1990

Nucleic Acids Research, Vol. 18, No. 3

Oxford University Press 493

The synthesis of polyamide oligonucleotide conjugate molecules -

Jim Haralambidis*, Lucy Duncan, Karin Angus and Geoffrey W.Tregear Howard Florey Institute of Experimental Physiology and Medicine, University of Melbourne, Parkville, Victoria 3052, Australia Received October 24, 1989; Revised and Accepted December 18, 1989

ABSTRACT We have developed methods for the synthesis of peptide-oligodeoxyribonucleotide conjugate molecules in particular, and polyamide-oligonucleotide conjugates in general. Synthesis is carried out by a solid-phase procedure and involves the assembly of a polyamide on the solid support, conversion of the terminal amino group to a protected primary aliphatic hydroxy group by reaction with ce,-hydroxycarboxylic acid derivatives, and finally oligonucleotide synthesis using phosphoramidite chemistry. The conjugate molecules can be used as DNA probes, with the polyamide component carrying one or more non-radioactive markers. These conjugates also have the potential to be used as anti-sense inhibitors of gene expression, with the peptide segment acting as a targeting moiety. INTRODUCTION Synthetic oligonucleotides are widely used as probes for specific nucleic acids, by selectively hybridizing to

a

complementary

target sequence" 2. The usual method of detecting hybridization

is by the attachment of a radioactive label (eg. 32p or 35S) to the oligonucleotide. Non-radioactive labels offer a number of significant advantages including greater safety and stability of the probe. In cases where tissue sections are probed they also offer the ability to directly detect hybridization under the microscope.

A major problem with attempts to develop non-radioactive probes has been the lack of suitable synthetic procedures for the attachment of the appropriate labels. Although a number of procedures have been reported recently, they usually involve the addition of a single primary aliphatic amino3-8, or sulphydryl9-"1 group at either the 3' or, more commonly, the 5'-end of the oligonucleotide or an internal primary aliphatic amino group'2-13. This reactive moiety is then used for the subsequent attachment of a variety of labels such as fluorophores, biotin, chemiluminescent groups or enzymes. A limitation of this approach is that the probes often do not have the required sensitivity because only a single label can be attached to each oligonucleotide molecule. Higher sensitivity is required when

*

To whom correspondence should be addressed

probing nucleic acids rather than proteins, due to the much lower number of target molecules. We'4, and others'5 1 have reported procedures for the synthesis of oligonucleotides containing multiple amino groups attached to the nucleotide heterocycic base through a linker arm. However, we have found that even this procedure is limiting, since the number of nucleotides in the probe will dictate the maximum possible number of labels that can be attached. Furthermore the hybridization efficiency of the probe can be affected when a high degree of derivatization is used14. In this report we describe a general method for the preparation of oligonucleotides containing a multi-functional polyamide moiety at the 3'-end. This moiety is attached to the oligonucleotide by a stable link and its structure can be designed and manipulated to accommodate any number of labels at any distance from the oligonucleotide. Preliminary communications of this work have been publishedl8-'9.

RESULTS There are two major points to consider in the synthesis of polyamide-oligonucleotide conjugate molecules. Firstly, whether the oligonucleotide or the polyamide should be synthesized first. We decided to synthesize the polyamide first, since we planned to utilize the general procedures of solid phase peptide synthesis. These methods employ harsher conditions than those of oligonucleotide synthesis, and could cause degradation and/or premature deprotection of a preformed oligonucleotide. Secondly, the linkage between the two moieties had to be stable to the conditions of synthesis, deprotection, purification and storage. We decided to use a phosphodiester link, similar to the internal links in the oligonucleotide. In order to achieve this, we had to modify the amino terminus of the peptide so that it terminated in a primary aliphatic hydroxy group rather than an amino group.

Design of the linkage between the peptide and the oligonucleotide A number of compounds were prepared that could be used to form the linkage between the peptide and the oligonucleotide. These are essentially derivatives of a,w-hydroxycarboxylic acids, and were synthesized as outlined in Scheme I. These compounds

494 Nucleic Acids Research RCI NaO2C(CH 2)30H

Q- NH2

, NaO2C(CH2)30R

DCC

I a,b

NO2

HO

-.-o--02

C

NHC(CH 2)3 OH

(CH2)30R

Ia R=DMTr lb R=Px

I c,d,e

0 0 NHC(CH2)3 OCCHNH2

0 0

HO(CH2)6 NH2

+ O

OH3

HO(CH 26 NHC(CH2)2C02 | DMTrCI

j f,g,h

0

RO(CH2)6 NHC(CH 2) 2C02 DCC HO-

0-NH6(CH2)30O6CHNH['CCHNH] in CH3 [ R

NO2

I i,b 0

RO(CH2)6 NHC(CH2)

2CO-

NO2

2

SCHEME I

have the hydroxyl group protected as either a di-(p-methoxyphenyl)phenymethyl (dimethoxytrityl) or 9-phenylxanthen-9-yl (pixyl) ether and the carboxy terminus activated as the p-nitrophenyl active ester. They can react with the amino terminus of the peptide, giving rise to a protected primary aliphatic hydroxy group as the new terminus. The hydroxy protecting groups were chosen so that they are similar to those used in conventional DNA synthesis. Reaction of the deprotected primary hydroxy group with the first nucleoside phosphoramidite will give rise (after oxidation) to a phosphate triester initially, which after deprotection will be converted to a phosphodiester. In this way, the link between the peptide and oligonucleotide moieties contains only functionalities that are normally found as internal links in peptides and oligonucleotides, i.e. amides and phosphodiesters. Of the three compounds prepared as linkers, lb is preferred since it is a crystalline compound that can be made analytically pure, unlike la and 2 that are oils. It does have the disadvantage however that at low resin loadings the colorimetric measurement of the pixyl group after acid release is unreliable, and so quantitative ninhydrin tests are carried out to determine the extent of amino reaction. We have also found that these compounds are excellent for the initial derivatization of the Aminopropyl Controlled Pore Glass (AP-CPG) resin, giving a readily quantifiable loading of protected hydroxy group on to which (after deprotection) the polyamide is attached. Unreacted amino groups can be capped by acetylation after addition of the linker. We also tested the use of -y-butyrolactone in place of the above mentioned compounds. However, we found that although the use of this reagent in the initial derivatization of the CPG did not give rise to any problems, its use as the linker between the peptide and the oligonucleotide gave rise to byproducts, probably because of the long time and elevated temperatures (7 days at 60°C) necessary for reaction. Synthesis of peptide-oligonucleotide conjugates We chose to synthesize a conjugate containing the peptide (Ala-

N HC(CH 2)3 0CCHNHiCCHNH

C(CH2)30H

; DNA SYNTHESIS

SCHEME 1 Reagents: (a) 1 or 2, DMAP; (b) C12CHCO2H/CH2C12; (c) (Boc-Ala)20, DMAP; (d) TFA/H20; (e) Et3N/CH2CI2; (f) Fmoc-Xaa-OPfp, HOBt; (g) piperidine/DMF; (h) repeat steps f and g n times; (i) 1 or 2, HOBt.

Lys-)5Ala to illustrate the synthesis of such compounds. We used the Fmoc (fluoren-9-ylmethoxycarbonyl) peptide synthesis methodology20 to prepare the peptide segment. The peptide was designed to include multiple lysine residues that could subsequently be used as attachment sites for non-radioactive labels and alanine residues that served as spacers between the lysines. We derivatized Aminopropyl CPG by reaction with la (Scheme H). The amount of hydroxy group introduced could be quantitated by colorimetric assay of the dimethoxytrityl cation released upon acid treatment (trityl test). Following deprotection of the hydroxyl group with dichloroacetic acid (DCA), the first amino acid (Ala) was introduced as the tert-butoxycarbonyl (Boc) symmetrical anhydride, with dimethylaminopyridine (DMAP) as catalyst. The Boc amino protecting group was then removed by trifluoroacetic acid (TFA) and the amines neutralized with triethylamine in dichloromethane. Standard Fmoc peptide synthesis methodology was then utilized, using the pentafluorophenyl (Pfp) active esters of N-ci-Fmoc protected amino acids (lysine and alanine) as the monomers. Following removal of the last N-az-Fmoc amino protecting group, the solid support was reacted with 2. This substrate was then used for oligonucleotide synthesis in an Applied Biosystems Inc. automated DNA synthesizer, using methyl N,N-diisopropyl nucleoside phosphoramidites21,22. The first phosphoramidite was linked to the deprotected terminal aliphatic hydroxyl group. This reaction was quantitative as assayed by the trityl test. Following the synthesis of the 30mer oligonucleotide d(GGGCTTCACAACATCTGTGATGTCAGCAGG) (KPIB, complementary to a region of mouse kallikrein mRNA that is common to all the mouse kallikreins23), the methyl protecting groups on the phosphotriesters were removed using thiophenoxide ion, and the acid sensitive protecting groups (e-Boc on Lys residues and 5'-DMTr) removed by treatment with 90% TFA/10% ethanedithiol'4. Normal

Nucleic Acids Research 495 H2N(CH20,C02H + FmocNHS

0

FmocNH(CHF5CO H DCC

A

B C D E F G H

ROH

FmocNH(CH2)5COR

H2N(CH2)5C°2H

3 R 4 R

= =

Pfp NHS

FmocNH(CHK5CONH(CH2)5CO2H DCC

5

HOPfp

FmocNH(CH2)5CONH(CHK)5CO2Pf p 6

SCHEME HI

workup with aqueous ammonia'4'24 gave the peptideoligonucleotide conjugate. The ammonia treatment removes the standard protecting groups on adenine, guanine and cytosine residues, and also cleaves the peptide-ester bond to the CPG resin, releasing the oligonucleotide-polyamide conjugate probably as a mixture of the C-terminal amide and carboxylate. Analysis of the crude product by polyacrylamide gel electrophoresis (PAGE) gave the product as the major band (Figure 1, lane B), running slower than the underivatized 30mer (Figure 1, lane A). It was purified by preparative PAGE. Amino acid analysis of the product gave the expected ratio of 6 Ala:5 Lys, with 1 mole of (AlaLys-)5Ala per mole of KPIB (Table I). This indicates that the peptide bonds are stable to oligonucleotide synthesis and deprotection. The product was resistant to snake venom phosphodiesterase (blocked 3'-end) and it was only partially digested (= 10 nucleotides from 5'-end) by spleen phosphodiesterase. This enzyme is known to be very sensitive to secondary structure25, and the interaction of the positively charged lysine residues on the peptide with the phosphodiester backbone may be inhibiting further digestion of the DNA. Appropriate conditions were found however for the complete digestion of the product with PI nuclease, which gave the same HPLC profile of nucleoside and nucleotides as digestion of the underivatized oligonucleotide. A number of other peptide oligonucleotide conjugates were prepared using these procedures, mainly containing various numbers of lysine residues with a variable number of spacer amino acids. Synthesis of non-peptide polyamide oligonucleotide conjugates It was realised that the use of non-peptidic amino acids would introduce greater flexibility in the design of the molecular architecture of the polyamide. Even though lysine residues were found to be quite appropriate for the attachment of labels (see following paper), the natural et-amino acids provided too small a distance between peptide bonds to act as efficient spacers between the lysine residues. A spacing of appropriate dimensions would require multiple residues of at-amino acids, which could give rise to synthetic problems as the size of the peptide grew too large. However, any c,w-aminocarboxylic acid could act as a spacer. The readily available 6-aminohexanoic acid (eAhx) was

Figure 1. Autoradiograph of a 20% polyacrylamide gel on which various conjugates were electrophoresed. Lane A, KPIB; lane B, KPIB-LL-(Ala-Lys-)5 Ala; lane C, KPIB-LL-(eAhx-)4Lys-Ala, synthesized by the Fmoc method; lane D, KPIB-LL-(EAhx-)4Lys-Ala, Boc method; lane E, KPIB-SL-(EAhx-Lys-)1OAla; lane F, KPIB-SL-[(eAhx-)2Lys-]1OAla; lane G, KPIB-SL-Ala-[Lys-(EAhx-)4]9Lys-Ala; lane H, KPLB. LL stands for the linker derived from 2 and SL that from 1.

Table I. Amino acid analysis results of oligonucleotide-polyamide conjugates

KPIB-LL-(Ala-Lys-)5Ala KPIB-LL-(EAhx-)4Lys-Ala1 KPIB-LL-(EAhx-)4Lys-Ala2 KPIB-SL-Ala-[Lys-(EAhx-)4]9Lys-Ala

Ala

Lys

6.1 1.3 0.9 2.4

4.9 1.2 1.1 10.4

EAhx nPeptide nKPIB 1.1 3.5 0.9 4.0 1.2 35.2 0.8

' By Fmoc method

2

By Boc method

chosen as the standard unit, but any other similar aminocarboxylic acid could be used. Initially, the N-Fmoc pentafluorophenyl active ester derivatives 3 and 6 were synthesized and used successfully using standard Fmoc peptide synthesis methodology. The dimer 6 could be prepared in good yield from the monomer 4 and 6-aminohexanoic acid (Scheme III). It was found that the use of the N-hydroxysuccinimide (NHS) active ester 4 in this reaction gave a much higher yield of the acid 5 (100%) than the Pfp derivative 3 (50%). The acid 5 was also used directly in solid phase synthesis by utilizing the BOP (benzotriazol-1-yl-oxy-tris(dimethylamino)phosphonium hexafluoroborate) methodology26,27. This proved to be a very efficient way of introducing this spacer avoiding the need to prepare the active ester. To illustrate the synthesis of an oligonucleotide containing a single primary aliphatic amino group well removed from the oligonucleotide segment, the polyamide (eAhx-)4Lys(Boc)-Ala was synthesized using the methodology described in the previous section, and utilizing two couplings of the active ester 6 to provide the spacing to the oligonucleotide. KPIB was then synthesized, after addition of the linker 2. It was deprotected as before and on PAGE analysis gave the product as the major band, running slower han KPIB (Figure 1, lane C, and Table I). This was found

496 Nucleic Acids Research to be a convenient way of introducing labels while the polyamnue is still on the solid support (see following paper), but we sought an even more straightforward way to prepare the same compound. For this we chose to use solely Boc chemistry for the synthesis of the polyamide. We used a-Boc-e-Fmoc-Lys-OPfp (or the corresponding carboxylic acid with BOP methodology) as the lysine derivative and Boc-(eAhx-)2OPfp (7) as the spacer. Boc-(eAhx-)2OPfp could be prepared from the corresponding monomer Boc-eAhx-OPfp (8) by reaction with 6-aminohexanoic acid followed by acidification, extraction, concentration and recondensation of the product Boc(-EAhx)2 and the released pentafluorophenol with DCC. The protected polyamideoligonucleotide conjugate was deprotected in exactly the same manner as a normal oligonucleotide, to give the same product as before (Figure 1, lane D and Table I). In this case the Fmoc protecting group on the lysine residue is cleaved during the ammonia deprotection step. A series of conjugates containing long polyamide moieties with ten lysine residues in each were also prepared by the Fmoc method, using a manual peptide synthesizer for polyamide synthesis. In this series the spacing between the lysine residues was varied from four to one aminohexanoyl residues. PAGE analyses of these compounds, after purification, are shown in Figure 1, lanes E-G.

DNA synthesis with methyl versus cyanoethyl protected nucleoside phosphoramidites The use of the 0-methyl protected phosphoramidites in the synthesis of these conjugate molecules proceeds with few apparent side reactions, giving the desired conjugate as the major species in the crude product. However, we found that when the 0cyanoethyl phosphoramidites were used the yield of product was low, sometimes zero, with a large number of smaller molecules replacing the main product. This was the case irrespective of the amino acid sequence of the peptide and oligonucleotide components. After a number of attempts to find the reason for this difference, it was found that the key variable was the length of the capping time during oligonucleotide synthesis. The protocols recommended by Applied Biosystems for use on their DNA Synthesizer with the methyl protected phosphoramidites have a 120 sec capping wait step with the DMAP/Ac2O solution, whereas those for use with the cyanoethyl protected phosphoramidites have a 5 sec wait step. When we changed this latter wait step to 120 sec, we obtained the same result as when using the methyl protected phosphoramidites (Figure 2). We then tried a variety of intermediate times so as to minimize any removal of the cyanoethyl groups by DMAP and finally decided to use a 60 sec wait step. It appears that extended contact with DMAP during the capping step is required for good results. A possible explanation for this observation is that the phosphoramidites react with the carbonyl oxygen of the peptide bonds, and DMAP is required to reverse this reaction. Such reaction is known to take place with the amides on the protected heterocyclic bases of the nucleotides28'29. Therefore, in all work in which cyanoethyl protected nucleoside phosphoramidites are used, a 60 or 120 sec capping wait step was employed.

DISCUSSION The use of a linking synthon that essentially converts the Nterminus of a polyamide from a primary aliphatic amino group to a primary aliphatic hydroxy group has enabled us to synthesize

A

'I.

I.

1

Figure 2. Autoradiograph of a 20% polyacrylamide gel on which the crude product mixtures from synthesis with variable capping times were 5'-end labelled and electrophoresed. Lanes A and H have normal KPLOB oligonucleotide. Lanes B -G are KPIB-LL-(A1a-LyS-)5Ala conjugates except for lane F which is KPIB-SL(EAhx-)4(LyS-)2,AJa. In lane B the synthesis was carried out using 0-methyl protected phosphoramidites. The products in lanes C - G were synthesized using 0-cyanoethyl protected phosphoramidites with the following capping times: C, 5 sec; D, 15 sec; E, 30 sec; F, 1 min; G, 2 min. The syntheses in lanes B- E and G were carried out on the same batch of peptide resin solid support.

variety of stably linked polyamide-oligonucleotide conjugates. Both Boc and Fmoc solid phase peptide synthesis methodologies have been used for the synthesis of the polyamide, which usually contains one or more lysine residues, suitable for the attachment of non-radioactive labels. The use of non-peptidic amino acids has added a much greater flexibility to the design of the polyamide component, and gives control over both the inter-lysine and thus inter-label spacing and the distance to the oligonucleotide. Once the polyamide has been synthesized on the solid support and the linker added, aliquots of this can be used for the synthesis of any oligonucleotide without having to derivatize each oligonucleotide separately. In its simplest form this means that oligonucleotides containing a single primary aliphatic amino function as part of a polyamide moiety attached at the 3'-end at an appreciable distance from the oligonucleotide can be synthesized on an Automated DNA Synthesizer without having to perform any additional steps. These will be very useful for the attachment of larger molecules to the probes, for example, enzymes. Finally, peptide-oligonucleotide conjugates can also be envisaged where the peptide is a specific sequence that will impart certain desirable properties to the oligonucleotide, for example, to facilitate its passage through cell membranes, in cases where a physiological or pharmacological effect of the oligonucleotide a wide

is desired.

EXPERIMENTAL General 'H and 13C NMR spectra were recorded on a Bruker AM300 at 300 and 75 MHz respectively with tetramethylsilane as internal

Nucleic Acids Research 497 reference. Dimethylformamide (DMF) was distilled under reduced pressure and used within 1-2 days. Pyridine was distilled from potassium hydroxide and stored over 5A molecular sieves. Microanalyses were determined by Amdel, Melbourne. Melting points were determined in open ended capillaries on an Electrothermal Melting Point Apparatus. Amino acid analyses were carried out on a Beckman System 6300 Amino Acid Analyser. Snake venom phosphodiesterase, bovine spleen phosphodiesterase and P1 nuclease were from Pharmacia. Infrared spectra were recorded on a Perkin-Elmer 983G infrared spectrophotometer. HPLC was carried out on an Altex system, using a Vydac C18 51t 25 cm x4.6 mm column. Buffers used were A, 0.1 M triethylammonium acetate, pH 7.0 and B, 0.1 M triethylammonium acetate containing 45% CH3CN, at pH 7.0. Flash chromatography30 was carried out using Merck Kieselgel # 9385.

p-Nitrophenyl 4-[di-(p-methoxyphenyl)phenyImethyloxy]butyrate (la) Sodium 4-hydroxybutyrate (1.26 g, 10 mmol) and 4,4'-dimethoxytrityl chloride (DMTrCl) (3.39 g, 10 mmol) were stirred in 30 mL of pyridine for 16 h, p-nitrophenol (1.39 g, 10 mmol) and dicyclohexylcarbodiimide (DCC) (2.06 g, 10mmol) were added and stirred for a further 2 days. The reaction mixture was filtered, the solution was concentrated and flash chromatographed on 70 g of silica gel with 25% EtOAc/petroleum ether to give a light yellow oil (5.0 g, 95 %). 1H NMR (CDCl3) 6 2.04 (m, 2H, H3), 2.7 (t, J=7.2 Hz, 2H, H2), 3.2 (t, J=5.9 Hz, 2H, H4), 3.77 (s, 6H, OCH3), 6.8-7.5 (m, 15H, ArH), 8.2 (d, J=9.2 Hz, 2H, PhNO2 m-H). '3C NMR (CDCl3) 6 25.2 (C3), 31.6 (C2), 55.2 (OCH3), 62.0 (C4), 86.0 (CAr3), 113.1 126.8, 127.1, 127.8, 127.9, 128.1, 129.1, 130.0, 136.2, 145.0, 158.4 (DMTr), 122.4, 125.1, 145.2, 155.4, (PhNO2), 171.1 (CO2). The product contained some EtOAc solvent impurity. p-Nitrophenyl 4-[(9-phenylxanthen-9-yl)oxy]butyrate (lb) This was prepared using the same method used to prepared la by substituting pixyl chloride for dimethoxytrityl chloride, to give lb in 80% yield, mp 130-130.5 °C (EtOAc). 'H NMR (CDCl3) 6 1.98 (m, 2H, H3), 2.7 (t, J=7.3 Hz, 2H, H2), 3.0 (t, J=5.8 Hz, 2H, H4), 7.0-7.5 (m, 15H, ArH), 8.2 (d, J=7.1 Hz, 2H, PhNO2 m-H). 13C NMR (CDCl3) 6 25.0 (C3), 31.5 (C2), 61.8 (C4), 75.4 (CPh3), 116.3, 123.2, 123.5 126.4, 126.6, 127.9, 129.1, 129.4, 148.9, 151.3 (Px C), 122.4, 125.1, 145.2, 155.4 (PhNO2 C), 171.0 (CO2). Anal. Calcd for C29 H23 NO6: C, 72.3; H, 4.8; N, 2.9. Found: C, 72.0; H, 4.4; N, 3.2.

p-Nitrophenyl 3-[6-(di-(p-methoxyphenyl)phenylmethyloxy)hexylcarbamoyl]propanoate (2) A solution of succinic anhydride (1.0 g, 10 mmol) and 6-aminohexanol (1.17 g, 10 mmol) in pyridine (10 mL) was stirred for 4 d. DMTrCl (3.39 g, 10 mmol) was then added, it was stirred for a further 4 h, andp-nitrophenol (1.39 g, 10 mmol) and DCC (2.06 g, 10 mmol) added and was stirred for a further 2 d. The reaction mixture was filtered, the solution concentrated and flash chromatographed on 100 g of silica gel with 50% EtOAc/petroleum ether to give a light yellow oil (4.09 g, 64%). 1H NMR (CDC13) 6 1.2-1.7 (m, 8H, CH2), 2.5 (t, J=6.5 Hz, 2H, CH2), 2.93 (t, J=6.5 Hz, 2H, CH2), 3.01 (t, J=6.4 Hz, 2H, CH2), 3.2 (t, J=6.4 Hz, 2H, CH2), 3.76 (s, 6H, OCH3), 6.8 - 7.5 (m, 1SH, AMH), 8.2 (d, J=9.2 Hz, 2H, PhNO2 m-H).

13C NMR (CDCl3) 6 25.83, 26.69, 29.53, 29.87, 30.55 (CH2), 39.68 (CH2NHCO), 55.14 (OCH3), 63.16 (DMTrOCH2), 85.60 (CAr3), 112.91, 126.53, 127.64, 127.70, 127.79, 128.10, 129.07, 129.95, 136.60, 145.32, 158.26, (DMTr C), 122.4, 125.1, 145.31, 155.8 (PhNO2 C), 170.49, 170.75 (C=O). The compound contained some EtOAc solvent impurity that could not be easily removed. Pentafluorophenyl N-(fluoren-9-ylmethoxycarbonyl)-6aninohexanoate (Fmoc-EAhx-OPfp,(3) 6-Aminohexanoic acid (2.62 g, 20 mmol) and Na2CO3 (5.30 g, 50 mmol) were dissolved in 60 mL of H20, 25 mL of dioxan added, followed by N-(fluoren-9-ylmethoxycarbonyloxy)succinimide (Fmoc-NHS) (6.75 g, 20 mmol). It was stirred vigorously for 16 h. The cloudy reaction mixture was then poured into 1.2 L of H20 to give a clear solution. This was extracted with EtOAc (2 x 300 mL) and the rapidly stirring aqueous layer acidified to pH 3 using = 10 mL of conc. HCI, to give a voluminous precipitate. This was kept at 4°C for 16 h, and was then filtered to give 6.05 g (86%) of Fmoc-eAhx. To a solution of Fmoc-EAhx (1.77 g, 5 mmol) and pentafluorophenol (1.01 g, 5.5 mmol) in 8 mL of DMF was added a solution of DCC (1.03 g, 5 nmmol) in 2 mL of DMF. This was stirred for 16 h, filtered, and the filtrate evaporated in vacuo to dryness giving the crude ester. It was recrystallized from 95% EtOH/1% AcOH (= 10 mL) to give 2.41 g (93%) of white needles, mp 128-129°C. 'H NMR (CDCl3) 6 1.4-1.8 (m, 6H, CH2), 2.7 (t, J = 7.2 Hz, 2H, CH2CO2), 3.2 (m, 2H, NHCH2), 4.2 (t, J = 6.7 Hz, 1H, Fmoc CH), 4.4 (d, J = 6.8 Hz, 2H, Fmoc CH2), 4.8 (bs, 1H, Fmoc NH), 7.3-7.8 (m, 8H, Fmoc ArH). 13C NMR (CDCl3) 6 24.36 (C4), 25.93 (C3), 29.62 (C5), 33.18 (C2), 40.73 (C6), 47.31 (Fmoc CH), 66.54 (Fmoc CH2), 119.98, 125.00, 127.02, 127.67 (Fmoc Aromatic CH), 141.34, 143.99 (Fmoc Aromatic C), 156.46 (Fmoc C=O), 169.36 (ester C02). IR (KBr) 1687 (Fmoc C=O), 1782 (ester C=O) cm'-. Anal. Calcd for C27HIIN04F5: C, 63.8; H, 2.2; N, 2.8. Found: C, 63.9; H, 1.8; N, 3.2. N-[N-(Fluoren-9-ylmethoxycarbonyl)-6-aminohexanoyl]-6aminohexanoic acid (Fmoc(-EAhx)2,(5) To a solution of Fmoc-EAhx (prepared as above, 1.77 g, 5 mmol) and N-hydroxysuccinimide (0.575 g, 5 mmol) in 8 mL of DMF was added a solution of DCC (1.03 g, 5 mmol) in 2 mL of DMF. It was allowed to stir for 16 h, filtered, and the fitrate evaporated in vacuo to a syrup. This was recrystallized from isopropanol (= 10 mL) to give 1.94 g (86%) of 4. To a solution of 4 (0.912 g, 2 mmol) in 10 mL dioxan was added dropwise a solution of 6-aminohexanoic acid (0.524 g, 4 mmol) and Na2CO3 (0.424 g, 4 mmol) in 10 mL of H20. The resulting suspension was stirred vigorously for 48 h, and was then poured into 100 mL of H20 to give a clear solution. The pH of this rapidly stirring solution was reduced to 3 by the dropwise addition of 10 mL of 1 M KHSO4. Voluminous precipitate formed, it was kept at 4°C for 24 h, and then filtered to give a quantitative yield of the acid. This crude product was recrystallized from EtOAc to give 0.679 g (73%) of a white powder, mp 116.5-117.5°C. 'H NMR (CD30D) 6 1.3-1.7 (m, 12H, internal CH2), 2.1 (t, J = 7.4 Hz, 2H, CH2CONH), 2.3 (t, J = 7.3 Hz, 2H, CH2CO2H), 3.0-3.2 (m, 4H, FmocNHCH2 and CH2CONHCH2), 4.2 (t, J=6.8 Hz, 1H, Fmoc CH), 4.3 (d, J=6.8 Hz, 2H, Fmoc CH2), 7.2-7.8 (m, 8H, Fmoc ArH). IR (KBr) 1632 (amide C=O), 1696 (Fmoc

498 Nucleic Acids Research C =0), 1718 (acid C=0) cm-t. Anal. Calcd for C27H34N205: C, 69.5; H, 7.4; N, 6.0. Found: C, 69.2; H, 7.3; N, 5.8.

Pentafluorophenyl N-[N-(fluoren-9-ylmethoxycarbonyl)-6aminohexanoyl]-aminohexanoate (Fmoc-(EAhx-)2OPfp, 6) To a solution of 5 (233 mg, 0.5 mmole) and pentafluorophenol (101 mg, 0.55 mmole) in 1 mL of DMF was added DCC (103 mg, 0.5 mmole). This was allowed to stir for 2 d, was then filtered, the filtrate evaporated in vacuo to a cream solid which was recrystallized from 95 % EtOH/ 1 % AcOH (1 mL) to give 180 mg (57%) of pure 6, mp 126-127°C. 'H NMR (CDCl3) 6 1.3-1.8 (m, 12H, internal CH2), 2.2 (t, J=7.4 Hz, 2H, CH2CONH) 2.7 (t, J-7.3 Hz, 2H, CH2CO2), 3.1-3.3 (m, 4H, FmocNHCH2 and CH2CONHCH2), 4.2 (t, J=6.9 Hz, 1H, Fmoc CH), 4.4 (t, J = 6.9 Hz, 2H, Fmoc CH2), 4.8 (bs, 1H, Fmoc NH), 5.5 (bs, 1H, CONH), 7.2-7.8 (m, 8H, Fmoc ArH). IR (KBr) 1639 (amide C =0), 1689 (Fmoc C =0), 1785 (ester C=O) cm-l. Anal. Calcd for C33H33F5N205: C, 62.7; H, 5.3; N, 4.4. Found: C, 62.5; H, 5.1; N, 4.7.

Pentafluorophenyl N-tert-butoxycarbonyl-6-aminohexanoate (Boc-EAhx-OPfp, 7). To a solution of N-Boc-6-aminohexanoic acid (4.78 g, 20.8 mmol) and pentafluorophenol (3.68 g, 20 mmol) in 50 mL of EtOAc was added DCC (4.12 g, 20 mmol). This was allowed to stir for 16 h, was filtered and the filtrate evaporated in vacuo to a syrup. On standing, this crystallized to give 7.46 g (94%) of the ester. Recrystallization from isopropanol/ 1 % acetic acid gave 6.26 g of white needles, mp 81-83°C. 'H NMR (CDCl3) 6 1.4-1.6 (m, 15H, Boc CH3 and internal CH2), 1.8 (m, 2H, CH2CH2CO2), 2.67 (5, J = 7.3 Hz, 2H, CH2CO2), 3.1 (m, 2H, NHCH2), 4.5 (b, 1H, NH). 13C NMR (CDC13) 6 24.38 (C4), 26.00 (C3), 28.39 (Boc CH3), 29.70 (CS), 33.20 (C2), 40.28 (C6), 79 (Boc C), 156 (Boc C =0), 169 (C 1). Anal. Calcd for C17H20N04F5: C, 51.4; H, 5.1; N, 3.5. Found: C, 51.5; H, 5.0; N, 3.4. Pentafluorophenyl N-(N-tert-butoxycarbonyl-6-amino-

hexanoyl)-6aminohexanoate (Boc-(Ahx-)20Pfp, 8) To a solution of 6-aminohexanoic acid (1.31 g, 10 mmol) in 5 mL of 1 M NaOH (5 mmol) and 5 mL H20 was added a solution of 7 (1.99 g, 5 mmol) in 10 mL of dioxan. The resulting fine suspension was stirred vigorously for 3 d, by which time it was clear. It was added to 200 mL of H20, and the pH decreased to 3.5 by the dropwise addition of = 10 mL of 1 M KHSO4, the resulting solution extracted with EtOAc (3 x 100 mL), dried (Na2SO4), and concentrated in vacuo to 3 mL. Another 10 mL of EtOAc was then added, followed by DCC (1.03 g, 5 mmol). It was stirred for 16 h, was filtered, the filtrate evaporated to dryness in vacuo, and the product was recrystallized from EtOH/H20 containing 1 % AcOH ( = 10ImL), to give 1.75 g (67%) of white needles, mp 88-89°C. 'H NMR (CDC13) 6 1.3-1.9 (m, 2H, CH2 and Boc CH3), 2.2 (t, J = 7.5 Hz, 2H, CH2CONH), 2.7 (t, J = 7.3 Hz, 2H, CH2CO2), 3.1 (m, 2H, Boc NHCH2), 3.3 (m, 2H, CONHCH2), 4.6 (bs, 1H, Boc NH), 5.6 (bs, 1H, CONH). '3C NMR (CDCl3) 6 24.35, 25.31, 26.10, 26.41, 29.31, 29.83, 33.17, 36.62, 39.14, 40.35 (CH2), 28.44 (Boc CH3), 79.12 (Boc central C), 156.05 (Boc C =0), 169.39 (ester C =0), 172.84 (amide C =0). Anal. Calcd for C23H3 I N205F5: C, 54. 1; H, 6. 1; N, 5.5. Found: C, 53.9; H, 5.9, N, 5.8.

Solid phase synthesis Solid phase synthesis using the Fmoc methodology was carried out on a Cambridge Research Biochemicals (CRB) manual peptide synthesizer. All other reactions were performed in a reaction cell previously described'4. DNA synthesis was carried out on an Applied Biosystems 380A DNA Synthesizer. Derivatization of the CPG To Aminopropyl Controlled Pore Glass (AP-CPG, Fluka, pore size 500 A, 0.5 g, 20 yimole of amino groups) was added either la, lb or 2 (250 ,imol) and dimethylaminopyridine (DMAP) (30.5 mg, 250 ymol) in 2 mL of DMF. This was either shaken for 3 h or left standing for 16 h. The CPG was then washed (DMF, CH2Cl2) and dried. The degree of functionalization was quantitated by spectrophotometric assay of the dimethoxytrityl or pixyl (X = 445 nm, E= 4770 M-') cation released on acid treatment of a small amount of CPG. Residual amino groups ( = 10-20%) were then acetylated by treating the CPG with acetic anhydride (0.5 ml, 2.5 mmol) and DMAP (50 mg, 0.4 mmol) in pyridine (2 mL) for 15 min. When -y-butyrolactone was used to derivatize the CPG, CPG (0.5 g) and -y-butyrolactone (3 mL) were placed in an oven at 60°C for 7 days. The extent of reaction was monitored by the disappearance of the primary amino groups on quantitative ninhydrin analysis3'.

Coupling of the first amino acid The CPG obtained in the previous step was treated with 3% dichloroacetic acid in CH2Cl2 (2 x S min) and washed (CH2Cl2). It was then reacted with a solution of N-Boc-alanine symmetrical anhydride and DMAP (0.2 M in each) in DMF for 20 h. After washing, the extent of reaction was determined by quantitative ninhydrin assay on a small amount of deprotected CPG. Any residual hydroxyl groups were then acetylated as before. Peptide synthesis The Boc group was removed from the first amino acid by treatment with 90% trifluoroacetic acid (TFA)/H20 (30 min), followed by washing (CH2C12), neutralization (20% triethylamine/CH2Cl2), washing (CH2Cl2) and drying. Further peptide synthesis was then carried out on the manual CRB peptide synthesizer, using standard Fmoc chemistry20, by utilizing a fourfold molar excess of Fmoc-amino acid pentafluorophenyl ester and 1-hydroxybenzotriazole (HOBt) in DMF. Addition of linker synthon to peptide Following deprotection of the terminal amino group, the CPG was reacted with la, lb or 2 (0.2 mmol) and 1-hydroxybenzotriazole (0.2 mmol) in DMF (0.5 mL) for 16 h. Residual amino groups were acetylated and the CPG used for DNA synthesis. Reaction with la or lb gives rise to the short linker (SL) and with 2 the long linker (LL). When -y-butyrolactone was used, the CPG (100 mg) was reacted with 2 mL of -y-butyrolactone at 60°C for 7 days.

Polyamide synthesis Synthesis of polyamides containing a,w-aminocarboxylic acid residues was carried out as for normal peptide synthesis but using four equivalents of N-Fmoc amino acid active esters 3 or 6 at the appropriate stage. Synthesis was also carried out using the N-Fmoc amino acid 5 utilizing the BOP (benzotriazol-1-yl-

oxy-tris-(dimethylamino)phosphonium hexafluoroborate)

Nucleic Acids Research 499

procedure26'27. Briefly, a fourfold molar excess of 5, BOP reagent, N-methylmorpholine and HOBt was used, in DMF. Alternatively, in cases were Boc methodology was used, Boc protected amino acid active esters 7 and 8 were used, utilizing a fourfold molar excess of the active amino acid and HOBt in DMF for 0.5 h. Deprotection of conjugates Following synthesis of the desired oligonucleotide on an Applied Biosystems 380A DNA synthesizer'4 using either methyl or cyanoethyl protected nucleoside phosphoramidites, the conjugate was deprotected as required. If the lysine residues contained a side chain Fmoc group, then no extra steps were necessary, as the N-Fmoc group is removed during the aqueous ammonia base deprotection step. If the conjugate contained Boc groups on the lysine side chains then the syntheses were carried out leaving the oligonucleotide fully protected, the solid support transferred to the manual reaction cell and treated with 90% TFA/10% ethanedithiol (1 mL) for 5 min, with shaking. It was then rinsed (CH2Cl2) and neutralized by repeated washes with 20% Et3N/CH2Cl2, followed by rinsing (CH2Cl2). Normal oligonucleotide cleavage and deprotection with aqueous ammoma then followedl4'24, to give the fully deprotected conjugate. In cases where methyl protected phosphoramidite nucleosides were used, a thiophenoxide deprotection step'4'32 was used prior to the amino acid side chain deprotection.

Purifi'cation of conjugates Purification was carried out by polyacrylamide gel electrophoresis (PAGE) using 10% gels2. Purification of the single lysine containing conjugate prepared by Boc chemistry was conveniently carried out by reverse phase HPLC on a Vydac C18 column, since the 5'-DMTr group of this conjugate is intact. The DMTrcontaining conjugate was initially purified using the following conditions: isocratic at 33.3% B for 20 min, and then a gradient to 66.6% B over 30 min. DMTr-KPIB-SL-(eAhx-)4Lys-Ala elutes at 44.0 min. Detritylation of the eluate (equal volume of acetic acid, 15 min) and rechromatographing on a gradient of 0 to 66.6% B over 30 min gave the pure product, eluting at 26.0 min.

Analysis of conjugates Conjugates were routinely analysed by a number of methods. A sample was 5'-end labelled2 and subjected to PAGE, on 20% gels, to determine homogeneity and apparent size. Amino acid analysis was used to determine the amounts and ratios of amino acids in the conjugate. Digestion with nuclease enzymes and HPLC of the digest was used to determine the nucleoside/nucleotide composition. Digestions with PI nuclease (conjugate containing 1 ytg of DNA in 20 ,ul of H20, 2 ,ul of 0.5 M NaOAc, pH 6.0 and 5 1tg of nuclease PI (5 yd, in 0.05 M NaOAc, pH 6.0, 50% glycerol, stored at -20°C)) at 37°C for 30 min gave complete digestion. HPLC was carried out using a gradient of 0 to 33.3% Buffer B over 30 min. Digestion with spleen phosphodiesterase (conjugate containing 1 yg of DNA in 20 1I H20, 2.5 p.l of 30 mM KH2PO4, pH 6.0, 2.5 !l of 3 mM EDTA, 0.025 units of bovine spleen phosphodiesterase (2 IL, in 1.5 mM KH2PO4, pH 6.0, 50% glycerol, stored at -20°C)) at 37°C for 16 h generally gave only partial digestion, from the

5'-end. Snake venom phosphodiesterase digestion was carried out with 1 Ag of enzyme in 0.05 M K2HPO4, pH 10.0 at 45°C for 4 h. The conjugates were resistant to these conditions (PAGE) with only a small amount of digestion taking place.

ACKNOWLEDGEMENTS We thank Scott Pownall for the radioactive labelling of conjugates and Dr. John Wade and Professor John Swan for valuable discussions. This work was supported by the National Health and Medical Research Council of Australia.

REFERENCES 1. Itakura, K., Rossi, J.J., Wallace, R.B. (1984) Ann. Rev. Biochem., 53, 323-356. 2. Penschow, J.P., Haralambidis, J., Aldred, P., Tregear, G.W., Coghlan, J.P. (1986) Methods in Enzymology, 124, 534-548. 3. Agrawal, S., Christodoulou, C., Gait, M.J. (1986) Nucleic Acids Res. 14, 6227-6245. 4. Chollet, A., Kawashima, E.H. (1985) Nucleic Acids Res. 13, 1529- 1541. 5. Wachter, L., Jablonski, J-A., Ramachandran, K.L. (1986) Nucleic Acids Res. 14, 7985-7994. 6. Sproat, B.S., Beijer, B., Rider, P. (1987) Nucleic Acids Res. 15, 6181-6196. 7. Li, P.; Medon, P.P.; Skingle, D.C.; Lanser, J.A.; Symons, R.H. (1987) Nucleic Acids Res. 15, 5275-5286. 8. Smith, L.M.; Fung, S.; Hunkapiller, M.W.; Hunkapiller, T.J.; Hood, L.F. (1985) Nucleic Acids Res. 13, 2439-2502. 9. Sproat, B.S.; Beijer, B.; Rider, P.; Neumer, P. (1987) Nucleic Acids Res. 15, 4837-4848. 10. Zuckerman, R.; Corey D.; Schultz, P. (1987) Nucleic Acids Res. 15, 5305-5321. 11. Connolly, B.A. (1985) Nucleic Acids Res. 13, 4485-4502. 12. Jablonski, E.; Moomaw, E.W.; Tullis, R.H.; Ruth, J.L. (1986) Nucleic Acids Res. 14, 6115-6128. 13. Urdea, M.S.; Warner, B.D.; Running, J.A.; Stempien M., Clyne, J.; Horn, T. (1988) Nucleic Acids Res. 16, 4937-4956. 14. Haralambidis, J.; Chai, M.; Tregear, G.W. (1987) Nucleic Acids Res. 15, 4857-4876. 15. Ruth, J.L.; Morgan, C., Pasko, A., (1985) DNA 4, 93. 16. Ruth, J.L. (1984) DNA, 3, 123. 17. Draper, D.E. (1984) Nucleic Acids Res. 12, 989-1002. 18. Haralambidis, J.; Duncan, L.; Tregear, G.W. (1987) Tetrahedron Lett., 28, 5199-5202. 19. Haralambidis, J.; Duncan, L.; Angus, K.; Chai, M.: Pownall, S.; Tregear, G. (1988) Nucleic Acids Res. Symposium Series, No.20, 115-116. 20. Atherton, E.; Sheppard, R.C. (1985) J. Chem. Soc., Chem. Commun., 165-166. 21. Beaucage, S.L.; Caruthers, M.H. (1981) Tebtrhedron Lett. 22, 1859-1862. 22. McBride, L.J.; Caruthers, M.H. (1983) Tetrahedron Lett. 24, 245-248. 23. Drinkwater, C.C.; Evans, B.A., Richards, R.I. (1988) Trends in Biochemical Sciences, 13, 169-172. 24. Atkinson, T.; Smith, M. In: Oligonucleotide Synthesis: A Practical Approach; Gait, M.J., Ed., IRL Press, Oxford, (1984) pp 35-81. 25. Bernardi, A.; Benardi, G. In: The Enzymes, 3rd edition, Volume 4; Boyer, P.P. Ed., Academic Press, New York, (1971) pp 359-336. 26. Castro, B.; Dormoy, J.R.; Evin, G.; Selve, C. (1975) Tetrahedron Lett. 1219-1222. 27. Hudson, D. (1988) J. Org. Chem. 53, 617-624. 28. Canrthers, M.H.; McBride, L.J.; Bracco, L.P.; Dubendorff, J.W. (1985) Nucleosides and Nucleotides, 4, 95-105. 29. Eadie, J.S.; Davidson, D.S. (1987) Nucleic Acids Res. 15, 8333-8349. 30. Still, W.C.; Kahn, M.; Mitra, A. (1978) J. Org. Chem. 43, 2923-2925. 31. Sarin, V.; Kent, S.; Tam, J.; Merrifield, R. (1981) Anal. Biochem. 117, 147-157. 32. Matteucci, M.D.; Caruthers, M.H. (1981) J. Am. Chem. Soc., 103, 3185-3191.