Autonomous Synthesis of Functional, Permanently Phosphorylated Proteins for Defining the Interactome of Monomeric 14-3-3ζ

14-3-3 proteins are dimeric hubs that bind hundreds of phosphorylated “clients” to regulate their function. Installing stable, functional mimics of phosphorylated amino acids into proteins offers a powerful strategy to study 14-3-3 function in cellular-like environments, but a previous genetic code expansion (GCE) system to translationally install nonhydrolyzable phosphoserine (nhpSer), with the γ-oxygen replaced with CH2, site-specifically into proteins has seen limited usage. Here, we achieve a 40-fold improvement in this system by engineering into Escherichia coli a six-step biosynthetic pathway that produces nhpSer from phosphoenolpyruvate. Using this autonomous “PermaPhos” expression system, we produce three biologically relevant proteins with nhpSer and confirm that nhpSer mimics the effects of phosphoserine for activating GSK3β phosphorylation of the SARS-CoV-2 nucleocapsid protein, promoting 14-3-3/client complexation, and monomerizing 14-3-3 dimers. Then, to understand the biological function of these phosphorylated 14-3-3ζ monomers (containing nhpSer at Ser58), we isolate its interactome from HEK293T lysates and compare it with that of wild-type 14-3-3ζ. These data identify two new subsets of 14-3-3 client proteins: (i) those that selectively bind dimeric 14-3-3ζ and (ii) those that selectively bind monomeric 14-3-3ζ. We discover that monomeric—but not dimeric—14-3-3ζ interacts with cereblon, an E3 ubiquitin-ligase adaptor protein of pharmacological interest.


Generating the Frb-v1 nhpSer biosynthetic pathway.
Assembly of T7 promoter mutant library. The sequence of six nucleotides within the T7 promoter known to control transcriptional efficiency 1 were mutated to all combinations (4 6 = 4096 mutants) and then cloned in front of a sfGFP reporter gene in the pET28 plasmid backbone. This was done by amplifying DNA fragments off the pET28-sfGFP plasmid (Addgene #85492) using primer pairs P1/P2 and P3/P4 (see Primer   Table below) to create two overlapping fragments that were then recombined using SLiCE 2 into pET28-sfGFP digested with BglII and NcoI. The assembled library of promoter mutants was electroporated into DH10b cells, which were recovered for 1 hr in SOC media and then transferred to a liquid culture of 2xYT media (16 g/L tryptone, 10 g/L yeast extract, 5 g/L NaCl) containing 50 g/mL kanamycin to grow transformants overnight at 37 o C. A small portion of the recovered cells was also plated on LB/agar with kanamycin, and counted the next day revealing 5 x 10 4 transformants were obtained during library assembly and transformation for complete library coverage. The resulting pET28-T7lib-sfGFP library was miniprepped from cells grown in overnight liquid culture.
Selection of T7 promoter mutants. The pET28-T7lib-sfGFP library was transformed into BL21(DE3) serC cells by electroporation. After 1 hr of recovering in SOC medium, cells were plated on LB/agar plates containing 1 mM IPTG, and grown for 24 hrs at which time 94 colonies displaying various levels of fluorescence were used to inoculate 0.5 mL of ZY-non-inducing media (ZY-NIM) 3 . Cells were grown at 37 o C with shaking at 300 rpm overnight, at which point 20 L of each culture was used to inoculate 0.5 mL of ZY-auto-inducing media (ZY-AIM). 3 After 24 hrs of growth at 37 o C, end-point in-cell sfGFP fluorescence was measured with a microplate reader. Twenty-five variants displaying a fluorescence range spanning two-orders of magnitude were selected for DNA isolation by miniprep. Isolated DNA was retransformed into DH10b cells. To confirm expression reproducibility at larger scales, isolated variants were again transformed back into BL21(DE3) serC cells and expressed in duplicate at the 50 mL scale in ZY-AIM media. Variants that did not reproduce the same sfGFP expression level as observed in the first round S3 were discarded. After a third round of expression at 50 mL scale, T7 promoter variants that displayed reproducible sfGFP expression levels at high (34%), medium (11%), low (4%) and trace (1%) levels, with wild-type designated as 100%, were sequenced and selected for downstream Frb pathway assembly.
FrbABCDE library assembly strategy. The FrbABCDE library assembly strategy was adapted from work by Zhao and colleagues 1 such that all permutations of transcriptional promoters in front of each of the five Frb genes were made via Golden Gate assembly (5 5 = 3125 combinations). Each gene was flanked by its own promoter and terminator so that each is transcribed independently. The plasmid backbone contained a CDF origin of replication and spectinomycin resistance to ensure compatibility and stable propagation with the nhpSer GCE machinery (pUC origin/kanamycin resistance) and target protein expression plasmids (p15a origin/ampicillin resistance). Since it has not yet been established whether FrbB or FrbE, or both, is responsible for the isocitrate dehydrogenase-like activity of the fourth step of the pathway, both were included in the library assembly.
Library assembly. Each of the five Frb genes from Streptomyces rubellomurinus were codon optimized for E. coli, synthesized, subcloned into NcoI/XhoI digested pET28 using SLiCE, and sequence verified.
Sequences of each Frb protein are included below. Each Frb gene was then PCR amplified using primers that flanked the T7 promoter and terminator (see Primer Table below). In each PCR reaction the forward primer was a mixture of five that encoded each of the four transcriptional mutants identified above plus wild-type, and a unique Type IIS BsaI restriction site. The reverse primer for each fragment also contained a unique BsaI restriction site. The CDF origin and spectinomycin resistance gene from pCDFduet (Novagen) were also amplified as a single fragment with flanking BsaI restriction sites. All BsaI overhangs were designed so that each of the five promoter-FrbX-terminator fragments and the CDF-Spec R backbone could be ligated directionally into a single circular plasmid. Forward primer mix P7-P11 and reverse primer P12 were used to amplify FrbA, P13-P17 and P18 for FrbB, P19-P23 and P24 for FrbC, P25-P29 and P30 for FrbD, P31-P35 and P36 for FrbE, and lastly P5 and P6 for CDF-Spec R . To assemble these 6 fragments into a circular plasmid, approximately 620, 1100 and 590 ng of gel-purified CDF-Spec R (3200bp), FrbA (2900bp) and FrbB (1355bp) DNA fragments, respectively, were mixed with 20 U of BsaI-HF v2 and 2000 U of T4 ligase (NEB) in 1x ligase buffer in a total of 25 L total volume. Similarly, 610, 500 and 610 ng of FrbC (1418bp), FrbD (1127bp) and FrbE (1394bp), respectively, were mixed with the same amounts of BsaI and T4 ligase in a separate tube. The two reaction mixtures were then incubated in a PCR thermocycler using the following protocol: Step 1: 37 o C for 5 min, Step 2: 16 o C for 5 min, repeat 15 times. Then, the two mixtures were combined and the reaction continued as follows: Step 1: 37 o C for 5 min, Step 2: 16 o C for 5 min, repeat 15 times; Step 3: 80 o C 20 min, Step 4: 12 o C 5 min. The ligation mixture was directly transformed into DH10b cells by electroporation, recovered in SOC for 1 hr and then grown in 2xYT containing spectinomycin to select for transformants. A small portion of the recovered cells was plated on LB/agar with spectinomycin, and counted the next day revealing 390,000 transformants were obtained for approximately 125-fold coverage of the library. DNA from cells grown overnight was miniprepped yielding the pCDF-FrbABCDE_lib DNA. Plasmid from 10 individual clones was miniprepped and digested with NcoI and NheI. Of these, 8 showed the predicted DNA fragment pattern following digestion demonstrating faithful assembly of the targeted library.
Screening for functional FrbABCDE assemblies. The pCDF-FrbABCDE_lib was screened for assemblies able to synthesize nhpSer for translational incorporation into sfGFP-150TAG. Approximately 0.5 g of pCDF-FrbABCDE_lib was electroporated with the nhpSer GCE machinery plasmid, pSF-nhpSer (see below), and the reporter expression plasmid, pRBC-sfGFP-150TAG, into BL21(DE3) serC cells. The serC mutation ensures the expression host cannot biosynthesize pSer, eliminating the chances that pSer could get incorporated into sfGFP by the GCE machinery. Thus, in this system cells will fluoresce only if nhpSer is biosynthesized by the FrbABCDE pathway and then incorporated at the 150TAG codon of sfGFP.
This was confirmed with a negative control transformation in which pCDF-FrbABCDE_lib was replaced with an "empty" pCDFduet vector. After transformation, cells were recovered for 1 hr in SOC media at 37 o C and plated onto LB/agar plates containing 0.5% glycerol, 1mM IPTG, 50 g/mL ampicillin, 25 g/mL kanamycin and 50 g/ml spectinomycin. Plates were incubated at 37 o C for 18 hrs, and then another 18 hrs S5 at room temperature. No fluorescence was observed for colonies containing the "empty" pCDFduet vector.
From cells transformed with pCDF-frbABCDE_lib, 92 fluorescent clones were selected to inoculate 0.5 mL ZY-NIM, which were grown for 16 hrs at 37 o C shaking at 300 rpm. Next, these cells were used to inoculated 0.5 mL 2xYT + 0.5% glycerol, which were grown at 37 o C for 3 hrs at which time expression was induced with 1 mM IPTG and temperature lowered to 30 o C. After 24 hr, in-cell fluorescence was measured and the top 6 clones were selected for follow up analysis. Plasmid DNA was isolated from these top 6 clones, and the pCDF-FrbABCDE pathway plasmids were isolated by transformation into DH10b cells, plating on LB/agar with spectinomycin only and selecting single colonies for subsequent plasmid isolation. T7 promoter identities for these six isolates were determined by sequencing. Pathway functionality of the six pathways were re-tested in subsequent expressions at 50 mL scale, in triplicate.

General protocol for expression of nhpSer containing proteins
Overview of plasmid architecture. nhpSer incorporation using the Frb-v1 pathway requires three plasmids. Contains a pBR322 origin of replication and confers chloramphenicol resistance. Proteins were expressed in either BL21(DE3) serB or B95(DE3) A fabR serB. 3 The latter, which is a partially recoded derivative of BL21(DE3), 7 lacks Release Factor-1 and therefore was used to avoid truncated protein. 3 General protocol for expression of pSer containing proteins. Fresh transformations were performed for every expression. Proteins were expressed exactly as described previously. 3 Briefly, BL21(DE3) serB or B95(DE3) A fabR serB cells were co-transformed with the appropriate pRBC/pRBCduet vector and the pKW2-EFSep vector, grown overnight on LB/agar plates with ampicillin and chloramphenicol, and then multiple colonies were used to inoculate an overnight ZY-non inducing media culture. After overnight growth at 37 o C, these non-inducing cells were used to inoculate a ZY-auto inducing culture, which was grown to an OD ~1.0 at 37 o C, at which point the temperature was adjusted as needed for target protein expression. Cells were harvested ~18-24 hrs later after temperature adjustment.

Expression and purification details of specific target proteins.
sfGFP: The pRBC-sfGFP wt, 150TAG and 134/150TAG plasmids containing C-terminal His6 affinity purification tags were as previously described. 3 pSer and nhpSer containing proteins were expressed as described above at 37 o C. Cells were lysed in Wash Buffer (50 mM Tris, 500 mM NaCl, 5% glycerol, 5 mM imidazole pH 7.4) containing a phosphatase inhibitor mixture (50 mM NaF, 5 mM sodium pyrophosphate and 0.5 mM sodium orthovanadate). Protein was bound to TALON metal affinity resin, washed with 20-50 column volumes of Wash Buffer, and eluted with Wash Buffer + 300 mM imidazole.
Proteins were desalted using a PD-10 column (GE Healthcare) into 50 mM Tris pH 7.4, 150 mM NaCl, frozen in liquid N2 and stored at -80 o C.
Mutations S16D and S16TAG were introduced into HSPB6 using SLiCE. Expressions were performed as described above, but at 25 o C, and purified similarly as sfGFP, except that protein complexes were eluted by proteolytic on-column cleavage with 30 nM bdSENP1 for 1 hr at 4 o C. 9 Proteins were gel-filtered on a Superdex-75 column (GE Healthcare) in 25 mM Tris and 150mM NaCl pH 7.4, flash frozen with liquid N2 and stored at -80 o C. Total protein yields after purification were ~ 10 mg/ L culture for the pSer dependent complex and ~ 8 mg/ L culture for the nhpSer dependent complex.

14-3-3 variants:
The same codon optimized 14-3-3 gene (residues 2-229) used for generating complexes with HSPB6 was cloned into pRBC with an N-terminal His6-bdSUMO fusion tag. Mutations S58E and S58TAG were introduced using SLiCE. Proteins were expressed at 25 o C, purified as done for the 14-3-3 / HSPB6 complexes, eluted using the same bdSENP1 on-column cleavage method, and then gel-filtered on a Superdex S-75 column in 25 mM Tris and 150 mM NaCl pH 7.4. Total protein yields after purification were ~ 12 mg/L culture for the pSer variant and ~ 8 mg/L culture for the nhpSer variant.

SARS-CoV-2 Nucleocapsid protein: Linker Np-sfGFP fusion proteins: A codon optimized SARS-CoV-2
Nucleocapsid Protein gene corresponding to residues 175-245 containing either a codon for Ser or a TAG codon at either S188 or S206 were cloned into pRBC with an N-terminal flanking bdSUMO fusion tag

FLAG-or Myc-14-3-3 variants:
The FLAG-or Myc-tags were cloned into the region between the Cterminal cleavage site of bdSUMO (GG), and the N-terminus Met of 14-3-3 in our original expression constructs for the 14-3-3 variants using SLiCE. Proteins were expressed and purified similarly to the 14-3-3 variants described above.  3 , while nhpSer incorporation was achieved using PermaPhos Frb-v1 as described in this manuscript.

SUPPORTING TABLES
(C) Phos-tag gel of the same proteins shown in panel B. In Phos-tag gel electrophoresis, phosphorylated proteins migrate slower than the non-phosphorylated counter-parts. 12 NhpSer-containing proteins migrate slower than the equivalent pSer-containing proteins. 13 Phos-tag electrophoresis of the pSer-incorporated protein shows ~50% of the protein was hydrolyzed during expression/purification. The slower electrophoretic mobility of sfGFP-nhpSer150 compared to sfGFP-pSer150 is consistent with previous reports. 13 The same data are shown in the main text Fig. 3B.    (as well as Ser) was incorporated into site S16 of a HSPB6 peptide consisting of residues 11-20, which was fused at its N-terminus to bdSUMO. (B) SDS-PAGE (top) and Phos-tag (bottom) of purified bdSUMO-HSPB6 11-20 proteins with Ser16, pSer16 and nhpSer16 are shown in the first three lanes. Upon incubation with -phosphatase (PPase), the pSer16 containing protein was hydrolyzed as evidenced by it migrating with the electrophoretic mobility of wild-type protein, while the nhpSer protein mobility remained constant. Interestingly, the phospho-proteins migrate more slowly on SDS-PAGE compared to wild-type. Slower electrophoretic mobility of phospho-proteins compared to non-phosphorylated proteins in SDS-PAGE has been well-documented previously. 3 Reasons for this are not well understood but presumably the negatively charged phospho-group limits SDS binding capacity to the protein, lowering its charge to mass ratio and causing it to migrate more slowly.                 Figure S25. 14-3-3 phosphorylated at Ser58 with pSer and nhpSer retain a functional phospho-peptide binding groove. (A) Construct design for His6-sfGFP and and His6-sfGFP fused with R18; R18 is an engineered peptide that binds specifically to all 14-3-3 isoforms in the same amphipathic binding grove that phosphorylated peptides of 14-3-3 clients bind 14 (B) His6-sfGFP and (C) His6-sfGFP-R18 proteins were mixed with un-tagged 14-3-3 WT, pSer58 and nhpSer58 variants in equal molar ratios. Protein mixtures were then incubated with TALON metal affinity resin, which was washed extensively to remove unbound proteins. Bound proteins were eluted with imidazole and run on SDS-PAGE. When His6-sfGFP was used as bait (panel B), none of the 14-3-3 forms co-eluted indicating no interaction as expected. When His6-sfGFP-R18 was used as bait (panel C), 14-3-3 WT, pSer58 and nhpSer58 forms co-eluted, indicating a specific interaction between 14-3-3 and R18 in all cases, though the interaction of R18 with the pSer58 and nhpSer58 forms of 14-3-3 was weaker than with WT. Figure S26. 14-3-3 pSer58 is hydrolyzed back to wild-type 14-3-3 in cell lysates. FLAG-tagged 14-3-3 WT, pSer58 and nhpSer58 expressed and purified from E. coli were incubated in buffer (no lysate) or HEK293T cell lysates for 120 min with calyculin A. Samples were run on SDS-PAGE (top) and Phos-tag (bottom) gels and probed with an -FLAG antibody. Phos-tag electrophoresis confirmed the pSer58 variant was fully dephosphorylated in cell lysates during the course of incubation, while the nhpSer58 variant was unmodified. Figure S27. Volcano plots comparing the pool of proteins identified in two biological replicates for (A and C) 14-3-3 WT and nhpSer58 pulldowns and the (B and D) 14-3-3 WT and pSer58 pulldowns. Proteins identified in red as statistically "enriched" were at least 2.8-fold (log2 1.5) (p-value < 0.05) more abundant in the 14-3-3 nhpSer58 or pSer58 pools compared to that of 14-3-3 WT. Similarly, proteins identified in blue as statistically "depleted" were at least 2.8-fold (log2 1.5) (p-value < 0.05) less abundant in the 14-3-3 nhpSer58 or pSer58 pools compared to that of 14-3-3 WT. Protein IDs, enrichment values and associated p-values, as well as raw peptide intensities from both biological replicates, can be found in the Supporting Dataset 1.