ω-Azido fatty acids as probes to detect fatty acid biosynthesis, degradation, and modification.

FAs play a central role in the metabolism of almost all known cellular life forms. Although GC-MS is regarded as a standard method for FA analysis, other methods, such as HPLC/MS, are nowadays widespread but are rarely applied to FA analysis. Here we present azido-FAs as probes that can be used to study FA biosynthesis (elongation, desaturation) or degradation (β-oxidation) upon their uptake, activation, and metabolic conversion. These azido-FAs are readily accessible by chemical synthesis and their metabolic products can be easily detected after click-chemistry based derivatization with high sensitivity by HPLC/MS, contributing a powerful tool to FA analysis, and hence, lipid analysis in general.

FAs are found in all known living organisms, playing a vital role in cell compartmentalization, energy storage, and secondary metabolite production. In bacteria, most FAs are found in the cell membrane as part of the lipid bilayer ( 1 ). The actual FA profi le of a cell can strongly vary depending on environmental and developmental conditions requiring de novo biosynthesis, degradation, and modifi cation of the FAs involved (2)(3)(4)(5). The monitoring of these metabolic pathways is usually conducted by GC-MS of FA methyl esters or other volatile FA derivatives ( 6 ). In spite of its merits, such as high sensibility and direct observability of especially volatile natural products, it is desirable to also have simple and effective methods of FA analysis via HPLC/MS, which today is a widespread tool in analytical chemistry as well. However, HPLC/MS-based FA analysis is diffi cult, mainly because of little ionizability and hence little signal strength. Small modifi cations introduced to certain FAs, however, can greatly increase ionizability water gradient ranging from 5 to 95% in 22 min at a fl ow rate of 0.6 ml/min ( 12 ). High resolution mass spectra were obtained from a MALDI LTQ Orbitrap XL (Thermo Fisher Scientifi c, Inc., Waltham, MA) equipped with a laser at 337 nm. A 4-chloro-␣cyanocinnamic acid matrix was used, and the sum formulas and according masses were internally calibrated using fl uorescein (monoisotopic mass = 332.068473) as a standard ( 13 ).

Bacterial strains and culture conditions
All Escherichia coli strains used in this study were grown on solid and liquid Luria-Bertani (LB) (pH 7.0) medium at 30°C and 180 rpm on a rotary shaker. The mutants and controls obtained from the Coli Genetic Stock Center were grown in the presence of kanamycin (40 g/ml). Psychrobacter urativorans (ATCC 15174) was obtained from the German Collection of Microorganisms and Cell Cultures (DSMZ) and grown on solid and liquid Oxoid nutrient broth (DSMZ medium 948) at 16°C on a rotary shaker at 210 rpm.

Feeding experiments for observation of FA degradation and desaturation
The feeding experiments for the observation of the degradation of FAs were conducted with 16-azidohexadecanoic acid. The compound was fed at a fi nal concentration of 1 mmol/l to a culture of fadE , fadA , and ilvE , respectively. Cultures were grown in 250 ml Erlenmeyer fl asks containing 30 ml of LB medium. Inoculation was performed from a preculture that was incubated overnight. The optical density was set to 0.1 and the culture was incubated at 30°C and 180 rpm for 3 h before feeding ensued and samples of 1 ml were taken . The samples were frozen in liquid nitrogen and freezedried before adding 1 ml 1 M NaOH and heated to 80°C for 4 h. Thereafter, the now clear solution was acidifi ed with 190 l of 6 M HCl and 1 ml of hexane was added. After vigorous stirring, the samples were centrifuged for 5 min at 4,000 rpm (3,220 g ) and 700 l of the supernatant were removed, dried, and taken up in 500 l of acetonitrile, after which 40 l of a 10 mM solution of the prepurifi ed raw product of 1 was added. The reaction was allowed to occur at room temperature for 24 h before HPLC/MS analysis. The experiments with P. urativorans were conducted in the same fashion under different incubation conditions as described above.

Feeding experiments for FA elongation observation
The feeding experiments for the observation of the elongation of FAs were either conducted with S -(5-azidopentanoyl)-Nacetylcysteamine or S -(6-azidohexanoyl)-N -acetylcysteamine, respectively. The compounds were fed at a fi nal concentration of 2 mmol/l to a culture of fadE , fadA , and DH10B. Cultures were grown in 100 ml Erlenmeyer fl asks containing 10 ml of LB medium, including kanamycin in the concentration mentioned above in the case of fadE and fadA . Inoculation was performed from a preculture that was incubated overnight. The optical density was set to 0.1 and the culture was incubated at 30°C and 180 rpm for 2 h before feeding ensued and samples of 700 l were taken . To these samples 200 l 4 M NaOH were added, after which the mixture was heated to 90°C for 1 h . Thereafter, the now clear solution was acidifi ed with 200 l of 6 M HCl and 800 l of hexane were added. After vigorous stirring, the samples were centrifuged for 5 min at 13,300 rpm (17,000 g ) and 600 l of the supernatant were removed, dried, and taken up in 100 l of acetonitrile, after which 10 l of a 0.1 M solution of the prepurifi ed raw product of 1 was added. The reaction was allowed to occur at 50°C for 1 h before HPLC/MS analysis.

RESULTS AND DISCUSSION
Prior to the application of AFAs as molecular probes for FA metabolism, a library of eight AFAs of varying chain such as dibenzocyclooctyne-amine, with comparable structures can be obtained commercially from Sigma-Aldrich and mostly share the abilities of the cyclooctyne chosen in this work. Silica chromatographic purifi cation was performed on a Biotage SP1™ Flash purifi cation system (Biotage, Uppsala, Sweden), using 40+M, 25+M, or 12+M KP-Sil cartridges (Biotage, Uppsala, Sweden) in combination with an UV detector. 1  C-HMBC NMR spectra for the synthesis products were recorded on a Bruker AV500 (500 MHz), AV400 (400 MHz), or AM250 (250 MHz) spectrometer using CDCl 3 or CD 3 CN as solvent and internal standard, using the chemical shifts described by Gottlieb et al. ( 11 ) with the exception of the 13 Cshift of CDCl 3 , which was set to 77.00 ppm. 1 H-NMR: CHCl 3 , ␦ = 7.24 ppm; CH 3 CN, ␦ = 1.96 ppm; 13 C-NMR: CDCl 3 , ␦ = 77.00 ppm; CD 3 CN, ␦ = 118.26 ppm. ESI HPLC/MS analysis was performed with a DionexUltiMate 3000 system coupled to a Bruker AmaZon X mass spectrometer and an Acquity UPLC BEH C18 1.7 m RP column (Waters) using a MeCN/0.1% formic acid in were also detectable when high concentrations of the corresponding nonoxidized AFAs were reached (supplementary Fig. III). The retention times of these metabolic intermediary products proved to be similar to their saturated counterparts for long chain AFAs, while being slightly lower for short chained AFAs and clearly different from the retention times of C n+1 -AFAs, which have the same nominal mass as the corresponding C n-3 -keto-AFAs.
Next, the degradation of 2h was analyzed over time in different E. coli strains with mutations in fadA (JW5578) and fadE (JW5020), with ilvE (JW5606) as a control. The fadA mutant supposedly lacked the ability to conduct the fi nal thiolysis step, whereas the fadE mutant lacked the acyl-CoA dehydrogenase. In the control strain (with a defect in leucine degradation not infl uencing FA metabolism), signals correlated to C 14 -and C 12 -AFA ( 2l and 2f , respectively) showed a maximum at 2.3 and 2.8 h after the addition of 2h , respectively, and also shorter degradation products could be observed ( Fig. 2B ). Notably, analysis of the fadA mutant showed no difference to the ilvE control strain (not shown), indicating the functional complementation of the fadA mutation, while the fadE mutant showed length was synthesized ( 2a-h , Fig. 1A ) from the corresponding -bromo-FAs in a one-step reaction (supplementary Scheme I). TDAC [obtained by a simple three step synthesis (supplementary Scheme II) with good click properties and formidable solubility in a wide range of solvents, as it is common to heterocyclic methoxylated cyclooctynes ( 14,15 )] was then reacted with these AFAs and subsequently analyzed by HPLC/MS ( Fig. 1 ). Because 15-bromopentadecanoic acid was not commercially avail able, it was synthesized from pentadecanolide by hydrolysis and subsequent bromination of the resulting 15-hydroxypentadecanoic acid for direct use in the corresponding AFA synthesis (supplementary Scheme III). The results showed tremendous signal amplifi cation by a factor of about 5,000 in the extracted ion chromatogram (EIC) of the clicked product as the [M+H] + ion in comparison to the free AFA (supplementary Fig. IA), along with clear differences in retention time according to the chain length of the corresponding AFA ( Fig. 1B ). Due to this strong increase in sensitivity resulting from the formation of the triazole unit, a relative quantifi cation between different TDAC-modifi ed AFAs also seems to be reasonable . Each signal consists of a characteristic double-peak, representing the two regioisomers formed in different amounts in accordance with the fi ndings of Starke, Walther, and Pietzsch ( 9 ) for 3b / 4b . This regioselectivity corresponds to the different steric demand of each of the transition states ( 9 ). It was observed that the more abundant isomers, 3a -h , had slightly higher retention times than the less abundant isomers, 4a -h . Analysis of the MS 2 fragmentation pattern revealed distinct neutral losses for each of the two isomers. In particular, the minor isomer always features a dominant loss of water, followed by loss of the carboxylic group in the form of a dihydroxycarbene, whereas the main isomer predominantly shows only the loss of the latter group (supplementary Fig. IB). Thus, by comparing the characteristic fragmentation patterns, retention times, and peak shapes, simple identifi cation of FA metabolites becomes possible.
In order to test whether AFAs can be used for the analysis of FA metabolism, ␤ -oxidation was investigated initially. ␤ -Oxidation is the primary FA degradation pathway in most organisms, consisting of desaturation, hydratization, oxidation, and fi nally thiolysis of the resulting CoA-bound 3-keto-FA residue, resulting in the formation of acetyl-CoA and an acyl-CoA residue shortened by two carbon atoms (supplementary Fig. II) ( 16,17 ). In most bacteria, the basic steps are performed by only three proteins: FadE (desaturation), FadB (hydratization and oxidation), and FadA (thiolysis) ( 18 ). Thus, a feeding experiment was conducted in which a long chain AFA was fed to E. coli DH10B. Hydrolysis of this culture and derivatization with TDAC allowed the combined detection of free and (previously) bound FA ␤ -oxidation degradation products, clearly showing that -azido FAs are tolerated by the ␤ -oxidation machinery. Indeed, C 16 -AFA ( 2h ) was readily degraded in C 2 -steps, leading to a degradation product as short as 4-azidobutanoic acid ( 2a ) resulting in the detection of 3a / 4a ( Fig. 2A ). Additionally, the 3-keto and the 3-hydroxy forms Fig. 2. A: EICs of click-labeled degradation products of 2h 12 h after initial feeding at 1 mM. The same amount of 2h was fed again right before the sample was taken. 3h / 4h = TDAC-C 16 -AFA, 3l / 4l = TDAC-C 14 -AFA, 3f / 4f = TDAC-C 12 -AFA, 3d / 4d = TDAC-C 10 -AFA, 3i / 4i = TDAC-C 8 -AFA, 3c / 4c = TDAC-C 6 -AFA, 3a / 4a = TDAC-C 4 -AFA. B: Relative peak area (in percent relative to the amount of 3h + 4h ) of 2h degradation products when fed to ilvE mutant. Increasing amounts of C 14 -and C 12 -AFA can be seen 1 h after feeding, followed by further degradation products. addition of 5a ( Fig. 3B ). No elongation could be observed when C 6 -azidoacyl-NAC ( 5b ) was fed (not shown), which implies that 5a more closely resembles the natural FA and is well-tolerated by the FA biosynthesis machinery.
Besides FA degradation and biosynthesis, FA modifi cation can also be studied using the described approach: P. urativorans , formerly known as Micrococcus cryophilus , is a Gram-negative psychrophilic bacterium found in cold habitats that possesses a simple membrane FA composition, consisting mostly of ⌬ 9-unsaturated palmitoleate and oleate residues ( 20,21 ). When adapting to temperature changes in the range of 0-20°C, P. urativorans changes its FA profi le by varying the ratio of the two main FA residues in the membrane ( 22,23 ). Feeding C 16 -AFA ( 2g ) to a culture of P. urativorans at 16°C followed by the subsequent detection of a mono-unsaturated FA with slightly reduced retention time ( 3m / 4m , Fig. 4 ) confi rms that a desaturase is indeed involved in this process .

SUMMARY
The use of AFAs in combination with modern and easy to synthesize cyclooctynes, such as TDAC, has shown to be a viable method for the analysis of FA metabolism, ranging from biosynthesis to degradation and desaturation. Because no degradation products smaller than 4-azidobutanoic acid ( 2a ) have been detected, it can be assumed that 2a cannot be degraded. This might lead to increased availability of labeled FAs, even after prolonged metabolization, as compared with completely degradable isotope labels ( 24 ). The azido label forms an easily detectable triazole ring after reacting with a cyclooctyne, and is readily introduced into the desired target compound, thus allowing a wide range of applications in a time-resolved manner at low cost and high sensitivity. Our method can also easily be applied to study FA uptake or FA activation, which can be coupled to FA ␤ -oxidation and can be detected by our approach. Moreover, it can not only be applied to FAs, but it can also be applied to lipid metabolism in general . For the latter, simply leaving out the hydrolytic step during sample preparation would enable the detection of acylcarnitines or CoA derivatives, as well as other lipid species carrying the respective AFA, not only in bacteria but also in all no degradation activity at all (supplementary Fig. IV) ( 19 ). The decrease in concentration of 2h in the latter experiment might be attributed to absorption on the glass wall of the cultivation fl asks or micelle formation in the medium. Similarly, C 15 -AFA ( 2g ) was also fed to the E. coli wild-type, and degradation products of uneven carbon chain length could be observed as expected (supplementary Fig. V). This also proved that AFA incorporation and degradation is not dependent on a specifi c chain length, but occurs with several long chain AFAs, furthermore indicating the broad applicability of AFAs for degradation studies.
Next, FA biosynthesis was studied, generally requiring the condensation of a given FA thioester with malonyl-ACP leading to an elongation in C 2 -steps in a fashion reverse to FA degradation (supplementary Fig. II). Therefore, acyl-ACP-mimicking S -azidoacyl-N -acetylcysteamines (azidoacyl-NACs) of five or six carbons in acyl chain length were synthesized ( 5a and 5b , respectively; see supporting information ) and fed to the E. coli fadE mutant and to the wild-type (DH10B) as control. The wild-type showed only trace amounts of 3g and 3k ( Fig. 3A ), whereas elongated AFA derivatives up to C 15 -AFA ( 3g ) could be observed in the fadE mutant during the fi rst 12 h after the Fig. 3. A: EICs of clicked uneven AFAs immediately after and 24 h after feeding of 5a (detected as 3b / 4a ) to fadE mutant. The other EICs prove the production of C 9 -AFA ( 3j ), C 11 -AFA ( 3e ), C 13 -AFA ( 3k / 4k ), and C 15 -AFA ( 3g / 4g ). Even after 24 h the wild-type (DH10B) only shows minimal production of C 13 -AFA ( 3k ) and C 15 -AFA ( 3g ). B: Relative peak areas (in percent relative to the area of 3b + 4b ) of various AFAs from the feeding experiment with the fadE mutant. C 15 -AFA (as click product 3g ) can clearly be seen as emerging main product of FA elongation. Production of C 7 -AFA could not be detected due to signal overlap with an impurity in the TDAC reagent. Fig. 4. Desaturase activity shown in P. urativorans . The EIC of the clicked AFAs shows that C 16 -AFA (shown as click products 3h and 4h ) is converted to the monodesaturated C 16:1 -AFA (shown as click products 3m and 4m , with the regiochemistry assumed to be the same as with 3a-l and 4a-l , respectively).