Bimodal fluorogenic sensing of matrix proteolytic signatures in lung cancer

We present a highly-specific, rapidly activatable dual-FRET-smartprobe for the independent optical detection of metalloproteinases and thrombin in lung cancer tissue.


Probe synthesis. General methods 173
The FRET peptide sequences for MMP and thrombin were individually synthesized by standard Fmoc 174 solid-phase peptide chemistry. Dyes and quenchers were coupled also by standard solid-phase methods.

175
General procedures are as follows: 176 Manual peptide synthesis was performed on Aminomethyl-ChemMatrix™ resin using an Fmoc-Rink 177 amide linker. 178 Coupling of Fmoc-Rink amide linker: The Fmoc-Rink-amide linker (0.54g, 1.0 eq) was dissolved in 179 DMF (10 mL) and Oxyma (0.14g, 1.0 eq.) was added and the mixture was stirred for 10 min. 180 Diisopropylcarbodiimide (DIC, 155 µL, 1.0 eq.) was then added and the solution stirred for 1 min before 181 adding it to Aminomethyl-ChemMatrix resin (1.0 g, 1.0 mmol/g). The resulting mixture was stirred at 182 50⁰C for 45 min and washed with DMF (3x10 mL), DCM (3x10 mL) and MeOH (3x10 mL). Finally 183 the resin was treated with Ac2O:Py:DMF (2:3:15) for 30 min in order to cap any remaining free amino 184 groups and it was washed again with DMF (3x10 mL), DCM (3x10 mL) and MeOH (3x10 mL). Resin 185 loading [6] was measured as ~0.58 mmol/g. 186 Fmoc deprotection: In general, to the resin pre-swollen in DCM was added 20% piperidine in DMF 187 and shaken (2x10 min). The solution was drained and the resin washed with DMF (3x10 mL), DCM 188 (3x10 mL) and MeOH (3x10 mL). In the cases were Fmoc deprotection was carried out on Cy5 189 containing peptides, a solution of 2% DBU in DMF (2 × 10 min) was used. 190 S11 Aminoacid coupling: A solution of the appropriate D-or L-amino acid (3.0 eq per amine) and Oxyma 191 (3.0 eq) in DMF (0.1M) was stirred for 10 min. DIC (3.0 eq) was added and stirred for 1 min. The pre-192 activated mixture was then added to the resin pre-swollen in DCM and the reaction heated at 50⁰C for 193 30 min. The solution was drained and washed with DMF (3x10 mL), DCM (3x10 mL) and MeOH 194 (3x10 mL). Completion of coupling reactions were monitored by a Kaiser test or Chloranil test (when 195 secondary amines are involved). The side chain protecting group used were Boc for Arginine, 196 Tryptophan and Lysine. Fmoc-Lys(Dde)-OH was used as orthogonal reagent to introduce the dyes.  To the resin pre-swollen in DCM was added 2% hydrazine in DMF and shaken (5x10 min). The solution 202 was drained and the resin washed with DMF (3x10 mL), DCM (3x10 mL) and MeOH (3x10 mL). (b) 203 Selective Dde deprotection [7] in Fmoc-protected peptides was achieved with a solution containing 204 Imidazole ( moieties. 232 233 S13

Thrombin probes containing the alkyne, FAM & Methyl Red groups. Characterization data 234 (a, b, c) 235
Thrombin substrates (Scheme S2) were built on the resin, cleaved and purified following the general 236 procedures  Hydroxysuccinimide (0.59 g, 1 eq) in EtOAc-Dioxane (1:1, 50 mL) was stirred at 0⁰C and DCC (1.0 g, 259 1 eq) was added allowing the mixture to reach room temperature and kept at these conditions for 12 h. 260 The DCU formed was removed by filtration and the filtrate concentrated under vacuum. EtOAc (100 261 mL) was added and washed with 5% NaHCO3 (2x40 mL), water (40 mL) and brine (40 mL). After 262 drying over anhydrous Na2SO4 and concentrating in vacuo the target compound was recrystallized from 263 DCM/Hexane to obtain a white solid that was used in the next step without further purification.   sulfo-Cy5: 1,3,3-trimethyl-2-((1E,3Z,5E)-3-(5-carboxypyridin-2-yl)-5-(1,3,3-292   trimethyl-5-sulfonatoindolin-2-ylidene)penta-1,3-dien-1-yl) eq) at 50⁰C for 1h. Once the activation was complete the solution was added to the resin together with 318 DIPEA (3 eq) and shaken overnight. The solution was drained and the resin washed with DMF until 319 colourless wash solution, DCM (3x5 mL) and MeOH (3x5 mL). N-terminal Fmoc deprotection was 320 carried out using 2% DBU in DMF (2 x 10 min). In the last step to introduce the quencher, N-terminal 321 capping with the QSY21-NHS ester (1 eq) was carried out in anydrous DMF containing DIPEA (3 eq) 322 for 12 h. The solution was drained and the resin washed with DMF until the wash solution was 323 colourless, DCM (3x5 mL), MeOH (3x5 mL) and finally ether (3x5 mL). Cleavage and purification 324 were done according to the general procedure to obtain the following compounds:  General procedure for the fabrication of the dual-probes by Cu-catalysed azide-alkyne cycloaddition 344 chemistry. Optimized conditions for the click reaction were used. [11] In an eppendorf tube the following 345 aqueous reagents were mixed: alkyne-peptide fragment (a, b or c) (   manufacturer's instructions. PRP was harvested from whole murine blood as described above for human 404 PRP. 405 Plate reader assays were performed as described in the main text, with the MMP buffer replaced with 406 lavage fluid or PRP. All experiments were carried-out in duplicate. Data was normalised by background 407 subtraction of intrinsic fluorescence. 408

PEPTIDE SEQUENCE OPTIMIZATION OF THE MMP-2/9/13 SUBSTRATE 410
To generate molecular probe sequences to optimally/specifically measure MMP activity the probes 411 were synthesised as shown below: 412 413 414 415 Table S3. Library of the FRET compounds generated/screened and iterated from the first to fourth 416 generation. The MMP cleavage site is indicated by italics. *Structure of the tail for each generation of 417 probe. 418

419
The initial library of FRET activatable probes (Table S3 -Generation 1 probes) contained the 420 fluorophore 5,6-Carboxyfluorescein and the quencher Methyl Red separated by a peptide sequence 421 acting as the substrate for the target enzyme. Sequence 1 was selected according to a previous proteomic 422 study using iTRAQ-TAILS. [15] Other sequences were designed and included in the initial G1 screen 423 after analysis of commercial and other reported MMP substrates. 424 The FRET peptides were synthesized by manual standard solid-phase Fmoc chemistry and evaluated as 425 MMP substrates. The response towards MMP and the specific inhibition with the MMP inhibitor 426 Marimastat was measured (Fig S14), with site specific cleavage confirmed for all the sequences (Table  427 S3) by MALDI-TOF MS analysis. The others MMPs tested shared the same cleavage site. 428 Specific inhibition of fluorescence signal using Marimastat (a pan-MMP inhibitor) was successful for 429 all the probes tested. Control probes (sequence 2 and 6) containing D-aminoacids in the cleavage site 430 showed no increase in fluorescence and no cleavage was detected by MALDI TOF MS. 431 Selectivity was determined by analysis of the change in fluorescence over background upon the addition 432 of different proteases (MMP -2, -9, -12, -13, Neutrophil elastase and Thrombin). In order to choose the 433 best probes for in vivo or ex vivo use a number of additional parameters were evaluated which included 434 fluorescence increase in the presence of other related inflammatory proteases such as Thrombin and 435 human Neutrophil Elastase (NE) (Fig S15 and Table S4). 436 437 Results from the generation 1 experiments indicated that sequences 3 and 8 were cleaved by NE, clearly 438 a major issue for a probe for use with tissue. MALDI TOF MS analysis performed after enzymatic 439 treatment confirmed the probes were being cleaved by NE at a different site to those found for the 440 MMPs. NE cleaved probe 3-G1 at -G-P-K-G-I↑K-G-and probe 8-G1 at P-F-G-I↑K-βA (Fig S16-S17).

441
The other sequences remained intact and Thrombin did not cleave any of the sequences.

443
The probes containing Methionine in the cleavage site (sequences 5 and 10-16), which were selectively 444 cleaved by MMPs were deprioritised due to the anticipated stability issues that thioether oxidation can 445 cause. Assays with with human tissue homogenate confirmed specific cleavage and inhibition with 446 Marimastat only for sequence 1 (-G-P-K-G↑L-K-G-) and sequence 9 (-P-F-G↑Nle-K-ßA-) (Fig S20-447 S21). All these assays were conducted using standard FRET peptides prior to lead optimisation for 448 incorporation into the dual-probes.   human Thrombin was used at 5U/ml. Cleavage site for each sequence is indicated in italics. As shown 461 below probe 1 was rapidly cleaved by Plasmin while 9 was observed to be stable to Plasmin. For the next generation probes sequences 1 and 9 were incorporated into peptides that were stabilised 494 with various hydrophilic tails (added in order to improve the aqueous solubility and prevent 495 exopeptidase cleavage. The two selected sequences 1 and 9 were thus flanked by ethylenglycol units 496 (8-amino-3,6-dioxaoctanoic acids) and Lys or D-Lys residues giving the Generation-2, -3 and -4 497 compounds ( Figure S22). 498 With the second generation of probes, the selectivity for MMPs and Plasmin was evaluated. 499 500 Figure S22. Structures of compounds in Generation-2, -3 and -4 501

a) MMPs vs Plasmin selectivity: 502
Experiments with MMP-9,-13 and plasmin were carried out in parallel with the selected sequences. 503 Comparison of results provided by compounds 1-G2 and 9-G2 indicated that sequence 1 was cleaved 504 by plasmin while sequence 9 was totally plasmin resistant. MALDI TOF MS analysis ( Figure S23-S24) 505 was carried out and plasmin cleavage site for sequence 1 was identified as (-G-P-K↑G-L-K-G-). 506 Attempts to make a resistant version replacing the lysine residue with D-aminoacids (-G-P-(D)K-G-L-507 K-G-) resulted in the failure of enzymatic recognition by the MMPs. The combination of PEG units with Lysine increased the aqueous solubility progressively with each 517 generation although the fragment -K-K-(PEG2)2-NH2 used in generation 3 (compounds 1-G3 and 9-518 G3) was unspecifically cleaved when the probes were assayed in tissue homogenate ( Figure S25). A 519 final iteration with the plasmin resistant sequence 9 (compound 9-G4) was synthesized containing as a 520 hydrophilic tail alternate D-Lys as a non-natural aminoacid and PEG units resulting in good solubility 521 and stability as confirmed by MALDI TOF MS and HPLC analysis. The fluorescence increase was 522 selective for MMPs over plasmin (Fig S26). Fluorescence signal of compounds 9-G3 and 9-G4 in the presence of different enzymes. 533

534
In summary from these studies the optimized structure for the MMP probe was selected as PFGNleKA 535 attached to the hydrophilic tail -[PEG2-(D)K]3-NH2 and this was used to construct the optimized dual-536 probe. The widley used MMP peptide GPKGLKG was shown to be non-viable due to its cleavage by 537 endogenous enzymes such as Plasmin. Whilst the initial MMP-peptide sequence was chosen for 538 gelatinases MMP-2 and MMP-9 selectivity [15] our final modified peptide sequence was also highly 539 selective for MMP-13, a collagenase which along with MMP-2 and MMP-9 is upregulated within 540 inflammatory microenvironments. [16] It is not wholly surprising that our peptide sequence was activated 541 by these three different MMPs as they share common targets such as gelatin and several collagen sub-542 types.