Simultaneous quantification of hemagglutinin and neuraminidase of influenza virus using isotope dilution mass spectrometry

Vaccine 30 (2012) 2475–2482

Contents lists available at SciVerse ScienceDirect

Vaccine journal homepage: www.elsevier.com/locate/vaccine

Simultaneous quantification of hemagglutinin and neuraminidase of influenza virus using isotope dilution mass spectrometry Tracie L. Williams, James L. Pirkle, John R. Barr ∗ Center for Environmental Health, Centers for Disease Control and Prevention, 4770 Buford Highway, Atlanta, GA 30341, United States

a r t i c l e

i n f o

Article history: Received 29 August 2011 Received in revised form 5 December 2011 Accepted 11 December 2011 Available online 22 December 2011 Keywords: Mass spectrometry Influenza Vaccine Hemagglutinin Neuraminidase Proteins

a b s t r a c t Influenza vaccination is the primary method for preventing influenza and its severe complications. Licensed inactivated vaccines for seasonal or pandemic influenza are formulated to contain a preset amount of hemagglutinin (HA), the critical antigen to elicit protection. There is currently no regulatory method that quantifies neuraminidase (NA), the other major membrane-bound protein thought to have protective capability. This is primarily due to the limitations both in sensitivity and in selectivity of current means to quantify these antigens. Current methods to establish the HA concentration of vaccines rely on indirect measurements that are subject to considerable experimental variability. We present a liquid chromatography-tandem mass spectrometry (LC/MS/MS) method for the absolute quantification of viral proteins in a complex mixture. Through use of an isotope dilution approach, HA and NA from viral subtypes H1N1, H3N2, and B were determined both directly and rapidly. Three peptides of each subtype were used in the analysis of HA to ensure complete digestion of the protein and accuracy of the measurement. This method has been applied to purified virus preparations, to monovalent bulk concentrates, to trivalent inactivated influenza vaccines, and even crude allantoic fluid with improved speed, sensitivity, precision, and accuracy. Detection of 1 ␮g/mL of protein is easily obtained using this method. The sensitivity of the method covers the range expected in vaccine preparations, including adjuvant-based vaccine. This LC/MS/MS approach substantially increases the selectivity, accuracy and precision used to quantify the amount of viral proteins in seasonal and pandemic influenza vaccines and reduce the time and effort to deliver influenza vaccines for public health use during the next influenza pandemic. Published by Elsevier Ltd.

1. Introduction Vaccination against influenza virus is the primary strategy to reduce the morbidity and mortality associated with the seasonal influenza. These vaccines are protective by inducing the production of antibodies directed against the hemagglutinin (HA) and the neuraminidase (NA) surface glycoproteins which protect in very different ways. Antibody against HA neutralizes viral infectivity either by interfering with attachment of the virus to the host cell surface or with interruption of the fusion process of the viral HA with endosomal membranes [1–3]. The neuraminidase of influenza is a surface glycoprotein that removes sialic acid, the viral receptor, from both viral and host proteins. The removal of sialic acid from viral proteins plays a key role in the release of the virus from the cell and prevents aggregation of the virus that would occur if hemagglutinin was able to bind to other viral proteins. Antibody against NA, therefore, allows the cell to be infected with the virus, but inhibits the release of progeny viruses from infected cells [4].

∗ Corresponding author. Tel.: +1 7704887848. E-mail address: [email protected] (J.R. Barr). 0264-410X/$ – see front matter. Published by Elsevier Ltd. doi:10.1016/j.vaccine.2011.12.056

While neutralizing antibodies to the hemagglutinin protein can directly block virus entry, as the protein will not be able to bind to its cell receptors, protective antibodies to the neuraminidase protein are thought to primarily aggregate virus on the cell surface effectively reducing the amount of virus released from infective cells. There have been reports that an intravirionic antigenic competition occurs between HA and NA and that the immune response is preferentially developed for HA, the most abundant surface antigen [5]. However, it is believed that both surface antigens can provide protection against influenza [6–8]. Seasonal influenza vaccines are currently required to contain a specified amount of HA, the trimeric glycoprotein most considered to provide protection against circulating influenza strains. The United States has no requirement that NA is present in commercial vaccines, while European nations require that evidence of the presence of NA be offered, but no specified amount is mandated. It is not clear that the current amount of NA in seasonal influenza vaccines is enough to elicit a strong protective antibody response, especially if there is antigenic competition with HA. Influenza viral strains undergo rapid mutations and may experience an abrupt change in the HA and NA which could result in a virus that is radically different from those previously

2476

T.L. Williams et al. / Vaccine 30 (2012) 2475–2482

Table 1 Target peptides employed for the quantification of neuraminidase. Underlined amino acids were those that were C13 - and N15 -labeled. Target peptide

Subtype

Observed m/z

ms/ms ion (quantification)

ms/ms ion (confirmation)

ms/ms ion (confirmation)

LNVETDTAEIR LNVETDTAEIR SGYSGIFSVEGK SGYSGIFSVEGK YNGIITDTIK YNGIITDTIK YNGIITETIK YNGIITETIK YNGIITGTIK YNGIITGTIK

B B H3N2 H3N2 H5N1, H1N1 (pandemic) H5N1, H1N1 (pandemic) H1N1 (seasonal) H1N1 (seasonal) H1N1 (seasonal) H1N1 (seasonal)

630.8 (+2) 634.3 (+2) 615.8 (+2) 618.8 (+2) 569.3 (+2) 572.8 (+2) 576.3 (+2) 579.8 (+2) 540.3 (+2) 543.8 (+2)

934.4 (y8) 941.5 (y8) 779.4 (y7) 785.4 (y8) 690.4 (y6) 697.4 (y6) 704.4 (y6) 711.4 (y6) 745.5 (y7) 752.5 (y7)

1033.5 (y9) – 836.5 (y8) – 803.5 (y7) – 817.5 (y7) – 802.5 (y8) –

1147.6 (y10) – 923.5 (y9) – 860.5 (y8) – 874.5 (y8) – 916.5 (y9) –

circulating in human populations [9]. Small changes in the amino acid sequence which result in the inability of antibodies to neutralize the virus is called an “antigenic drift”. “Antigenic shift”, on the other hand, is the formation of a new virus subtype that incorporates HA and NA from different subtypes. The mutations and their effects on the virus and its subsequent transmissibility to or between humans cannot be adequately foreseen. For this reason, the strains included in seasonal influenza vaccines must be annually updated. One of the major bottlenecks in the production cycle of vaccines is quantification of these antigenic proteins which have undergone an abrupt change (either a “shift” or a “drift”) which renders the reagents used for quantification unusable. Currently, the method used to quantify HA, is the single radial immuno-diffusion (SRID) assay which requires strain-specific reagents including an HA reagent and anti-HA sheep serum [10,11]. Production and characterization of these reagents can take several months. However, until the reagents are available, manufacturers are unable to formulate, fill, or deliver the final vaccine product. To add the requirement of quantification of NA using SRID would likely increase the vaccine production time schedule. In addition, SRID is not a sufficiently sensitive method as it cannot routinely quantify a sample concentration that is less than 4 ␮g/mL. HA is the most abundant protein on the surface of the virus with NA being at least 3 or more times fewer in number. Therefore, SRID would not be a suitable method for NA quantification in final presentations of most seasonal vaccines. We have reported a method that combines isotope dilution mass spectrometry (IDMS) and a multiplexed multiple reaction monitoring (MRM) approach to simultaneously quantify several HA subtypes in one sample [12]. The quantification method has been expanded to include the independent measurement of three peptides of each HA subtype to ensure that digestion of the protein is complete [13] since the IDMS method involves selecting a specific target peptide as a stoichiometric representative of the protein from which it is cleaved. The IDMS method has been further broadened to include the quantification of NA. With this method, both HA and NA of the three subtypes found in seasonal vaccines can be quantified in one experiment. It is rapid, accurate, and easily adaptable to changes in annual vaccine strains.

2. Methods and materials 2.1. Vaccines and virus strains The commercial 2007/2008 vaccine contains A/Solomon Islands/3/2006 (H1N1), A/Wisconsin/67/2005 (H3N2), and B/ Malaysia/2506/2004-like strains. The commercial 2009/2010 vaccine contains an A/Brisbane/59/2007 (H1N1), A/Uruguay/ 716/2007, and B/Brisbane/60/2008-like strains. The commercial 2010 vaccine contains A/California/7/2009, A/Perth/16/2009, and B/Brisbane/60/2008-like strains. The vaccines were from different

Collision energy (eV) 25 25 24 24 23 23 23 23 21 21

manufacturers, contained various surfactants and preservatives, and all were used without further purification. Whole virus preparations in crude allantoic fluid and MDCK cells were inactivated either by using ␤-propiolactone [14] or by performing the tryptic digestion protocol prior to providing the sample. These samples were provided by the National Center for Immunization and Respiratory Diseases and used without further purification. 2.2. Synthesis of native and labeled synthetic peptides Custom synthetic peptides were synthesized at a 1–5 mg scale by MidWest Bio-Tech (Fishers, IN) and are described in Tables 1 and 2. These lyophilized peptides were reconstituted, aliquoted in-house in 200 ␮L volumes into 1.5 mL vials using a Biomex NXP Laboratory Automation Workstation (Beckman Coulter, Fullerton, CA) and then lyophilized again and stored at −70 ◦ C until use. Each vial was targeted to contain 3–6 nmols of peptide. Accuracy of the peptide content in the vial was determined by an isobaric-tagged isotope dilution amino acid analysis (AAA) method developed in our laboratory [15]. This AAA method is calibrated using the certified standard mix of 17 amino acids provided by the National Institute for Standards and Technology (NIST). Labeled analogs of the target peptides VNSVIEK, STQAAIDQINGK, STQAAINQINGK, YNGIITETIK, YNGIITDTIK, YNGIITGTIK, and LNVETDTAEIR were made by incorporating the isoleucine closest to the carboxy-terminal with 13 C and 15 N to give a peptide that is 7 Da heavier than the native peptide. Leucine was 13 C and 15 N labled to give a peptide that is 7 Da heavier than the native peptide for DEALNNR, and SHFANLK. For TLDFHDSNVK, TLDYHDSNVK, NLNSLSELEVK, and SGYSGIFSVEGK, valine was 13 C and 15 N labeled resulting in a peptide that is 6 Da heavier than the native peptide. Proline and phenylalanine were similarly labeled to give peptides that are respectively 8 and 10 Da heavier than the native peptides for the peptides GVLLPQK, GILLPQK, and EQLSSVSSFER respectively. The HA peptides, and their location in the HA protein, are shown in Fig. 1. 2.3. Preparation of working stock, calibration, and labeled peptide solutions All labeled and unlabeled peptides were reconstituted as follows. One hundred ␮L of 10% (v/v) aqueous formic acid solution was added to the vial containing peptide. The vial was mixed on a vortex mixer to fully dissolve the peptide. An additional volume of 0.1% formic acid was added to make a 5 pmol/␮L stock solution. The volumes depended upon the peptide content in the vial as determined by AAA. In order to facilitate the sample preparation of unknowns, a cocktail of the 5 labeled NA peptides was made by adding 100 ␮L of each peptide and lyophilizing it to dryness. The same was done with the 5 NA native peptides and the 12 HA native and labeled peptides. Therefore, instead of having 34 vials in which to generate a calibration curve, there are only 4 (a vial

T.L. Williams et al. / Vaccine 30 (2012) 2475–2482

2477

Table 2 Target peptides employed for the quantification of hemagglutinin. Underlined amino acids were those that were C13 - and N15 -labeled. Target peptide

Subtype

Observed m/z

ms/ms ion (quantification)

EQLSSVSSFER EQLSSVSSFER TLDFHDSNVK TLDFHDSNVK TLDYHDSNVK TLDYHDSNVK VNSVIEK VNSVIEK STQAAIDQINGK STQAAIDQINGK STQAAINQINGK STQAAINQINGK DEALNNR DEALNNR EFSEVEGR EFSEVEGR NLNSLSELEVK NLNSLSELEVK SHFANLK SHFANLK GVLLPQK GVLLPQK GILLPQK GILLPQK

H1N1 H1N1 H1N1 (seasonal) H1N1 (seasonal) H1N1 (pandemic) H1N1 (pandemic) H1N1 H1N1 H3N2 H3N2 H3N2 H3N2 H3N2 H3N2 H3N2 H3N2 B B B B B (Yamagata lineage) B (Yamagata lineage) B (Victoria lineage) B (Victoria lineage)

634.8 (+2) 639.8 (+2) 588.3 (+2) 591.3 (+2) 596.3 (+2) 599.3 (+2) 394.7 (+2) 397.7 (+2) 623.3 (+2) 626.8 (+2) 622.8 (+2) 626.3 (+2) 416.2 (+2) 419.7 (+2) 476.7 (+2) 479.7 (+2) 623.3 (+2) 626.8 (+2) 408.7 (+2) 412.2 (+2) 377.7 (+2) 380.8 (+2) 384.8 (+2) 387.8 (+2)

811.4 (y7) 821.4 (y7) 961.4 (y8) 967.5 (y8) 977.4 (y8) 983.4 (y8) 575.3 (y5) 581.4 (y5) 929.5 (y9) 936.5 (y9) 928.5 (y9) 935.5 (y9) 587.3 (y5) 594.3 (y5) 676.3 (y6) 682.3 (y6) 1018.5 (y9) 1025.6 (y9) 592.3 (y5) 599.4 (y5) 697.5 (y6) 703.5 (y6) 711.5 (y6) 604.4 (y5)

of NA native peptides, a vial of NA labeled peptides, a vial of HA native peptides, and a vial of HA labeled peptides). By reconstituting these vials in 100 ␮L of 0.1% fomic acid, a 5 pmol/␮L solution of the native peptides used to make calibration curves could be easily prepared. Adding 1 mL of 0.1% formic acid to the labeled vials provided a 0.5 pmol/␮L solution of the labeled peptides which could be

ms/ms ion (confirmation) 898.4 (y8) – 846.4 (y7) – 862.4 (y7) – 689.4 (y6) – 858.5 (y8) – 857.5 (y8) – 716.3 (y6) – 589.3 (y5) – 904.5 (y8) – 729.4 (y6) – 598.4 (y5) – 598.4 (y5) –

ms/ms ion (confirmation) 1011.5 (y9) – 699.3 (y6) – 699.3 (y6) – 488.3 (y4) – 787.4 (y7) – 786.4 (y7) – 516.3 (y4) – 823.4 (y7) – 817.5 (y7) – 445.3 (y4) – 485.3 (y4) – 485.3 (y4) –

Collision energy (eV) 25 25 23 23 23 23 17 17 25 25 25 25 17 17 23 23 25 25 17 17 16 16 16 16

used as internal standards to spike unknown samples. Seven 0.5 mL stock calibration standards incorporating both HA and NA, ranging from 10 to 250 fmol/␮L, were prepared by carefully pipetting 1, 3, 5, 7, 9, 18 and 25 ␮L of each of the HA and NA 5 pmol/␮L unlabeled peptide solutions (5 pmol/␮L concentration), 50 ␮L of each of the two HA and NA labeled solutions (at 0.5 pmol/␮L concentration),

B/Brisbane/60/2008 MKAIIVLLMV IPSARVSILH NKTATNPLTI GKTGTITYQR ERGFFGAIAG ADTISSQIEL DDGLDNHTIL

VTSNADRICT EVRPVTSGCF EVPYICTEGE GILLPQKVWC FLEGGWEGMI AVLLSNEGII LYYSTAASS

GITSSNSPHV PIMHDRTKIR DQITVWGFHS ASGRSKVIKG AGWHGYTSHG NSEDEHLLAL

VKTATQGEVN QLPNLLRGYE DNETQMAKLY SLPLIGEADC AHGVAVAADL ERKLKKMLGP

VTGVIPLTTT HIRLSTHNVI GDSKPQKFTS LHEKYGGLNK KSTQEAINKI SAVEIGNGCF

PTKSHFANLK NAENAPGGPY SANGVTTHYV SKPYYTGEHA TKNLNSLSEL ETKHKCNQTC

GTETRGKLCP KIGTSGSCPN SQIGGFPNQT KAIGNCPIWV EVKNLQRLSG LDRIAAGTFD

KCLNCTDLDV ITNGNGFFAT EDGGLPQSGR KTPLKLANGT AMDELHNEIL AGEFSLPTFD

ALGRPKCTGK MAWAVPKNDK IVVDYMVQKS KYRPPAKLLK ELDEKVDDLR SLNITAASLN

LKGIAPLQLG SCSHNGESSF GRINYYWTLL VTGLRNIPSI ENLNKKVDDG LNREKIDGVK

NCSVAGWILG NPECELLISK YRNLLWLTGK NGLYPNLSKS EPGDTIIFEA NGNLIAPRYA QSRGLFGAIA GFIEGGWTGM FIDqIWTYNAE LLVLLENERT LESMGVYQIL AIYSTVASSL

SSSTGEICDS TSSSCIRRSN NIPSRISIYW KLATGMRNVP RIQDLEKYVE ALNNRFQIKG

PHQILDGENC NSFFSRLNWL TIVKPGDILL EKQTRGIFGA DTKIDLWSYN VELKSGYKDW

A/Brisbane/59/2007 (H1N1) MKVKLLVLLC ESWSYIVEKP YANNKEKEVL FALSRGFGSG VDGWYGYHHQ LDFHDSNVKN VLLVSLGAIS

TFTATYADTI NPENGTCYPG VLWGVHHPPN IINSNAPMDK NEQGSGYAAD LYEKVKSQLK FWMCSNGSLQ

CIGYHANNST HFADYEELRE IGDQKALYHT CDAKCQTPQG QKSTQNAING NNAKEIGNGC CRICI

DTVDTVLEKN QLSSVSSFER ENAYVSVVSS AINSSLPFQN ITNKVNSVIE FEFYHKCNDE

VTVTHSVNLL FEIFPKESSW HYSRKFTPEI VHPVTIGECP KMNTQFTAVG CMESVKNGTY

ENSHNGKLCL PNHTVTGVSA AKRPKVRDQE KYVRSAKLRM KEFNKLERRM DYPKYSEESK

A/Uruguay/716/2007 (H3N2) MKTIIALSYI QCDGFQNKKW NVTMPNNEKF RGYFKIRSGK GMVDGWYGFR HTIDLTDSEM FLLCVALLGF

LCLVFAQKLP DLFVERSKAY DKLYIWGVHH SSIMRSDAPI HQNSEGIGQA NKLFEKTKKQ IMWACQKGNI

GNDNSTATLC SNCYPYDVPD PGTDNDQIFP GKCNSECITP ADLKSTQAAI LRENAEDMGN RCNICI

LGHHAVPNGT YASLRSLVAS YAQASGRITV NGSIPNDKPF DQINGKLNRL GCFKIYHKCD

IVKTITNDQI SGTLEFNNES STKRSQQTVI QNVNRITYGA IGKTNEKFHQ NACIGSIRNG

EVTNATELVQ FNWTGVTQNG PNIGSRPRVR CPRYVKQNTL IEKEFSEVEG TYDHDVYRDE

TLIDALLGDP THLKFKYPAL INSTGNLIAP IAGFIENGWE AELLVALENQ ILWISFAISC

Fig. 1. The amino acid sequences of the HA protein of the strains included in the 2009/2010 commercial vaccines are shown with target peptides highlighted in red. Three target peptides are used to quantify the HA protein to ensure complete digestion of the protein and to ensure accuracy of the measurement. The arginine which is in bold font is where cleavage in to HA1 and HA2 subunits occurs. Ideally, a target peptide will be located on each subunit to ensure complete digestion across the protein and to ensure the presence of both subunits in a vaccine.

2478

T.L. Williams et al. / Vaccine 30 (2012) 2475–2482

and supplemented with enough 0.1% formic acid to make the final volume 0.5 mL. Therefore, the 7 calibration standards were 10, 30, 50, 70, 90, 180, and 250 fmol/␮L for each of the 17 peptides. 2.4. Preparation of vaccine digests and quantification by LC/MS/MS Three trivalent seasonal vaccines were used in this study. A 10 ␮L aliquot of 0.2% Rapigest (Waters Corporation, Milford, MA) was added to 40 ␮L of vaccine. The samples were heated for 5 min at 100 ◦ C. After cooling to room temperature, 5 ␮L (∼86 pmol) of sequencing grade modified trypsin (Promega, Madison, WI) were added to each sample and incubated at 37 ◦ C for 2 h. Digests were allowed to cool, and 10 ␮L of 0.475 M HCl was added to reduce the pH to 2.0 to cleave the acid labile surfactant [16]. To the resulting milieu, 10 ␮L each of the 0.5 pmol/␮L labeled cocktail solution for HA and for NA was added. A 15 ␮L aliquot of 0.1% formic acid was also added to make the final volume 100 ␮L. The digested samples were mixed, centrifuged for 10 s (3000 × g), and transferred to autosampler vials for LC/MS/MS analysis. 2.5. LC/MS/MS instrumentation parameters Peptide separations were performed on an Agilent 1200 Capillary LC system. The analytical column utilized was a 150 mm × 1 mm i.d. Symmetry300 reverse phase C18 (3.5 ␮m particle size, Waters Corporation, Milford, MA). The column temperature was held at 40 ◦ C while the autosampler held the samples at 4 ◦ C. The aqueous mobile phase (A) consisted of HPLC-grade water with 0.1% formic acid, while the organic phase (B) was acetonitrile (ACN) with 0.1% formic acid. A 2 ␮L full loop injection with 3-times loop overfill was utilized for injections. A gradient profile was utilized at a flow rate of 30 ␮L/min. Initially, the mobile phase consisted of 98% A and 2% B where it was held for 5 min. From 5 min to 20 min, the B solvent was increased to 20% and at 25 min to 25%. At 27 min the gradient was stepped to 2% A and 98% B for 8 min to clean the column, and then stepped to 98% A and 2% B for the next 20 min to equilibrate the column to initial conditions. The total run time was 57 min. The column eluent was introduced into a Thermo TSQ Vantage triple quadrupole mass spectrometer with an electrospray interface (Thermo Scientific, Waltham, MA). The instrument was operated in the positive ion mode with multiple reaction monitoring (MRM) m/z quantifying transitions pairs as shown in Tables 1 and 2. Two additional transition pairs utilizing the same conditions were monitored for peptide confirmation. Instrument parameters were as follows: spray voltage 3400 V, sheath gas 15, auxiliary gas 0, capillary tube temperature 270 ◦ C, and a collision gas of 1.5 m Torr. Collision energies and tube lens were optimized for each peptide. Data processing and instrument control were performed with the Thermo Scientific Xcalibur software 3. Results and discussion Absolute quantification by isotope dilution mass spectrometry (IDMS) involves selecting specific target peptides as a stoichiometric representative of the proteins from which they are cleaved. A known amount of synthetic peptide, in which one amino acid has been isotopically labeled, is spiked into the sample. Quantification is achieved by comparing the peak area of the isotopically labeled peptide with that of the endogenous target peptide that is generated by proteolytic cleavage of the target protein. This method employs enzymatic digestion to generate unique peptide fragments, which are then separated by liquid chromatography and quantified by multiple reaction monitoring. The accuracy of

this method is dependent upon complete digestion of the protein into the specific peptides used as quantification targets. The underlying assumption for protein quantification via this liquid chromatography-tandem mass spectrometry (LC/MS/MS) method is that there is one unique peptide per protein and that one mole of the measured peptide correlates to one mole of the intact protein. Therefore, steps were taken to ensure that the digestion parameters were optimized for efficient and complete protein digestion [13]. An incomplete digestion of the protein will lead to an underestimation of the true protein quantification. In order to ensure complete digestion of hemagglutinin which is the primary antigen of the influenza vaccine and the amount of which is regulated by government agencies, we quantify three unique peptides from different regions of the protein. Vastly different amounts of any of the peptides from a protein would suggest possible issues arising from steric hindrance of a target peptide due to the higher order structure of the protein or differences in optimum digestion ranges for the target peptides. Obtaining similar results using three peptides that are quantified independently indicates that the protein is completely digested in the region of the target peptides. The three peptides that are used for the quantification of HA for each influenza subtype in a 2009/2010 seasonal vaccine are shown in Fig. 1. Choosing three peptides that are suitable for protein quantification is not trivial. The general rules on what makes a good peptide for quantification have been described elsewhere [17,18]. Briefly, peptides containing oxidizable amino acids such as methionine [19] and tryptophan [20] should be avoided as multiple states of the peptide will exist requiring the monitoring and quantification of all peptide forms. This can be performed and does provide for an accurate determination of the protein quantity (data not shown), but the complexity of the analysis is increased 2- or 4-fold for that particular peptide. Peptides containing cysteine must be rejected as a choice of target peptide for quantification since reduction of the sample would be required to break the disulfide bonds and subsequent alkylation of the sample necessary to prevent the disulfide bonds from reforming. Alkylating agents such as iodoacetamide are known to alkylate other amino acids forming undesired byproducts [21,22]. If alkylation occurs at an unpredicted site on a target peptide, the quantification method would be adversely affected. Therefore, it is best to refrain from reducing and alkylating the sample, making the use of cysteine containing peptides impractical for targeted quantification methods. Peptides containing the consensus sequence Asn-X-Ser/Thr are also avoided as this is a potential site for N-linked glycosylation, a very heterogeneous post-translational modification in which the mass of the glycan cannot be accurately predicted [23]. Such a targeted quantification method requires knowledge of the accurate mass of the peptides to be used for quantification. For this reason, all amino acids and sample preparation protocols that can introduce unpredictable changes in the mass of the peptide must be avoided. Ideally, the peptides chosen would be distally separated on the protein to ensure that the entire protein was sufficiently denatured and digested prior to quantification. Since the HA is cleaved into HA1 and HA2 during the fusion of the hemagglutinin into the cell, we attempted to choose a peptide from both structural regions of the protein. Such is possible for H1N1 and influenza B strains in which two peptides are on one side of the cleavage site (HA1) and the third peptide is on the other (HA2). The arginine, which is the site of cleavage into the two HA subunits is in bold font in Fig. 1. However, this was not possible for H3N2 strains as there were no suitable target peptides in the HA1 portion of the protein due to multiple glycosylation sites and numerous cysteine residues. Therefore, all three peptides of H3N2 strains are located on the HA2 portion of the protein. We have focused our efforts on identifying peptides that are conserved among HA subtypes of concern, either seasonal influenza

T.L. Williams et al. / Vaccine 30 (2012) 2475–2482

2479

Fig. 2. The time-line showing the speed in which a new target peptide can be produced to allow an unpredicted sequence change to be adjusted for in the quantification method. On April 23rd, 2009, the sequence of the swine-origin H1N1 strain was made available. By May 7th, 2009, we had a third peptide standard prepared and ready for use.

viruses or emerging avian viruses with pandemic potential (i.e., H5 from H5N1). For example, in the case of H3N2 strains, DEALNNR is conserved in 665 of the 668 strains (99.6%) that have been sequenced in this region of the protein and for all available sequences deposited in the NCBI Influenza Virus Resource Database between 2007 and November 5, 2010 [24]. EFSEVEGR is conserved in 669 of the 767 strains (87.2%) and STQAAIDQINGK is conserved in 722 of the 734 strains (98.4%) that have been sequenced and deposited in the database. However, when this same database was searched for all strains that had been deposited prior to February 2007, only 485 of the 1739 strains (27.9%) had the sequence STQAAIDQINGK while 1204 strains (69.2%) had the sequence STQAAINQINGK. To be best prepared to quantify any H3N2 strain, both peptides were incorporated into the quantification method. In addition, there has been a change in the circulating influenza B strains from those of a Yamagata lineage to those in the Victoria

lineage. While two of the peptides used for quantification of HA of influenza B are conserved among strains of both lineages, GVLLPQK is found only in strains of the Yamagata lineage while GILLPQK is a peptide conserved by strains of Victoria lineage. If using only the two peptides that are conserved for quantification, then no distinction could be made between the two lineages. By including all four peptides, not only are three peptides used for accurate quantitative measurement, but the lineage of the influenza strain could be determined a priori. If vaccines began to include a B strain of each lineage (resulting in a tetravalent vaccine), then both strains could be quantified in a single experiment using the peptide that is lineage specific. Should a strain emerge that has a mutation which causes a change in the amino acid sequence of the selected target peptide, the sequence of a new peptide target can easily be identified and a new isotopically labeled peptide can be synthesized and accurately characterized in a matter of a few weeks. This was demonstrated

Fig. 3. Liquid chromatography-multiple reaction monitoring chromatogram of a commercial vaccine in which 17 peptide pairs were simultaneously monitored and all MRM transitions were co-added. The peptides in red are for quantification of H3N2 subtypes, the peptides in black are for quantification of H1N1 subtypes, and the peptides in blue are for the quantification of HA and NA from influenza B. The peptides that are underlined are for NA quantification while those that are not underlined are used for HA quantification.

19.6 ± 2.1 10.9 ± 1.1

The amount of HA determined using the individual peptides is shown as well as the average and standard deviation using all three peptides. An “X” indicates that the peptide was not present that in that particular strain of HA or NA. Units are ␮g/500 ␮L vaccine dose.

5.2 X 20.7 20.6 32.2 ± 3.5 17.3 1.1 10.3 X X 16.4

13.2

X

2.1

X

14.7 ± 1.9 10.5

11.9

4.0 35.0 X 34.3 19.4 ± 1.2 29.2 1.9 13.2 X X 22.7

19.3

2.4

X

X

19.3 ± 1.6 14.0

17

2.0 20.0 X 19.5 18.7 3.1 18.7 20.5 X 18.6 6.6 X X 17.6 20.6 X

2010 20.2 Commercial Vaccine 19.4 ± 1.9 19.1 2009/2010 Commercial Vaccine 20.4 ± 2.2 15.9 2007/2008 Commercial Vaccine 15.1 ± 2.1

SHFA NLK NLNSLSE LEVK SGYSGIF SVEGK STQAAIN QINGK STQAAID QINGK YNGIIT DTIK YNGII TETIK YNGIIT GTIK EQLSSVSS FER

TLDFHD SNVK

TLDYHD SNVK

VNSVIEK

H1N1 NA H1N1 hemagglutinin

Table 3 Three seasonal vaccines were analyzed and HA and NA were quantified together in one analytical run.

H3N2 hemagglutinin

DEAL NNR

EFSE VEGR

H3N2 NA

B hemagglutinin

GVLL PQK

GILL PQK

LNVETD TAEIR

T.L. Williams et al. / Vaccine 30 (2012) 2475–2482

B NA

2480

during the swine-originating H1N1 influenza pandemic of 2009. The sequence of A/California/7/2009 became available in April of 2009 and it was immediately clear that one of the three peptides, TLDFHDSNVK, which we had chosen for seasonal H1N1 quantification, had an amino acid change which required the synthesis of a new peptide. The other two peptides, EQLSSVSSFER and VNSVIEK, were conserved and could be used for the quantification of the HA in these pandemic strains in absence of the third peptide. However, TLDYHDSNVK was synthesized, aliquoted into vials, and in-house amino acid analysis was performed within two weeks. The timeline of this process is shown in Fig. 2. This flexibility as well as the sensitivity and selectivity of the LC/MS/MS method would enable rapid quantification of HA in newly emerging strains of influenza, and consequently, it would shorten development time for new vaccines. Quantification of neuraminidase was incorporated into the method, but currently only one peptide per subtype is employed. Neuraminidase is not currently a regulated component of vaccines distributed in the U.S. and the only criterion for vaccines sold in European markets is proof of the presence of neuraminidase. However, knowledge of how much neuraminidase is in a sample is useful for characterization of the virus including information on the variability of the amount of HA to NA in each particular strain. There is not much information on whether having NA in the influenza vaccine is helpful, detrimental, or ambivalent. The ability to accurately determine both HA and NA in an influenza strain will enable conclusions to be drawn on whether the NA should be considered an important aspect for the efficacy of influenza vaccines or if too much NA can lead to issues with vaccine safety. If regulatory agencies become interested in quantifying NA, then the time and cost required to add additional peptides to the method may be warranted. However, since NA is digested and quantified at the same time as HA, we assume that if digestion is complete for HA that it is likely complete for similarly sized NA. For protein quantification purposes related to research or characterization, one peptide is sufficient. The pitfall of having only one peptide is that if there is an antigenic change in the protein which is reflected in an amino acid mutation in the peptide used for quantification, a new peptide would need to be synthesized prior to analysis. We have shown that a new peptide can be prepared as a standard for quantification within 2 weeks (Fig. 2). For regulatory purposes, where amount of HA is directly related to efficacy of a vaccine, three peptides are used for triplicate independent verification of the protein amount. While only three influenza subtypes were the focus of this method, three individual peptides must be used for different N1 species. YNGIITDTIK is used for H5N1 strains and the newly circulating pandemic H1N1 strains while YNGIITETIK and YNGIITGTIK are used for seasonal H1N1 strains prior to 2009. YNGIITETIK is the sequence of most of the strains collected before 2007 while YNGIITGTIK is predominantly present in the more recent (2007–2009) strains not of swine origin. By including all of these peptides into the method, any sample can be analyzed without prior knowledge of its identity, and does not require any modifications to the method. The seventeen peptide targets are shown as a total ion chromatogram in Fig. 3. While there is some co-elution of peptides in the chromatographic run, the MRM transitions are specific for each peptide and there is no overlap of peaks in the extracted ion chromatograms, thereby permitting facile quantification of the target peptides. The peptides in red are for quantification of H3N2 subtypes, the peptides in blue are for quantification of influenza B, and the peptides in black are for quantification of H1N1 subtypes. Those that are underlined are for NA and the remainder are those of HA. All peptide standards used for spiking unknowns can be combined in one vial facilitating the sample preparation method in an infectious disease laboratory. Since the sample preparation includes heat, a

T.L. Williams et al. / Vaccine 30 (2012) 2475–2482

A

2481

10000 5000

VNSVIEK

10000

VNSVIEK

5000 4000

TLDFHDSNVK

Area Counts

2000 4000

TLDFHDSNVK

2000 4000

TLDYHDSNVK

2000 4000

TLDYHDSNVK

2000 15000 10000

EQLSSVSSFER

5000 15000 10000

EQLSSVSSFER

5000 0

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

Time (min)

B

40000

TIC

20000 15000 10000

YNGIITGTIK

5000 15000

YNGIITGTIK

Area Counts

10000 5000 15000 10000

YNGIITDTIK

5000 15000

YNGIITDTIK

10000 5000 15000 10000

YNGIITETIK

5000 15000

YNGIITETIK

10000 5000 0

2

4

6

8

10

12

14

16

18

20

22

24

26

28

Time (min) Fig. 4. Liquid chromatography-multiple reaction monitoring chromatogram of an H1N1 sample produced in MDCK cells. Prior to analysis, the strain and subtype was unknown, but could be determined based upon which peptides were detected in the analysis. With one analytical experiment (A) HA and (B) NA of any subtype and strain can be determined and quantified. Because TLDYHDNSVK of HA was detected rather than TLDFHDSNVK and YNGIITDTIK of NA was detected rather than YNGIITGTIK or YNGIITETIK, it can be concluded that this was an H1N1 strain of swine origin.

2482

T.L. Williams et al. / Vaccine 30 (2012) 2475–2482

denaturing surfactant, enzymatic digestion, and reducing the pH of the sample, inactivation of the virion is ensured. Therefore, there is no need for to inactivate the virus using formaldehyde or BPL prior to performing the sample preparation described herein. This makes the method valuable for the analysis of crude allantoic fluid. A sample that was produced in Madin Darby canine kidney (MDCK) cells was digested and the LC-IDMS method performed. Since HA and NA peptides are all monitored, it was a simple matter to determine that this influenza strain was an H1 strain due to the presence of two HA target peptides, EQLSSVSSFER and VNSVIEK as shown in Fig. 4. Also, it was obvious that this strain was a post2009 pandemic strain since TLDYHDSNVK was observed rather than TLDFHDSNVK. This was also confirmed by the chromatograms of the N1 target peptides in which YNGIITDTIK was observed and the other two N1 neuraminidase peptides were not. This allows for blind analysis of samples and does not require changing the method if the strain of interest changes. If a new strain were to emerge which required the analysis of another peptide, that peptide would simply be incorporated into this existing comprehensive method. Three commercial seasonal vaccines were analyzed using the same all-inclusive method. Data is shown in Table 3. In one analytical run, all three HA peptides from the three seasonal strains were analyzed as were the NA peptides. Agreement between the three HA peptides is routinely better than 85% and usually better than 90% for all subtypes indicating that the digestion method is robust and suitable for all strains included in vaccines of different years. This analysis allows for the ratio of HA to NA to be determined for each strain included in the annual vaccines. The HA to NA ratio varies from as low as 3 in the H1N1 strain used in the 2010 vaccine to as high as 10 in the H3N2 strain used in the 2007/2008 commercial vaccine. It is not clear that there is any biological significance to this ratio or that it affects the safety or efficacy of the vaccine, but it is the first time that the amount of both of these membrane proteins can be determined accurately and precisely and in one analytical method. This information may be useful for providing insight into drug resistant strains of influenza and fast growing strains to be used to make influenza vaccines. 4. Conclusions The current SRID method to quantify the HA content of seasonal and pandemic influenza vaccines relies on the use of strain-specific virus reagents, which have to be prepared every time each component of the vaccine is changed. SRID is not currently used to quantify NA and as such, there is no regulatory requirement which specifies the amount of NA expected to be present in a vaccine. The IDMS method offers the ability to quantify both membranebound antigens from all three influenza subtypes in one analytical run. The method not only provides much improved precision over the SRID method, but excellent accuracy and a shorter analysis time. The synthetic reference peptides used as the accuracy base for the IDMS method are quantified with precision and accuracy that is unattainable with traditional methods of protein quantification. We have focused our efforts on identifying peptides that are conserved among HA and NA of strains that are included in seasonal vaccines. However, should a strain emerge that has a mutation in the selected peptide, the sequence of a new peptide target can easily be identified and a new isotopically labeled peptide can be synthesized and accurately characterized in a matter of a few weeks. The flexibility and sensitivity of the LC/MS/MS method

would enable rapid quantification of HA in newly emerging strains of influenza, and consequently, it would shorten development time for new vaccines. The LC/MS/MS method is not limited to the quantification of influenza HA and NA. Peptides from other viral proteins including nucleoprotein and matrix protein have been identified (data not shown) and can be quantified either separately or incorporated into the method to be detected and quantified at the same time as HA and NA. The mass spectrometer offers a high level of selectivity based on the mass and sequence of the peptide so that multiple proteins from the three strain subtypes of the trivalent seasonal vaccine can be simultaneously quantified. Acknowledgments We thank Dr. Adrian Woolfitt and Ms. Maria Solano for AAA analyses of all peptides used as standards in this research. Viruses grown in MDCK cells were provided by Dr. Vasiliy Mishin of CDC’s Influenza Division. We also appreciate all of the members of the Biological Mass Spectrometry Laboratory who helped make the many vials of peptide standards. Reference in this article to any specific commercial products, process service, manufacturer, or company does not constitute an endorsement or a recommendation by the U.S. government or the Centers for Disease Control and Prevention. The findings and conclusions reported in this article are those of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention. References [1] Yoden S, Kida H, Kuwabara M, Yanagawa R, Webster RG. Virus Res 1986;4:251–61. [2] Kida H, Webster RG, Yanagawa R. Arch Virol 1983;76:91–9. [3] Virelizier JL. J Immunol 1975;115:434–9. [4] Kilbourne ED, Couch RB, Kasel JA, Keitel WA, Cate TR, Quarles JH, Grajower B, Pokorny BA, Johansson BE. Vaccine 1995;13:1799–803. [5] Johansson BE, Moran TM, Kilbourne ED. Proc Natl Acad Sci U S A 1987;84:6869–73. [6] Sylte MJ, Hubby B, Suarez DL. Vaccine 2007;25:3763–72. [7] Sylte MJ, Suarez DL. Curr Top Microbiol Immunol 2009;333:227–41. [8] Johansson BE, Bucher DJ, Kilbourne ED. J Virol 1989;63:1239–46. [9] Cox NJ, Tamblyn SE, Tam T. Vaccine 2003;21:1801–3. [10] Wood JM, Schild GC, Newman RW, Seagroatt V. Dev Biol Stand 1977;39:193–200. [11] Wood JM, Schild GC, Newman RW, Seagroatt V. J Biol Stand 1977;5:237–47. [12] Williams TL, Luna L, Guo Z, Cox NJ, Pirkle JL, Donis RO, Barr JR. Vaccine 2008;26:2510–20. [13] Norrgran J, Williams TL, Woolfitt AR, Solano MI, Pirkle JL, Barr JR. Anal Biochem 2009;393:48–55. [14] Perrin P, Morgeaux S. Biologicals 1995;23:207–11. [15] Woolfitt AR, Solano MI, Williams TL, Pirkle JL, Barr JR. Anal Chem 2009;81:3979–85. [16] Yu YQ, Gilar M, Lee PJ, Bouvier ES, Gebler JC. Anal Chem 2003;75:6023–8. [17] Stevenson SE, Houston NL, Thelen JJ. Regul Toxicol Pharmacol 2010;58:S36–41. [18] Andrews, P., Arnott, J., Farmar, J., Ivanov, A., Kowalak, J., Lane, W., Mechtler, K., Ogorzalek Loo, R., Raida, M. The Association of Biomolecular Resource Facilities: Salt Lake City, 2008. [19] Mo W, Ma Y, Takao T, Neubert TA. Rapid Commun Mass Spectrom 2000;14:2080–1. [20] Taylor SW, Fahy E, Murray J, Capaldi RA, Ghosh SS. J Biol Chem 2003;278:19587–90. [21] Yang Z, Attygalle AB. J Mass Spectrom 2007;42:233–43. [22] Lundell N, Schreitmuller T. Anal Biochem 1999;266:31–47. [23] Blake TA, Williams TL, Pirkle JL, Barr JR. Anal Chem 2009;81:3109–18. [24] Sayers EW, Barrett T, Benson DA, Bryant SH, Canese K, Chetvernin V, Church DM, DiCuccio M, Edgar R, Federhen S, Feolo M, Geer LY, Helmberg W, Kapustin Y, Landsman D, Lipman DJ, Madden TL, Maglott DR, Miller V, Mizrachi I, Ostell J, Pruitt KD, Schuler GD, Sequeira E, Sherry ST, Shumway M, Sirotkin K, Souvorov A, Starchenko G, Tatusova TA, Wagner L, Yaschenko E, Ye J. Nucleic Acids Res 2009;37:D5–15.