Next generation synthetic memory via intercepting recombinase function

Here we present a technology to facilitate synthetic memory in a living system via repurposing Transcriptional Programming (i.e., our decision-making technology) parts, to regulate (intercept) recombinase function post-translation. We show that interception synthetic memory can facilitate programmable loss-of-function via site-specific deletion, programmable gain-of-function by way of site-specific inversion, and synthetic memory operations with nested Boolean logical operations. We can expand interception synthetic memory capacity more than 5-fold for a single recombinase, with reconfiguration specificity for multiple sites in parallel. Interception synthetic memory is ~10-times faster than previous generations of recombinase-based memory. We posit that the faster recombination speed of our next-generation memory technology is due to the post-translational regulation of recombinase function. This iteration of synthetic memory is complementary to decision-making via Transcriptional Programming – thus can be used to develop intelligent synthetic biological systems for myriad applications.

Supplementary Figure 1: Illustrations and iconography for deletion and inversion synthetic memory.a (left) A granular description of the set if A118 recombinase attachment sites in the aligned (deletion) configuration.(right) The iconography for the two aligned orientations that result in deletion.b (top) A detailed description of anti-aligned (inversion) attachment sites for the A118 recombinase, and below is the iconography for the two anti-aligned orientations that result in inversion.Note: The icon for the recombinase is given as a monomer.Supplementary Fig. 2 Supplementary Figure 2: Definition and analysis of recombinase attachment half-site omission for A118, TP901, Int2, and Int3.Each recombinase recognizes and recombines two attachment sites, attB and attP.Each attachment site contains a central conserved region, shown in bold grey, such that attB and attP must be identical when aligned for deletion or complementary when aligned for inversion 1 (also see Supplementary Figure 1).Attachment site sequences are shown in blue, and central conserved regions are shown in bold grey.a Symbols corresponding to recombinase half-attachment-site sequence omissions for A118.For example, A118 B1 corresponds to a truncated attachment site lacking the indicated sequence, i.e.AAACGCAAAGAGGGAACTAAACACTT.In other words, the truncated triangle symbol refers schematically to the segment of the attachment site sequence that has been omitted.The DNA sequence on the original reporter construct (pSK001) upstream of attB is now present in place of the half-site B1.At right, data corresponding to the half-site omission experiment described in Fig. 2  Supplementary Fig. 3 Supplementary Figure 3: Definition and analysis of recombinase attachment half-site omission for Int5, Int8, Int12, and Bxb1.Each recombinase recognizes and recombines two attachment sites, attB and attP.Each attachment site contains a central conserved region, shown in bold grey, such that attB and attP must be identical when aligned for deletion or complementary when aligned for inversion 1 (also see Supplementary Figure 1).Supplementary Figure 5: Transcriptional repressor performance abstraction and sequence similarity between each operator substituted attP site relative to wildtype attP.At top left, the performance card for a general repressor (X + ) and an abstraction of its performance metrics to a logical BUFFER operation is shown.At top right, the performance card of a general anti-repressor (X A ) and an abstraction of its performance metrics to a logical NOT operation is shown.The metrology for a given single-INPUT single-OUTPUT (SISO) operation, we can model the induction profile for an experimentally verified SISO BUFFER operation via a coarse-grained binding function defined as where s is a constant representing the maximum fluorescence -relative to basal expression of the OFF-state, L + (I) is a coarse-grained Hill function that can assume a value of 0 or 1, and ε represents fluorescence in the absence of inducer -i.e., the OFF-state.
To model the performance of a given SISO NOT gate we used an analogous coarse-grained binding function -though for anti-repression -defined as

(-)
(+)  where s is a constant representing the maximum fluorescence, minus ligand -relative to basal expression of the OFFstate, L A (I) is a coarse-grained antithetical Hill-function for anti-repression where 0 INPUT corresponds to the ONstate, and 1 INPUT corresponds to the OFF-state, and ε represents fluorescence in the presents of inducer -i.e., the OFF-state.This set of models was used to study performance prediction in Milner et al. 2 .At bottom left, the front of a general performance card is shown detailing the plasmids, PROXIMAL operator position, and chassis used to measure the TF performance data.At bottom right, the wildtype A118 attP site shown in blue (with central conserved region shown in grey) is compared to the attP sites used for interception with operators substituted at position P+1 (also see Fig. 2j).Nucleotides that are altered by the inclusion of the given operator are shown in red, highlighting the alteration to the attP site incurred by substituting each operator.The sequence similarity between each operator substituted attP site and wildtype is given as Match %, with lower scores indicating lower similarity.Supplementary Fig. 6 Supplementary Figure 6: Transcriptional repressor (E + variants) performance compared to repressor interception.Comparing interception performance of E + variants to transcriptional repression performance of those same E + variants.At top left, assay data is shown for intercepted (minus ligand) circuits versus deprotected (induced) circuits using the repressor E + across six different DNA-binding domain/operator pairs.At top right, the performance card for a general repressor (X + ) and an abstraction of its performance metrics to a logical BUFFER operation is shown -detailed description given in Supplementary Figure 5. Below, transcriptional repression performance cards for each of the E + variants tested for interception performance are given.Source data are provided as a Source Data file.Data represents the average of n = 6 biological replicates.Error bars correspond to the SEM of these measurements Source data are provided as a Source Data file.Error bars correspond to the SEM of these measurements.Welch's ttest between (-) inducer and (+) inducer groups was conducted were ****: p-value < 0.0001, *: p-value < 0.05, ns: pvalue > 0.05.Comparing interception performance of R + variants to transcriptional repression performance of those same R + variants.At top left, assay data is shown for intercepted (minus ligand) circuits versus deprotected (induced) circuits using the repressor R + across six different DNA-binding domain/operator pairs.At top right, the performance card for a general repressor (X + ) and an abstraction of its performance metrics to a logical BUFFER operation is shown -detailed description given in Supplementary Figure 5. Below, transcriptional repression performance cards for each of the R + variants tested for interception performance are given.Source data are provided as a Source Data file.Data represents the average of n = 6 biological replicates.Error bars correspond to the SEM of these measurements Source data are provided as a Source Data file.Welch's t-test between (-) inducer and (+) inducer groups was conducted were ****: p-value < 0.0001, *: p-value < 0.05, ns: p-value > 0.05.Supplementary Fig. 8 Supplementary Figure 8: Transcriptional repressor (F + variants) performance compared to repressor interception.Comparing interception performance of F + variants to transcriptional repression performance of those same F + variants.At top left, assay data is shown for intercepted (minus ligand) circuits versus deprotected (induced) circuits using the repressor F + across six different DNA-binding domain/operator pairs.At top right, the performance card for a general repressor (X + ) and an abstraction of its performance metrics to a logical BUFFER operation is shown -detailed description given in Supplementary Figure 5. Below, transcriptional repression performance cards for each of the F + variants tested for interception performance are given.Source data are provided as a Source Data file.Data represents the average of n = 6 biological replicates.Error bars correspond to the SEM of these measurements Source data are provided as a Source Data file.Welch's t-test between (-) inducer and (+) inducer groups was conducted were ****: p-value < 0.0001, *: p-value < 0.05, ns: p-value > 0.05.Supplementary Fig. 9 Supplementary Figure 9: Transcriptional repressor (G + variants) performance compared to repressor interception.Comparing interception performance of G + variants to transcriptional repression performance of those same G + variants.At top left, assay data is shown for intercepted (minus ligand) circuits versus deprotected (induced) circuits using the repressor G + across six different DNA-binding domain/operator pairs.At top right, the performance card for a general repressor (X + ) and an abstraction of its performance metrics to a logical BUFFER operation is shown -detailed description given in Supplementary Figure 5. Below, transcriptional repression performance cards for each of the G + variants tested for interception performance are given.Source data are provided as a Source Data file.Data represents the average of n = 6 biological replicates.Error bars correspond to the SEM of these measurements Source data are provided as a Source Data file.Welch's t-test between (-) inducer and (+) inducer groups was conducted were ****: p-value < 0.0001, *: p-value < 0.05, ns: p-value > 0.05.Supplementary Fig. 10 Supplementary Figure 10: Transcriptional repressor (I + variants) performance compared to repressor interception.Comparing interception performance of I + variants to transcriptional repression performance of those same I + variants.At top left, assay data is shown for intercepted (minus ligand) circuits versus deprotected (induced) circuits using the repressor I + across six different DNA-binding domain/operator pairs.At top right, the performance card for a general repressor (X + ) and an abstraction of its performance metrics to a logical BUFFER operation is shown -detailed description given in Supplementary Figure 5. Below, transcriptional repression performance cards for each of the I + variants tested for interception performance are given.Source data are provided as a Source Data file.Data represents the average of n = 6 biological replicates.Error bars correspond to the SEM of these measurements Source data are provided as a Source Data file.Welch's t-test between (-) inducer and (+) inducer groups was conducted were ****: p-value < 0.0001, *: p-value < 0.05, ns: p-value > 0.05.Supplementary Fig. 14 Supplementary Figure 14: Transcriptional anti-repressor (I A variants) performance compared to antirepressor interception.Comparing interception performance of I A(5) variants to transcriptional anti-repression performance of those same I A(5) variants.At top left, assay data is shown for intercepted (induced) circuits versus deprotected (minus ligand) circuits using the anti-repressor I A(5) across six different DNA-binding domain/operator pairs.At top right, the performance card for a general anti-repressor (X A ) and an abstraction of its performance metrics to a logical NOT operation is shown -detailed description given in Supplementary Figure 5. Below, transcriptional anti-repression performance cards for each of the I A(5) variants tested for interception performance are given.Source data are provided as a Source Data file.Data represents the average of n = 6 biological replicates.Error bars correspond to the SEM of these measurements Source data are provided as a Source Data file.Welch's t-test between (-) inducer and (+) inducer groups was conducted were ****: p-value < 0.0001, *: p-value < 0.05, ns: p-value > 0.05.     1) variants to transcriptional anti-repression performance of those same R A (1) variants.At top left, assay data is shown for intercepted (induced) circuits versus deprotected (minus ligand) circuits using the anti-repressor R A (1) across six different DNA-binding domain/operator pairs.At top right, the performance card for a general anti-repressor (X A ) and an abstraction of its performance metrics to a logical NOT operation is shown -detailed description given in Supplementary Figure 5. Below, transcriptional anti-repression performance cards for each of the R A (1) variants tested for interception performance are given.Source data are provided as a Source Data file.Data represents the average of n = 6 biological replicates.Error bars correspond to the SEM of these measurements Source data are provided as a Source Data file.Welch's t-test between (-) inducer and (+) inducer groups was conducted were ****: p-value < 0.0001, *: p-value < 0.05, ns: p-value > 0.05.Supplementary Fig. 16 Supplementary Figure 16: Transcriptional anti-repressor (F A variants) performance compared to antirepressor interception.Comparing interception performance of F A (1) variants to transcriptional anti-repression performance of those same F A (1) variants.At top left, assay data is shown for intercepted (induced) circuits versus deprotected (minus ligand) circuits using the anti-repressor F A (1) across six different DNA-binding domain/operator pairs.At top right, the performance card for a general anti-repressor (X A ) and an abstraction of its performance metrics to a logical NOT operation is shown -detailed description given in Supplementary Figure 5. Below, transcriptional anti-repression performance cards for each of the F A (1) variants tested for interception performance are given.Source data are provided as a Source Data file.Data represents the average of n = 6 biological replicates.Error bars correspond to the SEM of these measurements Source data are provided as a Source Data file.Welch's t-test between (-) inducer and (+) inducer groups was conducted were ****: p-value < 0.0001, *: p-value < 0.05, ns: p-value > 0.05.    1) variants to transcriptional anti-repression performance of those same S A(1) variants.At top left, assay data is shown for intercepted (induced) circuits versus deprotected (minus ligand) circuits using the anti-repressor S A(1) across six different DNA-binding domain/operator pairs.At top right, the performance card for a general anti-repressor (X A ) and an abstraction of its performance metrics to a logical NOT operation is shown -detailed description given in Supplementary Figure 5. Below, transcriptional anti-repression performance cards for each of the S A (1) variants tested for interception performance are given.Source data are provided as a Source Data file.Data represents the average of n = 6 biological replicates.Error bars correspond to the SEM of these measurements Source data are provided as a Source Data file.Welch's t-test between (-) inducer and (+) inducer groups was conducted were ****: p-value < 0.0001, *: p-value < 0.05, ns: p-value > 0.05.Supplementary Fig. 18 Supplementary Figure 18: Transcriptional anti-repressor (P A variants) performance compared to antirepressor interception.Comparing interception performance of P A variants to transcriptional anti-repression performance of those same P A variants.At top left, assay data is shown for intercepted (induced) circuits versus deprotected (minus ligand) circuits using the anti-repressor P A across six different DNA-binding domain/operator pairs.At top right, the performance card for a general anti-repressor (X A ) and an abstraction of its performance metrics to a logical NOT operation is shown -detailed description given in Supplementary Figure 5. Below, transcriptional anti-repression performance cards for each of the P A variants tested for interception performance are given.Source data are provided as a Source Data file.Data represents the average of n = 6 biological replicates.Error bars correspond to the SEM of these measurements Source data are provided as a Source Data file.Welch's t-test between (-) inducer and (+) inducer groups was conducted were ****: p-value < 0.0001, *: p-value < 0.05, ns: p-value > 0.05.c-d).In the first iteration of the 2-INPUT circuit, the nested AND logic with interception was functional, requiring 2-INPUTs to produce GFPunless the attachment site was deprotected upon the addition of cellobiose see a-b.Moreover, in the absence of recombinase the circuit performed as a simple AND gate -independent of the presence of cellobiose (see b inset blue box).Likewise, the nested NOR iteration of the interception circuit was also functional -with a synonymous control feature regulating the deletion memory operation (see c-d).Namely, the protected circuit only produced GFP in the absence of both IPTG and ribose.However, upon the addition of cellobiose the circuit was deleted (see c-d). a Genetic schematic and mechanism of a transcriptional AND gate nested within a memory interception circuit (coded for deletion), cognate to the A118 recombinase.28, and the constructs are subjected to equivalent experimental conditions on day 1.Following day 1, the cells are passaged as described in Supplementary Figure 28; however, no inducers are added, preserving the memory state written on day 1.Source data are provided as a Source Data file.Data represents the average of n = 6 biological replicates.Error bars correspond to the SEM of these measurements.

TF ≡ X
is shown.b-d The attachment site sequences and corresponding half-site sequence omissions are shown with the relevant assay data for the recombinases A118, TP901, Int2, Int3, and Int5.Source data are provided as a Source Data file.Data in a-d represent the average of n = 6 biological replicates.Error bars correspond to the SEM of these measurements.

Supplementary
Attachment site sequences are shown in blue, and central conserved regions are shown in bold grey.a-d The attachment site sequences and corresponding half-site sequence omissions are shown with the relevant assay data for the recombinases Int5, Int8, Int12, and Bxb1.Source data are provided as a Source Data file.Data in a-d represent the average of n = 6 biological replicates.Error bars correspond to the SEM of these measurements.Figure 4: Qualitative genotype of 8 recombinases paired with cognate interception deletion circuits.a Schematic of an interception deletion circuit, in which a constitutive promoter, ribozyme, RBS, and fluorescent protein GFP is nested between an aligned attB and attP pair (STATE 1).A recombinase matched to those att sites catalyzes recombination between attB and attP, resulting in deletion of GFP expression (STATE 2).An O ttg operator is included at position P+1.b PCR primers that bind outside the reporter circuit can be used with gel electrophoresis to differentiate between an intact GFP circuit (state 1: 1,300 bp) and the recombined circuit (state 2: 200 bp).c Gel electrophoresis of colony PCR of the reporter circuit for cells transformed with only reporter and no recombinase (labeled "S1") versus cells transformed with reporter and recombinase (shown as "+ recombinase symbol").This gel data is representative of three repeats of this experiment, each with similar results.The 100 bp DNA Ladder (NEB #N3271) is used as the reference.Source data are provided as a Source Data file.

Supplementary Figure 11 :Supplementary Figure 12 :Supplementary
Detailed sequences for substituted attP sites.Nucleotides that are altered by the substitution with operator DNA are shown in red, highlighting the alteration to the attP site incurred by substituting with a given operator.Flow cytometry of select interception circuits.a Inset at left is a schematic summarizing the genetic construct used to assess A118 recombinase interception with variable repressors directed at different operators placed in the P+1 position.Inset at right is the subset of modular TF components used for flow cytometry analysis of interception performance.b-j Flow cytometry analysis of different TFs intercepting A118 via different operators in the P+1 position.The distribution of cells in the two possible states of fluorescent protein expression after growth in medium without inducers is shown.Source data are provided as a Source Data file.Data in b-j represent the average of n = 6 biological replicates.Error bars correspond to the SEM of these measurements.Figure 13: Ribosome binding site (RBS) tuning of A118 for three R + ADR.(top) A schematic summarizing the mechanism and genetic construct (deletion circuit) used to assess A118 recombinase interception with variable repressors directed at different operators placed in the P+1 position.(bottom left) Assay data collected under moderate conditions (see Supplementary Note 2) for R + , also given in Fig. 3. (bottom right) Assay data for three ADR/operator pairs, GKR/O gac , TAN/O tta , and HTK/O ctt following A118 expression modification via a prescribed RBS library.Source data are provided as a Source Data file.Data represents the average of n = 6 biological replicates.Error bars correspond to the SEM of these measurements.

Supplementary
Figure 15: Transcriptional anti-repressor (R A variants) performance compared to antirepressor interception.Comparing interception performance of R A(

Supplementary Figure 17 :
Transcriptional anti-repressor (S A variants) performance compared to antirepressor interception.Comparing interception performance of S A(

22 :
Interception synthetic memory with nested AND / NOR Boolean Logic.We constructed interception memory circuits with nested 2-INPUT logic.Namely, we constructed a circuit with a nested AND gate (see a-b) and a separate circuit with a nested NOR gate (see

SupplementarySupplementary
A118 and the regulating TFs (I + YQR, R + YQR, and E + HQN) are expressed constitutively.I + YQR and R + YQR regulate GFP expression by binding to an O gtg operator in the promoter's core position, and E + HQN intercepts A118 function at an O ttg promoter placed at the P+1 position.From left to right, the response to different inducers (INPUTs) is shown.b Assay data for the circuit shown in a.Cells exposed to cellobiose (i.e., post deletion) are boxed in red.Post deletion, cells were diluted 1:200 and grown in fresh minimal media for 20 additional hours with and without the transcriptional logic INPUTS IPTG and ribose to demonstrate deletion memory.Data in the red box labeled DAY 2 shows the resulting phenotypes.To the right (of the red box) boxed in blue is the control data for this circuit transformed with the TF-expression plasmid and without the recombinase expression plasmid.c Genetic schematic of a transcriptional NOR gate nested within a memory interception circuit (coded for deletion), cognate to the A118 recombinase.A118 and the regulating TFs (I A(9) YQR, R A(2) YQR, and E + HQN) are expressed constitutively.I A(9) YQR and R A(2) YQR regulate GFP expression by binding to an O gtg operator in the promoter's core position, and E + HQN intercepts A118 function at an O ttg promoter placed at the P+1 position.d Assay data for the circuit shown in c.Cells exposed to cellobiose (i.e., post deletion) are boxed in red.Post deletion, cells were diluted 1:200 and grown in fresh minimal media for 20 additional hours with and without the transcriptional logic INPUTS IPTG and ribose to demonstrate deletion memory.Data in the red box shows the resulting phenotypes.To the right (of the red box) boxed in blue is the control data for this circuit transformed with the TF-expression plasmid and without the recombinase expression plasmid.Source data are provided as a Source Data file.Data in b and d represent the average of n = 6 biological replicates.Error bars correspond to the SEM of these measurements.ttg (attP) @P-18 TP901 + O ttg (attP) @P-21 TTAGTTCCTTTTTGAGCGCTCAAAAAAGAAGAAGAAACGAGAAACTAAA A118 + O ttg (attP) @P-15 A118 + O ttg (attP) @P-18 A118 + O ttg (attP) @Pttg (attP) @P-21 TP901 + O ttg (attP) @P-16 TP901 + O ttg (attP) @P+3 Figure 25: Double-layer deletion circuit with two orthogonal attachment sites.Recombinase activities are measured with varying the central conserved region of the attachment sites.a Interception was tested on reporter circuits shown in Fig. 8a for central conserved regions AA, CA, GA, AC, and TT with O ttg operator inserted at the P+1 position.Each pair of bars on the plot show results for a different repressor (I + HQN, E + HQN, and R + HQN) intercepting the circuit at the O ttg operator in position P+1.On the left is the circuit with no ligand added (TFs are intercepting the recombinase from recombination), and on the right is the circuit with ligand added (TFs are detached from the operator, allowing recombination).A118 recombinase is constitutively expressed in all cases.b Schematic of a double-layer deletion circuit with two orthogonal attachment sites containing orthogonal O ttg and O agg DNA operators, enabling selective recombination: each attachment site pair with matching central conserved regions can recombine only when the ligand corresponding to their intercepting repressor is present.Repressor symbols are shown to demonstrate that these modular parts are combined in different ways to generate the data shown in c-f.c-f Assay data for different sets of repressors targeted at the circuit shown in b: c E + HQN and I + KSL, d E + KSL and R + HQN, e E + HQN and R + KSL, and f E + KSL and I + HQN.Source data are provided as a Source Data file.Data in a, and c-f represent the average of n = 6 biological replicates.Error bars correspond to the SEM of these measurements.Figure 26: Flow cytometry of a two-output circuit.a Diagram showing the genetic states corresponding to the four quadrants of each plot shown in b.At top right, the circuit is fully protected by the intercepting TFs E + HQN and I + KSL.E + HQN binds at O ttg in attP site with CA central dinucleotide, protecting the red channel (mKate) deletion, and I + KSL binds at O agg in attP site with AA central dinucleotide, protecting the green channel (sfGFP) deletion.At top left, I + KSL has been induced and sfGFP has been deleted.At bottom right, E + HQN has been induced and mKate has been deleted.At bottom left, both TFs have been induced and both fluorescent proteins have been deleted.b Flow cytometry data for each of the four inducer states, at top left no inducer, at top right + IPTG, at bottom left + cellobiose, and at bottom right + both IPTG and cellobiose.See Supplementary Figure 12 for the individual performances of I + KSL and E + HQN when intercepting a single-output deletion circuit as quantified by flow cytometry.Data for the 2-OUTPUT circuit summarized in Fig. 8b, and Source data are provided as a Source Data file.Data represents the average of n = 6 biological replicates.Error bars correspond to the SEM of these measurements.

Supplementary
Figure 29: Comparing type-I memory and type-II memory kinetics with transient INPUT.a-b display identical schematics to Supplementary Figure

Supplementary
Figure 30: Comparing type-I memory and type-II memory kinetics.Plate reader kinetic data for type-I and type-II memory.a At top, schematics for the type-I memory circuits used, addressed in different experiments by I + YQR+, R + YQR, or E + YQR.Cells were transformed with the relevant TF and an orthogonal deletion circuit (with O agg at P+1) and assayed for fluorescence every 10 minutes as described in Methods, Recombinase plate reader kinetic assays.b At top, schematics for the type-II memory circuits used, addressed in different experiments by I + YQR+, R + YQR, or E + YQR.Cells were transformed with the relevant TF and a constitutive pSK001 A118-expression and assayed for fluorescence every 10 minutes as described in Methods, Recombinase plate reader kinetic assays.c Control experiment without the addition of recombinase.Source data are provided as a Source Data file.Data represents the average of n = 6 biological replicates.Error bars correspond to the SEM of these measurements.