cisplatin can reduce p53 to turn it off. In this experiment, the transfected HCT116 p53 +/+ cells were incubated for approximately 9 h to allow cell recovery and attachment (Fig. 4a). Then time-lapse imaging commenced under different treatment conditions. As shown in Fig. 4b and c, and ESI Video, † under the treatment of nutlin-3a, the uorescence signals gradually increased and reached a plateau (2.6-fold) with a slight decrease upon 10 h treatment. The slightly decreased signal might be due to the reduction of DO-1 epitope of p53, and/or the degradation-or cell-division-derived Q-body decreases. When the medium was washed off and replaced with cisplatin aer 16 h treatment with nutlin-3a, the signal reduced from 2.4-fold to 1.0-fold within 3 h and kept at low levels. In the non-treated, but electroporated cells, the uorescent signal was high at the beginning and then reduced time-dependently (Fig. 4c), which likely reects increased p53 levels induced by electroporation stress that gradually revert to normal aer time-dependent recovery. 48 Overall, these observations were consistent with the results of the corresponding xed cell Q-body staining assay (Fig. S12 †). In HCT116 p53 À/À cells, the signals were almost unchanged and maintained at a low level (Fig. S13 †). These data indicate that the Q-body is stable enough for long-term live-cell imaging and enables visualization of p53 dynamics in live cells. These pilot studies demonstrate that Q-body technology can be utilized to localize intracellular POIs in viable cells, but also allows visualization of the dynamic changes in intracellular targets in living cells.

Intracellular antigen-specic live-cell sorting using Q-body
To investigate the feasibility of applying this technology to intracellular antigen-specic live-cell sorting, a proof-of-concept experiment was performed. Three cell groups, including HCT116 p53 +/+ cells transfected with or without C11_Fab Qbody, and HCT116 p53 À/À cells transfected with C11_Fab Qbody, were used for ow cytometric analysis to evaluate the cell distributions of each group. As shown in Fig. 5a, a clear F.I. increment in Q-body-transfected HCT116 p53 +/+ cells was observed in comparison with the other two groups. Next, a mixture of the positive (p53 +/+ + C11_Fab Q-body) and negative (p53 À/À + C11_Fab Q-body) cells at a 6.9% proportion was used for live-cell sorting. Aer sorting, the ratio of HCT116 p53 +/+ cells transfected with the Fab Q-body increased from 6.9% to 94% (Fig. 5b), which corresponds to approximately 14-fold enrichment of p53 +/+ cells. Additionally, the F.I. changes before and aer sorting showed a signicant increase (4.6-fold) (Fig. 5c). Taken together, these results indicate that the application of Q-body technology in intracellular antigen-specic live-cell sorting is possible. Fig. 4 Time-lapse observation of p53 dynamics in HCT116 p53 +/+ using C11_Fab Q-body. (a) Schematics of time-lapse imaging assay. The cells were incubated for 9 h to allow cell recovery and adherence following electroporation. After the addition of 12 mM nutlin-3a, time-lapsed observation commenced. After 16 h, the old medium containing nutlin-3a was removed and gently washed twice with fresh medium, then treated with fresh medium containing 12 mM nutlin-3a or 6 mM cisplatin. Cisplatin is an anti-cancer chemotherapy drug. (b) Representative images of HCT116 p53 +/+ cells transfected with C11_Fab Q-body after being treated with nutlin-3a, cisplatin, or non-treated as indicated. Red color, TAMRA channel. DIC, differential interference contrast. Scale bar, 20 mm. (c) Time-dependent TAMRA intensity changes in nuclei. Normalized F.I. of the nucleus, the mean fluorescence intensity (F.I.) of nucleus areas subtracted to that in the blank area, then normalized to the F.I. of the start point (9 h). Data are presented as the mean AE SEM of 18 cells. All images were acquired as Z-stacks (five sections at 2.5 mm intervals), and the most representative Z-stack images were presented and used for F.I. analysis.
Fulllment of intracellular antigen-specic live-cell sorting is valuable for improving cell therapies. However, to the best of our knowledge, current approaches used to isolate intracellular antigen-specic live cells are not available. For instance, the isolation of regulatory T cells (Tregs), which are crucial in the treatment of human diseases that include graversus-host disease and autoimmune disease 49 is based on cell surface markers. Although it is clear that Foxp3 (an intracellular protein) is the best and most specic marker of Tregs, people still use cell surface markers (such as CD25, CD4, and CD127) for Treg isolation 50 because current technologies are not able to isolate viable Tregs using Foxp3 directly. In this study, we provide the rst demonstration that intracellular antigen-specic live-cell sorting using Q-body technology is possible.

Conclusions
In general, imaging of intracellular POIs by extrinsic probes remains a challenge due to typically low levels of endogenous proteins and unbound or unspecic bound probes that result in a low S/B ratio. In particular, these probes are unable to distinguish cells based on uorescence intensities, as they display equivalent uorescence even in the absence of analyte binding. Here, we developed an anti-p53 C11_Fab Q-body that detects human p53 in both xed and live cells which allows visualization of the long-term dynamics of human p53 in living cells. Crucially, in our proof-of-concept study, the p53 Fab Qbody was successfully used to isolate p53 positive cells from a mixture of p53 positive and negative cells. As most antibodies can be engineered to construct Q-bodies, Q-bodies targeted to many other intracellular biomarkers could be developed to overcome the challenges of live-cell imaging of intracellular POIs.
Transcription factors have been regarded as targets for cancer therapy and are used as biomarkers for the development of anti-cancer drugs. 51 However, the intracellular localization of transcription factors and lack of useful biosensors hinder functional studies and drug development in live cells. Here, we proved that Q-body technology can overcome these challenges and that the application of Q-body technology to live-cell imaging or sorting based on intracellular biomarkers is possible. This approach opens a new avenue for live-cell imaging and intracellular target-based live-cell sorting.

Data availability
The authors conrm that the data supporting the ndings of this study are available within the article and its ESI. † Raw data that support the ndings of this study are available from the corresponding author, upon reasonable request.

Author contributions
Y. D. carried out and designed the experiments and wrote the manuscript dra. H. U. and F. J. G. conceived the study, designed the experiments, and edited the manuscript. Y. S., B. Z., T. K., H. K., and F. J. G. supported performing experiments, analysing data, and editing manuscript. B. Z. performed MD simulation and analysis. All authors have given approval to the nal version of the manuscript.

Conflicts of interest
There are no conicts to declare. The ratio of HCT116 p53 +/+ cells transfected with C11_Fab Q-body before and after sorting. Their ratio was determined based on the mean F.I. of cell nuclei from their fluorescence images. The cells with mean F.I. below the red dotted line (F.I. < 200) in (c) were counted as negative cells. Representative photos before and after sorting are presented. Merge, TAMRA (red) + Hoechst (blue) + Bright field (gray). Scale bar, 20 mm. (c) Box plot of mean TAMRA intensities in nuclei subtracted to that in the blank areas (before sorting, n ¼ 535; after sorting, n ¼ 246). The median F.I. was used to calculate their fluorescence changes between groups. Welch's t-test. ****p < 0.0001. For the box plot, the white line indicates the median, the box indicates 25-75% range, whiskers indicate 1.5 interquartile range, and circles indicate data distributions.