Chemistry, structural insight and applications of β-sheet forming lipopeptides

lipopeptides

Cavalli, S.

Citation

Cavalli, S. (2007, January 25). Chemistry, structural insight and applications of β-sheet forming lipopeptides. Retrieved from https://hdl.handle.net/1887/9452

Version: Corrected Publisher’s Version

License: Licence agreement concerning inclusion of doctoral thesis in the Institutional Repository of the University of Leiden

Downloaded from: https://hdl.handle.net/1887/9452

Note: To cite this publication please use the final published version (if applicable).

Chemical Modification of Liposome Surface via

Copper(I)-Mediated [3+2] Azide-Alkyne

Cycloaddition

Abstract. A generic method for the efficient in situ modification of liposomes, based on “click” chemistry, is described. Fluorescence resonance energy transfer (FRET) experiments demonstrated that the reaction occurred at the surface. Furthermore, a simple colorimetric assay was developed for monitoring the reaction. CD spectroscopy was also used to follow the reaction in time, exploiting the change in conformation from random coil to β-sheet of the (Leu-Glu)4 peptide motif upon conjugation to liposome surfaces. Finally, an example of a possible application of this strategy for the preparation of immunogenic liposomes is given. Toll-like receptor 7 ligands (TLR7-Ls) were conjugated to the vesicle outer membrane and the immunogenicity was investigated.

* Part of this work has been published: S. Cavalli, A. R. Tipton, M. Overhand and A. Kros Chem. Commun.

2006, 30, 3193-3195.

5.1 Introduction

In the search for “high-tech” liposomes as drug carriers,¹ surface modification of lipid vesicles by peptides or proteins has been particularly exploited to target specific cells.² Exceptionally appealing appears the preparation of immunoliposomes as new synthetic peptides-based vaccines.^3,4 Covalent anchorage of (poly)peptides onto phospholipid bilayers has been accomplished by functionalizing the peptide part with a lipidic moiety.^5,6,7 However, lipidation of peptides render them less soluble, thereby complicating their efficient incorporation and distribution on the exterior of liposomes. Alternatively, soluble peptides, in their active conformation, have been covalently attached to functionalized lipid anchors already inserted in the membrane of vesicles.⁴ Chemical connectivity has been achieved by this approach with variable success using amide⁸ or thiol-maleimide coupling^9,10 as well as by imine¹¹ or hydrazone linkage.¹² In many cases there is a lack of specificity resulting in the uncontrolled formation of the number of covalent bonds between the liposome and the (poly)peptide of interest.

The [3+2] cycloaddition¹³ between azides and strained¹⁴ or terminal¹⁵ alkynes has recently emerged as a highly useful chemical handle for conjugation in both a non copper-mediated^14,16 and a copper(I)-catalyzed^15,17 manner, also known as “click” reaction.

The unreactive nature of both azides and alkynes towards functional groups present in biomolecules as well as the thermal and hydrolytical stability of their cycloaddition product established this chemical approach as a powerful means for a wide range of biomolecular applications. For example the selective modification of proteins,¹⁸ enzymes,¹⁹ virus particles²⁰ and cells,²¹ has been shown. Besides chemoselectivity and stability, particularly intriguing was also found the fact that the copper(I)-catalyzed “click” reaction, can occur efficiently in aqueous media at room temperature.^15a Based on these considerations, the copper-mediated [3+2] azide-alkyne cycloaddition was investigated as novel generic chemical tool for the facile in situ surface modification of liposomes.

A schematic representation of the general approach is shown in Figure 5.1. First, small unilamellar lipid vesicles are prepared bearing terminal alkyne groups at their surface. The number of functional groups can be controlled within a wide range without altering the properties of the liposomes, allowing the proper ratio between a terminal alkyne functionalized lipid and natural lipids that have the tendency to form vesicles easily to be chosen. Subsequently, azido-modified peptides are coupled to the liposome via

copper(I)-mediated [3+2] azide-alkyne cycloaddition, resulting in lipid vesicles which have been chemically modified at their surface by peptides.

DOPE-COC≡CH

DOPE- ≡

=

O O O O

O P OO NH OH

= O NH

O DOPE

Cu(I)-mediated [3+2]

azide-alkyne cycloaddition azido-modified peptide

NH O

DOPE NH

O DOPE

liposomes bearing terminal alkyne groups

NH O

N NN NH

O

DOPE DOPE

chemically modified liposome surface by peptides

N3

DOPE-COC≡CH

DOPE- ≡

DOPE-COC≡CH

DOPE- ≡

=

O O O O

O P OO NH OH

= O NH

O

DOPE =

O O O O

O P OO NH OH

= O NH

O NH

O DOPE

Cu(I)-mediated [3+2]

azide-alkyne cycloaddition azido-modified peptide

NH O

DOPE NH

O DOPE NH

O NH

O

DOPE NH

O NH

O DOPE

liposomes bearing terminal alkyne groups

NH O

N NN NH

O

DOPE DOPE N

H O

N NN NH

O N NN NH

O NH

O

DOPE DOPE

chemically modified liposome surface by peptides

N3 N3 N3

Figure 5.1. Schematic representation of the general approach for the chemical modification of a liposome surface by peptides via copper(I)-mediated [3+2] azide-alkyne cycloaddition.

To anchor terminal alkyne moieties to the lipid bilayer, a terminal alkyne derivative of 1,2-Dioleoyl-sn-Glycero-3-Phosphoethanolamine (DOPE-COC≡CH 1) was prepared by reacting propiolic acid with 1,2-Dioleoyl-sn-Glycero-3-Phosphoethanolamine (DOPE) under the influence of 1-[3-(Dimethylamino)propyl]-3-ethylcarbodiimide (EDC), as described in Scheme 5.1. The lipid derivative 1 can be mixed in different ratios with 1,2-Dioleoyl-sn- Glycero-3-Phosphoethanolcholine (DOPC) to prepare the desired liposomes, bearing alkyne functions at their outer membrane, which can react with azido-modified peptides.

O O O O

O P O O

NH₂ OH

O O O O

O P O O

NH OH

O a

DOPE

1

Scheme 5.1. Reagents and conditions: (a) CH≡CCOOH (3 equiv.), EDC (3 equiv.) in DCM.

To demonstrate the versatility of the “click” chemistry approach for surface modification of vesicles, three examples are discussed in detail in the following sections of this chapter. In the first example, fluorescence resonance energy transfer (FRET) is exploited to prove that the cycloaddition takes place at the surface. Furthermore, based on the observation that a color change occurs upon reaction, a colorimetric assay is developed to follow the reaction in time without the need of any equipment. In the second example, based on the conformational change from random coil to β-sheet of the (Leu-Glu)4 motif upon conjugation to liposome surfaces, CD spectroscopy is used as another powerful tool to easily follow the reaction in time and demonstrate that the cycloaddition truly occurs at the surface.

The last example describes an application of this chemical strategy for the preparation of immunoliposomes. Toll-like receptor²² 7 ligands (TLR7-Ls) are covalently bonded to the surface of unilamellar liposomes and the immunogenicity of these conjugates is investigated.

5.2 Colorimetric Assay for Chemical Modification of Liposome Surface via a Copper-Mediated [3+2] Azide-Alkyne Cycloaddition

Copper-mediated [3+2] azide-alkyne cycloaddition was used to functionalize the surface of lipid vesicles. In the majority of the examples of bioconjugation via copper-catalyzed triazole, the Cu(I) species is generated in situ from the soluble Cu(II) source CuSO4 and an exogenous reducing agent, either a copper wire or sodium ascorbate as well as tris-(carboxyethyl)phosphine (TCEP). Direct addition of Cu(I) salts to proteins or cells to effect bioconjugation has likely been avoided because of the extremely low solubility of these salts in aqueous media and their tendency to disproportionate quickly to Cu(0) and Cu(II).^18b,19b,20 In order to find a general easy protocol for the “click” reaction, several conditions were investigated. For the in situ generation of Cu(I), 1 mol% of CuSO4.5H2O and 5 mol% of sodium ascorbate were used as described by Sharpless and co-workers.^15a However, this method was not always successful in modifying the liposome surface.

Alternatively, Copper(I) salts have been used directly in the absence of a reducing agent (e.g. CuI, CuOTf^.C6H6 and [Cu(NCCH3)4][PF6]). These reactions usually require a co-solvent and one equivalent of a nitrogen base (i.e. 2,6-lutidine, triethylamine, diisopropylethylamine or pyridine). However, formation of undesired by-products, primarily diacetylenes, bis-triazoles, and 5-hydroxytriazoles, was often observed.²³ The complications encountered with the direct use of CuI species was minimized when 2,6-lutidine was used. In addition, exclusion of oxygen further improved product purity and yield.^15a Tirrell and co-workes²¹ have shown the efficient use of CuBr for copper-catalyzed triazole formation on the cell

surface. Based on these considerations, an aqueous suspension of CuBr was added directly to the vesicles. As expected, when CuBr was used, liposomes modified by peptides at their outer membrane were easily obtained with good reproducibility. CuBr was therefore selected as the best copper source. It should be noted that the purity of the CuBr is crucial to the success of the experiment (only 99.999% pure CuBr stored under dry conditions was used).²¹ Although this synthetic protocol can be applied to any azido-functionalized compound for its conjugation to the outer membrane of liposomes, in this work, azido-modified peptides were mainly used, as lipopeptides constitute the general research topic discussed in this thesis. To start with an easy compound, a single amino acid was chosen, N-α-azido-modified lysine (N3-Lys), in order to simplify the protocol and find the best conditions. Furthermore, to prove that the reaction truly occurs at the surface, the fluorescence resonance energy transfer (FRET) or Förster type energy transfer²⁴ was exploited. For this purpose, 7-nitrobenzofurazan (NBD) and lissamine rhodamine (LR) were chosen as donor-acceptor pair.

=

O O O O

O P O O

NH OH

O

DOPE-COC≡CH 1

=

DOPE- ≡CH 1 NH

O DOPE

Cu(I) catalyzed reaction

in water

NH₂ O N3

NH

NO₂ NO N

FRET

2

Cu(I) catalyzed reaction

in water

NH O

N NN NH2 O

NH

NO2 NO N DOPE-LR

1 NH

O

DOPE NH

O DOPE

1

DOPE-LR 1

NH O

DOPE DOPE

3

O N

N

S O₂ S O₃

O

O N H

O

O

DOPE-LR = _{O P O}^O

OH

=

O O O O

O P O O

NH OH

O

DOPE-COC≡CH 1

=

DOPE- ≡CH 1 NH

O NH

O DOPE

Cu(I) catalyzed reaction

in water

NH₂ O N3

NH

NO₂ NO N

NH₂ O N3

NH

NO₂ NO N

FRET

2

Cu(I) catalyzed reaction

in water

NH O

N NN NH2 O

NH

NO2 NO N NH

O

N NN NH2 O

NH

NO2 NO N DOPE-LR

1 NH

O NH

O

DOPE NH

O NH

O DOPE

1

DOPE-LR 1

NH O NH

O

DOPE DOPE

3

O N

N

S O₂ S O₃

O

O N H

O

O DOPE-LR =

O N

N

S O₂ S O₃

O

O N H

O

O

DOPE-LR = _{O P O}^O

OH O P O

O

OH

Figure 5.2. Schematic representation of the general approach based on FRET effect occurring upon functionalization of a liposome surface via “click” chemistry.

In the general approach, small unilamellar lipid vesicles, containing terminal alkyne groups and a fluorescent acceptor molecule at their surface, are prepared (Figure 5.2). As a proof of principle, an azido- and fluorescent probe modified lysine (2) is coupled to the liposome using a catalytic amount of CuBr. When the reaction occurs, donor 2 is linked in close proximity to the acceptor (DOPE-LR) present in the lipid bilayer, favouring energy transfer (FRET). The terminal alkyne functionalities in the internal membrane of the liposomes are screened and do not take part in the reaction, thus allowing the energy transfer to occur only with the acceptor molecules available at the outer surface.

N3-Lys(NBD)-NH2 (2) was prepared as described in Scheme 5.2.

NH O FmocNH

NHMtt NH

O N₃

NHMtt O

N₃

NH

NO₂ N

O N

NH₂ FmocNH

Rink Amide resin

e-g c-d a-b

h

O

O O P O

OH N

H O

O

O

O

N NN NH₂ O

NH

NO₂ NO N O

O O P O

OH N

H O

O

O

O

mixed to DOPC in the liposome 1

3 II I

2

Scheme 5.2. Reagents and conditions: (a) 20% piperidine in NMP. (b) PyBOP (4 equiv.), DIPEA (8 equiv.) in NMP. (c) 20% piperidine in NMP. (d) CF3SO2N3, CuSO4·5H2O (cat.) in DCM. (e) 94/1/5 (v/v) DCM/TFA/TIS.

(f) NBD-Cl (4 equiv.), 0.5% DiPEA in dry THF. (g) 90% TFA in H2O. (h) CuBr (0.5 equiv.) in H2O.

Removal of the Fmoc group using 20% piperidine in NMP furnished unprotected Rink Amide resin, which was reacted with a mixture of N-α-Fmoc-N-ε-4-methyltrityl-Lys-OH (Fmoc-Lys(Mtt)-OH) and benzotriazole-1-yl-oxy-tris-pyrrolidino-phosphonium hexafluorophosphate (PyBOP) in NMP to yield protected Lys (I). The Fmoc group was

subsequently removed as described above and diazo transfer was performed on the solid support²⁵ to generate azido-modified Lys (II). After deprotection of the methyltrityl (Mtt) group, by rinsing the resin with a solution of 94/5/1 (v/v/v) of DCM/triisopropylsilane (TIS)/trifluoroacetic acid (TFA), a mixture of NBD-Cl and 0.5% DIPEA in dry THF was added to the resin. N-α-azido-Lys(NBD) 2 was collected after cleavage from the solid support by treatment with 9/1 (v/v) of TFA/H2O.

In this example, two solutions of unilamellar liposomes with an average diameter ranging between 110-120 nm (A in Figure 5.3) were prepared in water by sonication of a mixture of the readily prepared terminal alkyne derivative of 1,2-Dioleoyl-sn-Glycero-3- Phosphoethanolamine (DOPE-COC≡CH 1), DOPC and a commercially available lissamine rhodamine derivative of DOPE (DOPE-LR).

0 50 100 150 200 250 300 350 0

10 20 30 40 50

% in class

Diameter (nm)

b a

0 50 100 150 200 250 300 350 0

10 20 30 40 50

% in class

Diameter (nm)

0 50 100 150 200 250 300 350 0

10 20 30 40 50

% in class

Diameter (nm)

c

0 50 100 150 200 250 300 350 0

10 20 30 40 50

% in class

Diameter (nm)

d

Figure 5.3. Size distribution of liposomes determined by Photon Correlation Spectroscopy (PCS). Vesicle solution A (a) before reaction, cumulant Z average = 122.9 nm; polydispersity = 0.359. (b) after reaction, cumulant Z average = 133.4 nm; polydispersity = 0.257. Size distributions for negative control solution B (c) before addition of N3-Lys(NBD)-NH2 (2), cumulant Z average = 111.5 nm; polydispersity = 0.373. (d) after addition of 2, cumulant Z average = 129.6 nm; polydispersity = 0.325.

Both solutions appeared purple in color due to the presence of the LR. When the vesicles solution was added to 2 equiv. of N3-Lys(NBD)-NH2 (2), the color changed to red/orange. To one solution a catalytic amount of CuBr (0.5 equiv.) was added (vesicle solution A). The second solution did not contain any CuBr (negative control B). After

reaction, which was allowed to occur for 20 hrs, no dramatic effect on the size of the liposomes (average diameter = 120-130 nm, Figure 5.3 B) was observed. Subsequently, samples were dialyzed and the color of the solution B turned back to purple, while solution A remained orange, indicative of the covalent attachment of the NBD derivative 2 to the outer membrane forming the cycloaddition product 3 (Scheme 5.2 and Figure 5.2).

FRET spectra of the reaction mixture before dialysis are shown in Figure 5.4. Because of the presence of the NBD derivative 2, both samples appeared red (see Figure 5.5, top and color images on the back of the cover). Interestingly, in the case of the vesicle solution A, the intensity of the LR emission clearly increased due to the energy transfer, demonstrating that the NBD fluorophore is located near the DOPE-LR. This implies the occurrence of the “click”

reaction at the liposome surface. In contrast, only a weak FRET effect was observed in the negative control B due to the random presence of non reacted compound 2.

Figure 5.4. FRET for (⎯) vesicle solution A and (⋅⋅⋅⋅⋅) negative control solution B, which did not contain CuBr.

Samples were measured after 20 hrs reaction before dialysis.

Based on the observation that a color shift occurred after dialysis, a colorimetric assay was developed to follow the reaction in time (Figure 5.5, a color image is on the back of the cover). Samples were taken from the reaction at several time intervals and dialyzed. Already after 30 min a color change was observed in the case of the vesicle solution A (Figure 5.5, row A, see color image on the back of the cover). In contrast, the vesicle solution B did not show any color change in time (Figure 5.5, row B, see color image on the back of the cover).

300 400 500 600 700 800

0 500

In t. (a . u .)

λ (nm)

The relevance of this easy test is based on its practical use, allowing fast monitoring of the reaction, simply by observing the color change by the naked eye, whereas the detection of the lipid-peptide conjugate 3 by LC-MS was rather time consuming and laborious.

Figure 5.5. (top) Images of the color change (left) before reaction, (middle) during reaction and (right) after reaction. In all cases A corresponds to the vesicle solution A and B to the vesicle solution B. (bottom) Colorimetric test for (row A) the vesicle solution A and (row B) negative control solution B, which did not contain CuBr. Samples were taken after 0, 0.5, 1, 2, 4 and 20 hrs and dialyzed for 2 hrs (color images are on the back of the cover).

The rate of formation of product 3 was followed by measuring UV absorbance at 472 nm as a function of time (Figure 5.6). In the case of vesicles solution A, the intensity increased quickly and leveled off after approximately 4 hrs, indicative of a fast reaction. The absence of the Cu(I) catalyst (vesicle solution B) resulted in no reaction at the surface of the liposomes.

Finally, HPLC analysis performed after the reaction, revealed that 57% of the DOPE- COC≡CH (1) had reacted, implying that only the outer layer of the liposome membrane was accessible and took place in the formation of the product 3.

A B A B A B

during reaction before reaction

Time (hrs)

0 5 10 15 20 0,0

0,1 0,2 0,3

Abs. (a. u.)

Time (hrs)

a

0 5 10 15 20

0 20 40 60 80 100

% o f 1

Time (hrs)

b

Figure 5.6. (a) UV absorbance of NBD at 472 nm as function of time for the vesicle solution A. Absorbance values were corrected using the total lipid concentration obtained performing a phosphorous test.²⁶ (b) Normalized HPLC data. Decrease of DOPE-COC≡CH (1) as a function of time.

In conclusion, the liposome surface has been efficiently modified using copper- mediated [3+2] azide-alkyne cycloaddition. The reaction proceeded at room temperature and was finished within 4 hrs. FRET proved that the chemical modification truly occurred at the surface of the liposomes and a colorimetric assay was developed that allowed monitoring of product formation in time exploiting the visible color change upon reaction. The generic nature of this approach consents to any azido-functionalized peptide to be straightforwardly conjugated to the outer membrane of liposomes, using a similar synthetic protocol as shown for compound 2. Additionally, NBD can be easily introduced²⁷ in order to follow the reaction using the colorimetric assay.

5.3 β-Sheet Folding Induced by Liposome Conjugation via Copper(I)-Catalyzed [3+2] Azide-Alkyne Cycloaddition

Amphiphilic lipopeptide DOPE-(Leu-Glu)4-NH2 (octa-ALP, Chart 5.1) was shown to form two-dimensional β-sheet monolayers at the air-water interface (see Chapter 2).

However, octa-ALP did not form well-defined assemblies with a regular size and morphology upon dispersion in aqueous solutions. Also the preparation of liposomes by mixing different amounts of octa-ALP with DOPC proved rather difficult to achieve (see Chapter 4). The complications encountered in the preparation of uniform lipid vesicles decorated with β-sheet oligopeptides could be overcome using the “click” protocol described above for the post-modification of pre-formed liposome directly at the surface.

Chart 5.1. Chemical structure of the amphiphilic lipopeptide DOPE-(Leu-Glu)4–NH2 (octa-ALP).

In this section the conjugation via “click” chemistry of the N-α-azido modified octapeptide N3-(Leu-Glu)4-NH2 (4) is shown (Scheme 5.3), demonstrating the possibility to functionalize the liposome outer membrane using peptide sequences. Furthermore, based on the results already discussed in the previous Chapters, which showed that the ramdon coil (Leu-Glu)4 motif folded into β-sheet upon conjugation to a lipid tail, CD spectroscopy was used to visualize this transition during the occurrence of the reaction, in order to follow the reaction in time and demonstrate again the formation of the conjugate at the surface, as illustrated in the schematic representation in Figure 5.7. The octapeptide N3-(Leu-Glu)4-NH2

(4) was therefore selected as suitable peptide sequence and the random coil to β-sheet transition, due to the conjugation of the peptide to the liposome, was visualized by CD spectroscopy, which resulted in another powerful tool to easily follow the reaction in time.

4

HN N H

NH₂ O HO O

O O

O O O

O P O OH O

NH O

O octa-ALP

Cu(I) catalyzed reaction

in water

PE N H O

PE PE

PE

random coil peptide

β-sheet peptide

N3

NH O

N NN NH

O N NN NH

O

N NN PE N

H O

PE PE

NH O PE

NH O NH

O

Cu(I) catalyzed reaction

in water

PE N H O

PE PE

PE

random coil peptide

β-sheet peptide

N3

N3

NH O

N NN NH

O N NN NH

O N NN NH

O

N NN PE N

H O

PE PE

NH O PE

NH O NH

O

PE N H O

PE PE

PE N H O

PE PE

NH O PE

NH O NH

O

Figure 5.7. Schematic representation of the approach based on random-coil to β-sheet transition observed for the (LeuGlu)4 peptide motif upon its conjugation to the liposome surface via “click” chemistry.

Immobilized octapeptide III was prepared on a Rink amide resin by standard Fmoc solid phase peptide synthesis (SPPS) using a peptide synthesizer (Scheme 5.3),²⁸ followed by diazo transfer on solid support.²⁵ Compound III was subsequently cleaved from the resin by treatment with a solution of 9/1 (v/v) TFA/H2O yielding N-α-azido functionalized octapeptide N3-(Leu-Glu)4-NH2 (4).

Two solutions of unilamellar liposomes (with an average diameter ~ 100 nm) were prepared in water by sonication, using a mixture of 50 mol% DOPE-COC≡CH 1 and 50 mol% DOPC (to a total lipid concentration of 1 mM). Subsequently, the vesicle solutions were added to N3-(Leu-Glu)4-NH2 (4) and CuBr (vesicle solution C), while the negative control did not contain any CuBr (vesicle solution D). CD spectra were recorded at different time intervals (t = 0, 15, 30 and 45 min and t = 1, 2 and 4 hrs).²⁹ After reaction no dramatic effect on the size of the liposomes was observed (average diameter ~ 120 nm). This observation shows the possibility of loading liposome surface with a high amount of a peptide. It is important to note that in general a much lower molar percentage of lipid anchor is used in the other examples of the modification of liposome outer membranes. Speculations on the possible orientation of the β-sheets on the surface contemplate two possible arrangements. In one case, the β-sheet peptides may lie flat on the surface. In the other case, the peptides could stick into the water. The fact that the average liposome diameter did not change much before and after reaction indicates that the first possibility might be more realistic compared to the second one. However, no data is available thus far to confirm this hypothesis.

N₃ O

HN O

HO O

NH 3 N

H O

HN O

O O

NH

4

Fmoc O

HN O

HO O

NH₂ a FmocNH

e

O

O O P O

OH N

H O

O

O

O

O

O O P O

OH N

H O

O

O

O

N NN O

HN O

HO O NH O

HN O

HO O NH₂

3

1

mixed to DOPC in the liposome

5

Rink Amide resin b-d

4 III

Scheme 5.3. Reagents and conditions: (a) Fmoc removal: 20% piperidine in NMP. (b) Solid Phase Peptide Synthesis (SPPS): (i) coupling: Fmoc-Glu(OtBu)-OH or Fmoc-Leu-OH, 0.45 M HBTU/HOBt in DMF and 2 M DiPEA in NMP (for details see experimental section); (ii) capping: 0.5 M Ac2O and 0.125 M DiPEA in NMP (for detail see experimental section). (iii) Fmoc removal: 20% piperidine in NMP. Iterated 4 times.

(c) CF3SO2N3, CuSO4·5H2O (cat.) in DCM. (d) 90% TFA in H2O. (e) CuBr (0.5 equiv.) in H2O.

CD spectra were measured at different time intervals, as shown in Figure 5.8. The occurrence of the reaction was monitored by following the increase of absorbance at

~ 200 nm and the decrease of absorbance at ~ 218 nm in time (see also Figure 5.9, b), which indicated the transition from random coil to β-sheet upon formation of the lipopeptide conjugate 5. Interestingly, already after 15 min a change in the peptide folding was observed.

However, the β-sheet conformation started to become dominant after 45 min and after 2 hrs a strong β-sheet CD signal was recorded. In the case of the negative control solution D, the CD signal remained unmodified even after 4 hrs as shown in Figure 5.9, demonstrating that the peptide was not conjugated at the outer membrane of the vesicles and remained in solution in a random coil conformation.

200 220 240 260 -10

-5 0 5 10 15 20 25 30

Abs. ( a . u.)

λ (nm)

Figure 5.8. CD spectra at several time intervals for solution C. Symbols denote measurement at () t = 0 min;

() t = 15 min; () t = 30 min; (%) t = 45 min; (0) t = 1 h; (]) t = 2 hrs and () t = 4 hrs. Arrows indicate increasing of absorbance at ~ 200 nm and decreasing of absorbance at ~ 218 nm, showing the occurrence of the transition from random coil to β-sheet in time.

Figure 5.9. (A) CD spectra at several time intervals for negative control solution D, which did not contain CuBr. Symbols denote measurement at () t = 0 min; () t = 15 min; () t = 30 min; (%) t = 45 min and (0) t = 1 h. (B) Absorbance at 218.5 nm as a function of time for () vesicle solution C and (u) negative control solution D, which did not contain CuBr.

Plotting the decrease in value of the absorbance at ~ 218 nm (typical minimum for β-sheets)³⁰ against time, information about the occurrence of the reaction in time could be obtained (Figure 5.9, B). Already after 15 min some of the N3-(Leu-Glu)4-NH2 (4) was

A B

200 220 240 260

-6 -5 -4 -3 -2 -1 0

abs. (a. u.)

λ (nm)

0 1 2 3 4

-8 -6 -4 -2 0 2

abs. (a. u.)

Time (hrs)

conjugated to the outer membrane of the lipid vesicles. The reaction proceeded at room temperature, reaching good conversions already after 45 min and was finished within 2 hrs.

In summary, β-sheet folding of the random coil (Leu-Glu)4 peptide sequence has been induced by conjugation to liposome outer membrane via the “click” chemistry protocol described in the previous section, demonstrating the possibility to functionalize the liposome exterior using peptide sequences rather than just amino acids. Furthermore, liposome surface was loaded with an high amount of the peptide 4 without affecting the average diameter of the vesicles. Finally, exploiting the random coil to β-sheet change in conformation, the reaction could be monitored by CD spectroscopy in real time without the need of any workup. Already after 15 min some of the N3-(Leu-Glu)4-NH2 (4) was conjugated to the outer membrane of the lipid vesicles, demonstrated by the change in the CD signal. The reaction proceeded at room temperature and was finished within 2 hrs.

5.4 Copper(I)-Catalyzed [3+2] Azide-Alkyne Cycloaddition towards Immunoliposomes

In the development of potent and selective vaccines toll-like receptors (TLRs)²² are important targets,³¹ as they recognize specific molecules associated with common pathogens, such as bacteria and viruses and generate an inflammatory response.³² Although a range of TLRs (TLR1 through TLR11) have been discovered, only a limited number of ligand molecular structures have been established until now.³³ For examples, lipopeptides having a terminal N-palmitoyl-S-[2,3-bis(palmitoyloxy)-(2R)-propyl]cysteine (Pam3Cys-OH) moiety and oligodeoxynucleotides containing the CpG motif (GACGTT) have been shown as ligands of TLR2³⁴ and TLR9,³⁵ respectively. Interestingly, TLR7 is expressed intracellularly in the endosomes and can bind small drug-like compounds such as imidazoquinolines³⁶ and guanine derivatives.³⁷ A series of toll-like receptor 7 ligands (TLR7-Ls), the alkylazide derivatives of the potent interferon-alpha inducer 2-alkoxy-8-hydroxyadenine (6a-b, Chart 5.2), have recently been prepared by Weterings et al.³⁸. Biological studies have shown that TLR7-L1 (6a) and TLR7-L2 (6b) induced production of the T helper cell (Thl)-activating cytokine, interleukin-12 (IL-12p40) to comparable levels.³⁸ The conjugation of the TLR7-L2 (6b) to a peptide sequence, the well-known major histocompatibility complex (MHC) class I epitope from ovalbumin (SIINFEKL), led to complete abolishment of the IL-12p40 production.³⁸ The lack of dendritic cell (DC) activation by the conjugate was attributed to either the poor

or to the impaired intracellular trafficking of the conjugated ligand compared to free TLR7-L2 (6b), or both. According to this line of reasoning, the stimulatory capacity of the conjugate could be restored by the introduction of a cleavable linker between the TLR-ligand and the peptide moiety, allowing release of the ligand after internalization. The internalization itself could be improved by inclusion of functionalities known to enhance endosomal uptake.

Connecting the TLR7-Ls (6a-b) to the outer membrane of liposomes was therefore believed to improve their cellular uptake and consequently enhance the dendritic cell (DC) activation.

N

N N

HN NH₂

O N₃ O

n = 0 6a n = 3 6b n

Chart 5.2. Chemical structure of toll-like receptor 7 ligands (TLR7-Ls). Both compounds contain the 2-alkoxy- 8-hydroxyadenine unit modified with two different lengths of the alkylazide linker, n = 0 for TLR7-L1 (6a) and n = 3 for TLR7-L2 (6b).

In this section, the copper(I)-catalyzed [3+2] azide-alkyne cycloaddition protocol developed previously was used for the in situ preparation of the liposome-TLR7-L conjugates (7a-b, Scheme 5.3).

O

O O P O

OH N

H O

O

O

O

1

a O

O O O

O P O O

NH OH

O N

N N

HN NH₂

O

O

n = 0 7a n = 3 7b

N NN n

Scheme 5.3. Reagents and conditions. (a) 2 equiv. TLR7-L1 (6a) or TLR7-L2 (6b), 0.5 equiv. CuBr in phosphate buffer (pH = 7.4).

Solutions of unilamellar liposomes (with an average diameter ranging between 80 and 100 nm) were prepared by sonication in phosphate buffer (pH 7.4) using pure DOPE-COC≡CH (1); 50 mol% of 1 and 50 mol% of DOPC or 10 mol% of 1 mixed with 90 mol% of DOPC (to a total lipid concentration was 1 mM). A detailed description of the samples is given in Table 5.1. Liposome solutions E,F,G,³⁹ H and I were used for the reaction with the TLR7-Ls, while liposome solutions J, K and L were used as controls. Subsequently, the vesicle solutions were added to TLR7-L1 (6a) or TLR7-L2 (6b) and CuBr (solutions E, F, G, H and I), as described in Table 5.1 and Scheme 5.3.

Table 5.1. Lipid composition of the different solutions. All samples were prepared using a total lipid concentration of 1 mM in phosphate buffer (pH 7.4).

Solution Composition 6a

added

6b added

CuBr added

E DOPE-COC≡CH (1) yes yes

F DOPE-COC≡CH (1) yes yes

G³⁹ 50 mol% DOPE-COC≡CH (1) mixed with 50 mol%

DOPC yes yes

H 10 mol% DOPE-COC≡CH (1) mixed with 90 mol%

DOPC yes yes

I 10 mol% DOPE-COC≡CH (1) mixed with 90 mol%

DOPC yes yes

J DOPE-COC≡CH (1)

K 50 mol% DOPE-COC≡CH (1) mixed with 50 mol%

DOPC

L 10 mol% DOPE-COC≡CH (1) mixed with 90 mol%

DOPC

In this case, upon conjugation of the TLR7-Ls, an effect on the size of the liposomes was observed (after reaction the average diameter of the vesicles was ranging between 120 and 180 nm). Liposome solutions that were used as negative controls did not contain either TLR7-Ls or CuBr (solutions J, K and L ).

CpG (0 ,5 uM

)

solution E (1 uM)

solution F (1

uM )

solution G (0 .5 u

M)

solution H (0.1 u

M)

solution I ( 0.1 uM) 0

5

ng/ml IL 12p40

Figure 5.10. Induction of Dendritic cell (DC) activation by liposome-ligand conjugates (solutions G, H and I).

For comparison also the lipid-ligand conjugates are shown (solutions E and F). Values were subtracted using the liposome solutions as negative controls (solutions J, K and M). Activity was compared to CpG (positive control). Concentration of the ligands are reported in µM and were estimated considering a total conversion of lipid 1.⁴⁰

The immunogenicity of the conjugates was investigated (Figure 5.10).

Liposome-TLR7-L1 (liposome containing 7a, solutions G and H) did not show any activity and only its lipid conjugate (7a, solution E) generated a weak response. In contrast, liposome-TLR7-L2 (liposome containing 7b, solution I) appeared rather potent, while its lipid conjugate (7b, solution F) did not induce any IL-12p40 production. This result was attributed to the fact that TLR7-L1 could have folded back into the bilayer due to the flexible and short short alkylazide linker (n = 0), while the longer spacer (n = 3)kept the TLR7-L2 at the exterior of the lipid membrane, leaving the active molecule available for recognition.

In summary, an example of the use of the “click” reaction for the preparation of immunoliposomes was discussed. Although conjugation of the TLR7-L1 to the liposome surface led to a loss of activity, the immunogenicity of TLR7-L2 was strikingly enhanced upon conjunction to the liposome. Furthermore, the lipid-ligand conjugates alone were not active, demonstrating the importance of the presence of the liposomes.

5.5 Conclusions and Future Prospects

The [3+2] azide-alkyne cycloaddition has been used as a successful tool for the modification of the liposome surface. Three different examples have been discussed in detail to validate the generic nature of the “click” chemistry approach and to demonstrate that the chemical modification truly occurred at the outer membrane of the lipid vesicles.

Furthermore, easy monitoring of the progress of the reaction has been achieved using a colorimetric assay (Section 5.2) or CD spectroscopy (Section 5.3). Interestingly, β-sheet folding of the random coil (Leu-Glu)4 motif has been induced by its conjugation to the outer membrane of the liposomes. It is important to note that, when the lipidated peptide (octa-ALP) was mixed with DOPC, the preparation of well-defined β-sheet liposome surfaces has been rather difficult to achieve (see Chapter 4). This type of β-sheet loaded liposomes represent attractive three-dimensional membrane models, which could be used for investigation of CaCO3 mineralization in solution, as was done already at the air-water interface using two-dimensional β-sheet monolayers of the octa-ALP (see Chapter 3). Finally, in the last example (Section 5.4), the successful conjugation of toll-like receptor 7 ligands (TLR7-Ls) to the liposome surface was reported. The biological tests demonstrated that the immunogenecity of one of the TLR7 ligands (TLR7-L2) increased upon conjugation to the liposome outer membrane. Further optimization of the ligand properties, for example tuning the length of a spacer or increasing the liposome stability, for instance, using liposome containing phospholipids bearing units that can be polymerized by UV irradiation (i.e. diyne-phosphatidylcholine or -phosphatidylethanolamine),⁴¹ could lead to the preparation of potent synthetic liposome-based vaccines using the “click” chemistry protocol developed in this Chapter.

5.6 Experimental Section

General materials and methods. All reagents and solvents were commercial products purchased from Sigma-Aldrich B.V. or Biosolve B.V. and used as received. Lipids were purchased from Lipoid and Avanti Polar Lipids, Rink Amide resin (0.78 mmol/g), PyBOP and Fmoc protected amino acids were purchased from Novabiochem. LC-MS spectra were recorded on a JASCO RP-HPLC system, with simultaneous UV detection at 214 and 254 nm, coupled to a PE/SCIEX API 165 mass spectrometer equipped with a custom-made electronspray interface (ESI). HPLC analysis was performed on a Shimadzu system connected to a ELSD-detector. For LC-MS an analytical Vydac C4 column (Grace Vydac, 4.6 mm x 250 mm, 5 µm particle size, flow 1 ml/min) was employed. Buffers for lipopeptides: A: 25% (v/v) H2O in CH3OH; B: CH3CN and C:

1% (v/v) TFA in CH3OH. A linear gradient with increasing percentage of B was applied in 5 column volumes

gradient with increasing percentage of B was applied in 5 column volumes (CV). For RP-HPLC a preparative Vydac C4 column (Grace Vydac, 22 mm x 250 mm, 10 µm particle size, flow 25 ml/min) and the same buffer system used for LC-MS was employed. For HPLC an analytical Alltech Silica Column (Alltech, Altima Silica Column 250 x 4.6 mm, Silica 5 µm particle size, flow 1 ml/min) was employed. Buffers: A: 60% CHCl3, 34.5%

CH3OH, 5% H2O, 0.5% ammonia (25%). B: CH3OH. A linear gradient of 75 → 0% B was applied in 5 CV.

Milli-Q water with a resistance of more than 18.2 MΩ/cm was provided by a Millipore Milli-Q filtering system with filtration trough a 0.22 µm Millipak filter. Phosphate buffer was prepared by dilution (10x) of a PBS solution, containing 80 g of NaCl, 2.0 g of KCl, 14.4 g of Na2HPO4 and 2.4 g of KH2PO4 in 800 ml distilled H2O.

After adjusting the pH to 7.4 and the volume to 1L, the PBS solution was sterilize by autoclaving and diluted for the experiments. The size distributions of the liposome was measured by Photon Correlation Spectroscopy (PCS) using a Zetasizer 3000 HSA, Melvern Instruments. Temperature = 25.0 °C, Viscosity = 0.890 cP; Angle = 90.0 deg RI medium = 1.33; RI particle 1.50 + Abs. 0.00. ¹H-NMR and ¹³C-NMR spectra were measured with a Bruker AC-200 (200 and 50.1 MHz respectively). Chemical shifts are reported in ppm downfield from internal tetramethylsilane (0.00 ppm). In the case of the ¹³C spectra, the solvent peak was used as a reference (CDCl3: 77.7 ppm). Abbreviations used are s = singlet, d = doublet, dd = doublet of doublets, m = multiplet, br = broad.

1,2-Dioleoyl-sn-Glycero-3-Phosphoethanolamine-N-2-pro-2’-yne (DOPE-COC≡CH 1).

1,2-Dioleoyl-sn-Glycero-3-Phosphoethanolamine (DOPE) (1.49 g, 2 mmol) in CH2Cl2 (10 ml) was treated dropwise with a preactivated solution (0°C) of propiolic acid (3 mmol, 0.2 ml) and N-(3-Dimethylaminopropyl)- N'-ethylcarbodiimide (EDC, 3 mmol, 288 mg) in CH2Cl2 (5 ml). After stirring for 20 hrs at room temperature the reaction mixture was again dropwise treated with a preactivated solution (0°C) of propiolic acid (3 mmol, 0.2 ml) and N-(3-Dimethylaminopropyl)-N'-ethylcarbodiimide (EDC, 3 mmol, 288 mg) in CH2Cl2 (5 ml). After at total of 40 hrs, TLC analysis revealed complete conversion. The mixture was concentrated under reduced pressure and purified by silica gel column chromatography (100/0 Æ 85/10 v/v CH2Cl2/CH3OH). Evaporation of the solvents yielded 1.52 g (1.9 mmol, 95%) of product as a “sticky” foam. Finally, the product was purified by RP-HPLC using a linear gradient of 0 → 90% B in 5 CV (see general materials and methods).

ESI-MS: Rt 14.4 min, m/z = 796.8 [M+H]⁺, m/z = 818.7 [M+Na]⁺, m/z = 1592.6 [2M+H]⁺, m/z = 604.⁴²¹H NMR (200 MHz, CDCl3, δ): 7.8 (br, 1H, NH), 5.4 (m, 5H, =CH and CH glycero), 4.5-4.4 (m, 1H, CH_A glycero), 4.2-4.0 (m, 5H, CHB glycero, CHA’B’ glycero, CH2O), 3.7-3.4 (m, 2H, CH2N), 3.0-2.9 (m, 2H, ≡CH), 2.7-2.5 (m, 4H, CH2 succinyl), 2.4-2.3 (m, 4H, CH2CO oleoyl), 2.1-1.9 (m, 8H, CH2CH=CH), 1.7-1.5 (m, 4H, CH2CH2CO oleoyl), 1.4-1.2 (m, 49H, CH2 oleoyl, CH3 TEA), 1.0-0.8 (m, 6H, CH3 oleoyl). ¹³C APT NMR (50.1 MHz, CDCl3, δ): 173.6 (CO), 153.1 (COC≡CH), 129.7, 129.4 (=CH), 79.9 (C≡CH), 84.2 (C≡CH), 70.2 (CH glycero), 64.2, 63.6, 62.4 (CH2 glycero, CH2O), 40.2 (CH2N), 33.8-22.4 (CH2 oleoyl), 13.8 (CH3 oleoyl).

IR (thin film from CH2Cl2, cm^-1): 3251 (w, NH stretch, Amide A band), 3008 (CH3 anti-symmetric stretch), 2923 (CH2 anti-symmetric stretch), 2852 (CH2 symmetric stretch), 2109 (CH≡CH stretch), 1736 (C=O stretch), 1652 (C=O stretch, Amide I band), 1538 (coupled NH deformation and C-N stretch, Amide II band).

N-α-azido-N-ε-7-nitrobenzofurazan-Lys-NH2 (2). Fmoc protected Rink Amide resin (0.78 mmol/g, 0.25 mmol, 321 mg) was washed with DCM (15 ml x 3) and NMP (15 ml x 3), followed by removal of the Fmoc group using 20% piperidine in NMP (15 ml x 1 min x 1, 15 ml x 2 min x 2) and washing with NMP (15 ml x 3). N-α-Fmoc-N-ε-4-methyltrityl-Lys-OH (Fmoc-Lys(Mtt)-OH, 625 mg, 1 mmol, 4 equiv.) in NMP (10 ml) was preactivated with benzotriazole-1-yl-oxy-tris-pyrrolidino-phosphonium

hexafluorophosphate (PyBOP, 4 equiv.) and diisopropylethylamine (DIPEA, 8 equiv.) for 2 min, added to the resin and the reaction vessel was shaken for 2 hrs. Then the resin was washed with NMP (15 ml x 3) and the Fmoc group removed as described above. Diazo transfer was performed on the solid support,²⁵ using an excess of triflyl azide (5 mmol, 20 equiv.) in DCM (23 ml) in the presence of a catalytic amount of CuSO4⋅5H₂O (5 mg, 20 mmol, 0.8 equiv.) in MeOH (1 ml). The reaction mixture was shaken for 16 hrs at room temperature.

After the Kaiser test confirmed completeness of the diazo transfer, the resin was washed first with NMP (15 ml x 3) and subsequently with 0.5% DIPEA in NMP (15 ml x 3), 0.05 M diethyldithiocarbamic acid sodium salt in NMP (15 ml x 3), NMP (15 ml x 5) and finally with DCM (15 ml x 3). The methyltrityl (Mtt) group was removed by rinsing the resin with 10 ml of 95/5/1 (v/v/v) DCM/triisopropylsilane (TIS)/trifluoroacetic acid (TFA) over a period of 2 min (3x) followed by washing with DCM (15 ml x 4) and dry THF (15 ml x 3). A mixture of NBD-Cl (200 mg, 1 mmol, 4 equiv.) and 0.5% DIPEA in dry THF (20 ml) was added to the resin and shaken for 16 hrs at room temperature.⁴³ The dark brown resin was rinsed in THF (15 ml x 3), NMP (15 ml x 3) and DCM (15 ml x 3). The N-α-azido-Lys(NBD) was cleaved from the solid support by treatment with 10 ml of 9/1 (v/v) TFA/H2O 90/10 for 10 min (3x) at room temperature. After filtration of the resin, the TFA solution was concentrated under reduced pressure and a dark brown oil was obtained, which was directly used in the next step. ESI-MS: m/z = 335.2 [M+H]⁺, m/z = 357.1 [M+Na]⁺.

General procedure for the Fmoc solid phase peptide synthesis (SPPS) of Fmoc-NH-(Leu-Glu)4-NH-resin (III): Fmoc-protected peptide III was synthesized using a peptide synthesizer on a Rink Amide Resin by standard Fmoc Solid Phase Peptide Synthesis strategy. A 0.25 mmol scale was applied (loading Fmoc-protected Rink Amide resin: 0.78 mmol/g). The consecutive steps performed during each cycle were: a) Deprotection of the Fmoc group with 20% (v/v) piperidine in NMP. b) Coupling of the appropriate amino acid (Fmoc-Glu(OtBu)-OH and Fmoc-Leu-OH), applying a four-fold excess, was achieved using proper dilution of a 2 M DiPEA solution in NMP and a 0.45 M HBTU/HOBt solution in DMF and shaking for 1 hour. c) The unreacted amino functions were capped using a solution of 0.5 M acetic anhydride and 0.125 M DiPEA in NMP.

General procedure for the preparation of triflic azide:⁴⁴ A solution of NaN3 (595 mg, 9.15 mmol) in 1.5 mL of H2O was cooled to 0 °C in an ice bath and treated with 2.5 mL of CH2C12. To the resulting biphasic mixture, Tf2O (523 mg, 1.85 mmol) was added under vigorous stirring over a period of 5 min.

Then the reaction was stirred at ice bath temperature for 2 h. The organic phase was separated and the aqueous phase was extracted twice with CH2C12. The organics were combined, extracted once with saturated Na2CO3

solution and used without further purification. The total volume of the reagent solution was ~ 9 mL.

N-α-azido-(Leu-Glu)4-NH2 (4). Fmoc-NH-(Leu-Glu)4-NH-anchored to the resin (III) was prepared as described above. After Fmoc removal by treatment of the resin with 20% piperidine in NMP, diazo transfer was performed on the solid support.²⁵ An excess of triflic azide (5 mmol, 20 equiv.) in DCM (23 mL) in the presence of a catalytic amount of CuSO4⋅5H₂O (12.5 mg, 0.02 mmol, 0.8 equiv.) in MeOH (2 mL) were added.

The reaction mixture was shaken overnight at room temperature. After the Kaiser test confirmed completeness of the diazo transfer, the resin was washed first with NMP (3x) and subsequently with 0.5% DIPEA in NMP (3x), 0.05 M diethyldithiocarbamic acid sodium salt in NMP (3x), NMP (5x) and finally with DCM (3x). The resin was rinsed in THF (3x), NMP (3x) and DCM (3x). The N-α-azido-peptide was cleaved from the solid support by treatment with 10 ml of 9/1 (v/v) TFA/H2O for 1 hour at room temperature. After filtration of the resin, the TFA solution was dropped into cold Et2O and centrifuged. Freeze-drying yielded 177 mg of crude peptide. The crude