Subscriber access provided by CORNELL UNIVERSITY LIBRARY
Article
Synthesis and Characterization of Micelle-Forming PEGPoly(Amino Acid) Copolymers with Iron-Hydroxamate CrossLinkable Blocks for Encapsulation and Release of Hydrophobic Drugs Kevin N. Sill, Bradford Sullivan, Adam Carie, and J. Edward Semple Biomacromolecules, Just Accepted Manuscript • Publication Date (Web): 05 May 2017 Downloaded from http://pubs.acs.org on May 7, 2017
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Biomacromolecules is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.
Page 1 of 41
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biomacromolecules
Synthesis and Characterization of Micelle-Forming PEG-Poly(Amino Acid) Copolymers with IronHydroxamate Cross-Linkable Blocks for Encapsulation and Release of Hydrophobic Drugs Kevin N. Sill*, Bradford Sullivan, Adam Carie, and J. Edward Semple Intezyne Technologies, 3720 Spectrum Blvd, Suite 104, Tampa, FL, 33612 KEYWORDS Block copolymers, poly(ethylene glycol), poly(amino acid), hydroxamic acid, mixed stereochemistry, drug encapsulation
ABSTRACT
Described is the development of a polymeric micelle drug delivery platform that addresses the physical property limitations of many nanovectors. The system employs triblock copolymers comprised of a hydrophilic poly(ethylene glycol) (PEG) block, and two poly(amino acid) (PAA) blocks: a stabilizing cross-linking central block, and a hydrophobic drug encapsulation block. Detailed description of synthetic strategies and considerations found to be critical are discussed. Of note, it was determined that the purity of the α-amino acid-N-carboxyanhydrides (NCA)
ACS Paragon Plus Environment
1
Biomacromolecules
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 2 of 41
monomers and PEG macroinitiator are ultimately responsible for impurities that arise during the polymerization. Also, contrary to current beliefs in the field, the presence of water does not adversely affect the polymerization of NCAs. Furthermore, we describe the impact of poly(amino acid) conformational changes, through the incorporation of D-amino acids to form mixed stereochemistry PAA blocks, with regard to the physical and pharmacokinetic properties of the resulting micelles.
INTRODUCTION Intelligently designed nanoscopic drug carriers, or ‘nanovectors’, offer the ability to selectively deliver therapeutic agents to the target tissue; thus reducing toxicity, improving quality of life, and increasing survival. A wide variety of nanovector technologies have been developed, including liposomal,1 micellar,2 dendrimers,3 nanocrystals,4 and many others.5 Among these strategies, polymeric micelles are one of the most extensively studied.6 Their ability to deliver a variety of payloads—including small molecules, nucleic acids, and proteins—coupled with improved stability compared to other colloidal carriers, make polymeric micelles an attractive nanovector system. Particularly relevant to applications in oncology, the nanoscopic size of polymeric micelles allows them to passively accumulate in solid tumors through the enhanced permeation and retention (EPR) effect.7-8 Additionally, the surface of polymeric micelles can be functionalized with cell-targeting groups9-10 or permeation enhancers that actively target diseased cells and aid in the delivery of their payload. Circulation can be improved by tailoring the size of polymeric micelles to avoid uptake by the liver and renal clearance, which effects particles >120 nm and 50g) was sporadic, often resulting
ACS Paragon Plus Environment
16
Page 17 of 41
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biomacromolecules
in the formation of a significant amount of a PEG-imine impurity. Consequently, use of the mPEG-NH2 free base was reexamined in light of reported successes with free amine initiators, albeit on significantly smaller scales than we required.47-48 Ultimately, the use of our in-house prepared PEGs, coupled with a solvent change from N-methyl-2-pyrrolidone to a mixture of N,N-dimethylacetamide and dichloromethane, allowed for the successful application of free amines in these polymerizations on large scale. Azeotropic Drying of PEG. After multiple years of focusing on this specific polymerization technique, we fully believe that the purity of the NCAs and the PEG macroinitiator determines the quality of the copolymer produced. This theory gave rise to a separate hypothesis: does the presence of water adversely affect the quality of the polymerization? As azeotropic drying of the PEG starting material is difficult on scales >200 g (i.e. 10+ L toluene), the removal of this step would be highly advantageous. Our measurements put the water content of our PEG-amines to be between 0.5–1.5% w/w before azeotropic drying. To determine the effect of the PEG water content, we conducted side-by-side polymerizations with identical starting materials and solvents. The PEG macroinitiator was dried by azeotropic distillation for one reaction, while this step was omitted for the other. The resultant polymers were found to be identical by GPC and 1H NMR analysis. To further test this hypothesis, we repeated this experiment, but added water (10% w/w) to the polymerization that omitted the azeotropic drying step. To our surprise, there was no substantial difference between the polymers (Figure 2). While we did not evaluate these polymers by mass-spectrometry, we could find no significant differences in the appearance, solubility, molecular weight, composition by 1H NMR (Supporting Information: Figures S7, S9, and S10), or performance of the polymers prepared that omitted the drying step or when water was added. Although NCAs have been purified via
ACS Paragon Plus Environment
17
Biomacromolecules
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 18 of 41
aqueous extraction techniques,37 we found it surprising that the polymerization of NCAs under non-anhydrous conditions was successful considering the breadth of literature directed towards excluding water.41-42,
46-47
One possible explanation is differences in nucleophilicity between
water and the amines leads to a significantly slower rate of NCA hydrolysis compared to the rate of propagation. Ultimately, we feel this result strengthens the hypothesis that monomer and initiator purity are the driving factors in obtaining narrow polydispersity PEG-poly(amino acids).
Figure 2. GPC chromatograms of identical mPEG12K-NH-b-p-[(L/D-Glu(OBn)10)]-b-p-[(DPhe15-co-L-Tyr(OAc)25)]-Ac copolymers prepared with and without drying, and with water added to the polymerization. Block Composition. The ability of PEG-PAA copolymers to form spherical micelles suitable for drug encapsulation is determined, in part, by the volume ratio between the hydrophobic and hydrophilic blocks. Polymers too rich in hydrophilic components will remain as unimers in aqueous solutions, while copolymers heavy in hydrophobic components will form non-micellar aggregates.49 The contribution of block composition and length to the morphology of PEG copolymers in solution has been determined experimentally.11,
50-51
We conducted similar
experiments, varying the ratio of each block, until we concluded that a 12 kDa PEG block with a
ACS Paragon Plus Environment
18
Page 19 of 41
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biomacromolecules
40–60 amino acid block would be suitable for forming spherical micelles of the requisite size, capable of drug encapsulation. In order to impart enhanced stability in our polymeric micelles, we sought to develop an amino acid-derived cross-linking block situated between the hydrophilic PEG and hydrophobic core of our copolymers. Covalently cross-linked polymeric micelles, especially in the shell, have been in development for more than 20 years.52-53 An alternative approach has been metal chelation crosslinking, commonly with dendritic polymers.54 Under these constrains, monomer selection for the stabilization block and the hydrophobic core-forming block were systematically evaluated. Ultimately, we elected to pursue cross-linking via metal chelation with the carboxylatecontaining aspartic acid and glutamic acid residues. As the carboxylic acid serves as an effective ligand for iron at high pH levels (pH >7), this chemistry imparts a pH-dependent trigger for micelle stability, allowing the micelle to rapidly destabilize and dissociate at the lower pH levels (pH 5–7) found in tumor environments.55-56 The core-forming drug encapsulation block was empirically chosen to include highly hydrophobic and more hydrophilic amino acids—as the ability to encapsulate hydrophobic drugs relies upon both hydrogen bonding and van der Waals interactions. Initially, leucine and tyrosine were chosen as the core forming amino acids. This led to the development of ITP-101, our first generation copolymer: N3-PEG12K-NH-block-poly(L-Asp10)-block-poly(D-Leu20-co-L-Tyr20)Ac. Cross-linking of the copolymer micelle with Fe(III) was shown to increase stability in plasma by several orders of magnitude over the uncross-linked copolymer.28 Of clinical importance, the cross-linking did not interfere with the encapsulation of a variety of oncology therapeutics, including doxorubicin, paclitaxel, and vinorelbine. The greatest success was the
ACS Paragon Plus Environment
19
Biomacromolecules
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 20 of 41
encapsulation of topoisomerase I inhibitor SN-38 (active metabolite of irinotecan) by ITP-101, which significantly increased circulation time, and tumor exposure.23 Physical Properties of Copolymers: Comparison of Homochiral vs. Heterochiral PAA Blocks. Throughout the development of ITP-101, we continually assessed the physical properties of various PEG-PAA triblock copolymers. During the initial phases of development, significant attention was focused on the differences between copolymers containing all L-isomer amino acids when compared to their
D/L-isomer
mixture counterparts. Often, the differences
were readily apparent, as tremendous variances in solubility were encountered between the polymers of different chirality. Also of note, polymers comprised of D/L-amino acids do not show any difference in cytotoxicity over their all L equivalents (IC50 >5 mg/mL in vitro).28 Many PEG-PAA copolymers developed for generating micelles contain a poly L-amino acid (PLAA) block.17,
20, 57
These copolymers can form three-dimensional secondary structures,
including α-helices and β-sheets, which are believed to impart stability.58 Naturally occurring peptides take advantage of these ordered secondary structures to form specific interactions with other peptides or small molecules. For PLAA copolymers—which are preferably non-specific and can thus encapsulate many different drugs—the limited rotational degrees of freedom of secondary structures are a disadvantage. Ideally, the poly(amino acid) core of a polymer micelle would have the freedom to adopt as many conformations as possible in order for the drug to find its lowest energy state. Thus, increasing the flexibility of the poly(amino acid) portion of our triblocks was an key objective. The use of racemic poly(amino acid) blocks in amphiphilic copolymers has been shown to disrupt α-helical formation and increase the overall flexibility.59-61 Indeed, the incorporation of approximately 10% of a structurally dissimilar residue such as proline,62 or a stereochemically
ACS Paragon Plus Environment
20
Page 21 of 41
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biomacromolecules
dissimilar residue such as a D-amino acid,63 is enough to produce random coils. Hence, we began preparing copolymers with mixed stereochemistry poly(amino acid) blocks. To confirm the presence of random coils, we used circular dichroism spectroscopy to determine the secondary structure of a series of block copolymers pairs with either
L-amino
acids or mixed
stereochemistry (Table 1 and Figure 3). Each homochiral copolymer contained telltale helical negative bands at 208 and 222 nm, the latter can be used to estimate the % alpha helix of a poly(amino acid).64 The spectra of the heterochiral copolymers resembles that of denatured, unfolded proteins with random coils devoid of ordered secondary structures.65 Table 1. Homochiral and heterochiral block copolymers used for circular dichroism analysis #
Block Copolymer
Structure
1a
mPEG12K-NH-b-p-(L-Glu(OBn)30)-Ac
helical
1b
mPEG12K-NH-b-p-(L-Glu(OBn)15-co-D-Glu(OBn)15)-Ac
random
2a
mPEG12K-NH-b-p-(L-Asp10)-b-p-(L-Leu13-co-L-Tyr17)-Ac
helical
2b
mPEG12K-NH-b-p-(L-Asp10)-b-p-(D-Leu13-co-L-Tyr17)-Ac
random
Figure 3. Circular dichroism plot for homochiral and heterochiral copolymer pairs 1a/1b and 2a/2b.
ACS Paragon Plus Environment
21
Biomacromolecules
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 22 of 41
The ability of amphiphilic copolymers to self-assemble in solution is a requisite property for effective drug encapsulation.66 This property can be quantified through measurement of the critical micelle concentration (CMC), which is the concentration of a surfactant above which micelles form in solution. CMC values can be determined through a pyrene fluorescent probe method which compares the ratio of pyrene excitation band when in solution and when encapsulated in the hydrophobic core of a micelle at varying polymer concentrations.21 To test the effect of incorporating D-amino acids on CMC, we prepared a series of di- and triblock copolymer homochiral and heterochiral pairs (Table 2). In each instance, the incorporation of D-amino acids increased the CMC compared to their all L-isomer counterparts, indicating a lower propensity to aggregate in solution. This data was expected and correlates well with the polymer solubility data observed; the higher degrees of freedom imparted by the more flexible polymer chain increases solubility, thus resulting in a higher CMC value. For each polymer, the relative CMC values between the pairs are inversely related to the turbidity of the polymers in aqueous solution as measured by nephelometry. The increased solubility is a consequence of freeing the peptide block from rigid secondary structures so it can freely interact with the media. The solubility of polymeric micelles is an important, and often overlooked, consideration for clinical applications. The increased solubility allows for more practical concentrations to be prepared, and also eases in sterilization via filtration.67 Micelle dimension is another critical consideration for any nanovector drug delivery system and thus we assessed the size of our di- and triblock copolymers via dynamic light scattering. This is a hydrodynamic technique, directly measuring the translation and/or rotational diffusion coefficients, which can be related to size via the Stokes-Einstein equation.68 In each instance, the hydrodynamic radius of the heterochiral block copolymers was smaller than their homochiral
ACS Paragon Plus Environment
22
Page 23 of 41
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biomacromolecules
counterparts. It appears that the extra rotational degrees of freedom allow poly(amino acid) blocks to pack more efficiently. Interestingly, this is the opposite phenomenon others have observed for PEG-poly(L-leucine) and PEG-poly(racemic-leucine),69 indicating that the amino acid composition plays a role in micelle dimensions.
Table 2. Physical properties of di- and triblock homochiral and heterochiral block copolymers #
Block Copolymer
CMC (µg/mL)
Turbidity (NTU)
Micelle Size (nm)
3a
mPEG12K-NH-b-p-(L-Asp(OBn)40)-Ac
6
71
157
3b
mPEG12K-NH-b-p-(L-Asp(OBn)20-co- D-Asp(OBn)20)Ac
24
17
65
4a
mPEG12K-NH-b-p-(L-Phe20-co-L-Tyr20)-Ac
19
221
192
4b
mPEG12K-NH-b-p-(L-Phe20-co-D-Tyr20)-Ac
42
11
73
5a
mPEG12K-NH-b-p-(L-Asp10)-b-p-(L-Glu(OBn)40)-Ac
6
434
88
5b
mPEG12K-NH-b-p-(L-Asp10)-b-p-(L-Glu(OBn)40-co-DGlu(OBn)20)-Ac
20
28
27
6a
mPEG12K-NH-b-p-(Asp10)-b-p-(L-Glu(OBn)20-co-LTyr20)-Ac
2
172
96
6b
mPEG12K-NH-b-p-(Asp10)-b-p-(D-Glu(OBn)20-co-LTyr20)-Ac
7
22
46
7a
N3-PEG12K-NH-b-p-(L-Glu(OBn)30)-Ac
6
434
87
7b
N3-PEG12K-NH-b-p-(L-Glu(OBn)15-co-DGlu(OBn)15)-Ac
7
28
43
8a
N3-PEG12K-NH-b-p-(L-Asp10)-b-p-(L-Leu13-co-LTyr17)-Ac
6
220
40
8b
N3-PEG12K-NH-b-p-(L-Asp10)-b-p-(D-Leu13-co-LTyr17)-Ac
21
22
19
ACS Paragon Plus Environment
23
Biomacromolecules
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 24 of 41
To test whether chirality affects drug-loading, we encapsulated irinotecan in the 7a/b and 8a/b copolymer pairs. In each, the drug-loading efficiency was significantly higher for the heterochiral copolymers—highlighting the effect of increasing the conformational freedom of these copolymers (Table 3).
Table 3. Drug (irinotecan) loading and efficiencies for homochiral vs. heterochiral copolymers #
Block Copolymer
Feed %
Final %
% Efficiency
7a
N3-PEG12K-NH-b-p-(L-Glu(OBn)30]-Ac
13.1
1.6
12.2
7b
N3-PEG12K-NH-b-p-(L-Glu(OBn)15-co-DGlu(OBn)15]-Ac
13.5
13.2
97.8
8a
N3-PEG12K-NH-b-p-(L-Asp10)-b-p-(L-Leu13-co-LTyr17]-Ac
13.1
8.0
61.1
8b
N3-PEG12K-NH-b-p-(L-Asp10)-b-p-(D-Leu13-co-LTyr17]-Ac
12.9
12.6
97.7
Hydroxamate Cross-linking. Although the first series of heterochiral copolymers (including ITP-101) effectively encapsulated several structurally unique agents, they were not a solution for all oncology drugs tested. In particular, aminopterin encapsulations with ITP-101 could not be cross-linked—most likely resulting from competing coordination with its two carboxylates. Therefore, we set forth to prepare an optimal polymer based upon the knowledge of all the synthetic and composition parameters and characterization info described above. A literature survey of other known metal coordinating species led to us to pursue hydroxamic acids as a possible solution for the cross-linking block. This functionality is commonly found in siderophores, a class of microbial secondary metabolites which sequester and transport insoluble Fe(III).70 Hydroxamates are superb bidentate chelating ligands, significantly stronger than
ACS Paragon Plus Environment
24
Page 25 of 41
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Biomacromolecules
carboxylates. For comparison, the binding constants in aqueous solution for acetohydroxamate with Fe(III) is 2.6×1011 M-1, while acetate is 2.4×103 M-1.71 As important, hydroxamic acids are accessible from glutamate and aspartate esters via aminolysis with hydroxylamine (vide infra). In order to determine the optimal block length for this type of cross-linking, we prepared a series of four copolymers with 2, 7, 10, and 20 D/L-mixed stereochemistry hydroxamic acids (specifically, mPEG12K-NH-block-poly(D/L-Glu(NHOH)#-block-poly(D-Phe15-co-L-Tyr25)-Ac). SN-38 pharmacokinetics in a rat model demonstrated a significant increase in plasma exposure when the number of hydroxamates was increased from 2 to 7 (Figure 4). The optimal length was 10, in this copolymer with 50 total amino acids. This number is identical to the optimal number of aspartic acid cross-linking residues used in earlier triblock copolymers.
Figure 4. SN-38 pharmacokinetics (rat model) for copolymers with various hydroxamic acid cross-linking block lengths. Ultimately, the aforementioned copolymer with 10 hydroxamates (ITP-102: mPEG12K-NHblock-poly[L-Glu(NHOH)5-co-D-Glu(NHOH)5]-block-poly[D-Phe15-co-L-Tyr25]-Ac.) proved to be one of our most successful drug encapsulating copolymers. The preliminary synthetic route to ITP-102 proceeded through a protected triblock intermediate 9 constructed with L- and
D-
glutamic acid γ-benzyl esters, D-phenylalanine, and O-benzyl-L-tyrosine NCAs. Conversion to
ACS Paragon Plus Environment
25
Biomacromolecules
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Page 26 of 41
the final hydroxamate triblock was achieved in two steps (Scheme 1). First, trifluoroacetic acid (TFA) and the cation scavenger pentamethylbenzene (PMB) were used to selectively cleave the tyrosine benzyl ether moieties over the glutamate benzyl esters to provide intermediate 10 (not depicted). This reaction required carefully monitoring by 1H NMR (DMSO-d6) for the selective loss of the tyrosine benzylic methylene signals (δ 4.90) compared to the glutamate benzylic methylene signals (δ 5.00). This was followed by aminolysis of the glutamate benzyl esters with hydroxylamine catalyzed by triazabicyclodecene (TBD) to yield ITP-102.72 Partial hydrolysis (up to 16%) of the glutamate benzyl esters during the first step—coupled with arduous product precipitation from the TFA slurry—led us to pursue an alternative approach. Switching from Obenzyl- to O-acetyl-L-tyrosine provided intermediate 11, whose living polymerization and construction was monitored by GPC via disappearance of the NCA peaks with concurrent increase in the molecular weight of the parent peak for each block (Supporting Information: Figure S11). We did not observe any lower molecular weight species resulting from premature chain terminations or other side reactions. This intermediate was converted to ITP-102 via a single tandem deprotection/aminolysis step with hydroxylamine and lithium hydroxide in tetrahydrofuran.73 1H NMR (DMSO-d6) was used to monitor the reaction progression through loss of the benzylic methylene protons (δ 5.00) and tyrosine acetate signals (δ 2.17) with concomitant appearance of the phenol signals (δ ~9.2). Importantly, the final copolymer contained only minor, insignificant amounts (
