12
Impact of a human CMP-sialic acid transporter on recombinant glycoprotein sialylation in glycoengineered insect cells Hideaki Mabashi-Asazuma 2 , Xianzong Shi 2 , Christoph Geisler 2 , Chu-Wei Kuo 3 , Kay-Hooi Khoo 3 , and Donald L. Jarvis 1,2 2 Department of Molecular Biology, University of Wyoming, Laramie WY 82071, USA and 3 Institute of Biological Chemistry, Academia Sinica 128, Nankang, Taipei 115, Taiwan Received on March 28, 2012; revised on October 3, 2012; accepted on October 7, 2012 Insect cells are widely used for recombinant glycoprotein production, but they cannot provide the glycosylation pat- terns required for some biotechnological applications. This problem has been addressed by genetically engineering insect cells to express mammalian genes encoding various glycoprotein glycan processing functions. However, for various reasons, the impact of a mammalian cytosine-5- monophospho (CMP)-sialic acid transporter has not yet been examined. Thus, we transformed Spodoptera frugi- perda (Sf9) cells with six mammalian genes to generate a new cell line, SfSWT-4, that can produce sialylated glycopro- teins when cultured with the sialic acid precursor, N-acetyl- mannosamine. We then super-transformed SfSWT-4 with a human CMP-sialic acid transporter (hCSAT) gene to isolate a daughter cell line, SfSWT-6, which expressed the hCSAT gene in addition to the other mammalian glycogenes. SfSWT-6 cells had higher levels of cell surface sialylation and also supported higher levels of recombinant glycopro- tein sialylation, particularly when cultured with low concen- trations of N-acetylmannosamine. Thus, hCSAT expression has an impact on glycoprotein sialylation, can reduce the cost of recombinant glycoprotein production and therefore should be included in ongoing efforts to glycoengineer the baculovirus-insect cell system. The results of this study also contributed new insights into the endogenous mechanism and potential mechanisms of CMP-sialic acid accumulation in the Golgi apparatus of lepidopteran insect cells. Keywords: baculovirus / CMP-sialic acid transport / glycoengineering / glycoprotein sialylation / insect cell glycosylation Introduction Insect cells produce N-glycoproteins with relatively simple oligosaccharide side-chains, or glycans, which lack the ter- minal sialic acid residues often found on the N-glycans pro- duced by mammalian cells (Marz et al. 1995; Altmann et al. 1999; Harrison and Jarvis 2006; Geisler and Jarvis 2009). One reason for this important structural difference between in- vertebrate and vertebrate N-glycoproteins is that insect cells generally lack adequate levels of the glycosyltransferases needed to elongate trimmed N-glycan processing intermedi- ates and synthesize complex, terminally sialylated end- products (Stollar et al. 1976; Butters et al. 1981; Hooker et al. 1999). Another is that insect cells have an additional N-glycan trimming enzyme that is not found in mammalian cells and that specically removes terminal β1,2-linked N-acetylglucosamine residues from the lower branch of a key N-glycan processing intermediate (Altmann et al. 1995; Geisler et al. 2008; Geisler and Jarvis 2010, 2012). Thus, the major processed N-glycans found on insect cell glycoproteins, including recombinant glycoproteins produced by insect cells infected with baculovirus expression vectors, are highly trimmed core structures consisting of Man 3 GlcNAc 2 (±Fuc). Substitution of the complex, terminally sialylated N-glycans normally found on native mammalian glycoproteins with these relatively simple, core N-glycans is a major problem associated with the production of recombinant glycoproteins in insect cells because terminal sialic acids inuence many biomedical applications of these products. For example, it is well known that the circulatory half-life and, therefore, the pharmacokinetic behavior and clinical efcacy of therapeutic glycoproteins are strongly inuenced by terminal sialic acids on their mammalian-type N-glycans (Byrne et al. 2007; Durocher and Butler 2009; Sola and Griebenow 2011). Previous research has revealed that glycoengineering approaches can be used to address the biotechnological limita- tions imposed by the relatively simple nature of endogenous insect cell protein glycosylation pathways. Most of this work has involved genetically transforming insect cells with mam- malian genes encoding the glycosyltransferases needed to convert trimmed N-glycans into complex, terminally sialylated structures (Breitbach and Jarvis 2001; Hollister and Jarvis 2001; Hollister et al. 2002; Yun et al. 2005). These efforts have yielded various transgenic derivatives of established lepi- dopteran insect cell lines, including Spodoptera frugiperda 1 To whom correspondence should be addressed: Tel: +1-307-766-4282; Fax: +1-307-766-5098; e-mail: [email protected] Glycobiology vol. 23 no. 2 pp. 199210, 2013 doi:10.1093/glycob/cws143 Advance Access publication on October 12, 2012 © The Author 2012. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: [email protected] 199 at University of Wyoming Libraries on December 27, 2012 http://glycob.oxfordjournals.org/ Downloaded from

Impact of a human CMP-sialic acid transporter on recombinant glycoprotein sialylation in glycoengineered insect cells

Embed Size (px)

Citation preview

Impact of a human CMP-sialic acid transporteron recombinant glycoprotein sialylationin glycoengineered insect cells

Hideaki Mabashi-Asazuma2, Xianzong Shi2,Christoph Geisler2, Chu-Wei Kuo3, Kay-Hooi Khoo3,and Donald L. Jarvis1,2

2Department of Molecular Biology, University of Wyoming, Laramie WY82071, USA and 3Institute of Biological Chemistry, Academia Sinica 128,Nankang, Taipei 115, Taiwan

Received on March 28, 2012; revised on October 3, 2012; accepted onOctober 7, 2012

Insect cells are widely used for recombinant glycoproteinproduction, but they cannot provide the glycosylation pat-terns required for some biotechnological applications. Thisproblem has been addressed by genetically engineeringinsect cells to express mammalian genes encoding variousglycoprotein glycan processing functions. However, forvarious reasons, the impact of a mammalian cytosine-5′-monophospho (CMP)-sialic acid transporter has not yetbeen examined. Thus, we transformed Spodoptera frugi-perda (Sf9) cells with six mammalian genes to generate anew cell line, SfSWT-4, that can produce sialylated glycopro-teins when cultured with the sialic acid precursor, N-acetyl-mannosamine. We then super-transformed SfSWT-4 with ahuman CMP-sialic acid transporter (hCSAT) gene to isolatea daughter cell line, SfSWT-6, which expressed the hCSATgene in addition to the other mammalian glycogenes.SfSWT-6 cells had higher levels of cell surface sialylationand also supported higher levels of recombinant glycopro-tein sialylation, particularly when cultured with low concen-trations of N-acetylmannosamine. Thus, hCSAT expressionhas an impact on glycoprotein sialylation, can reduce thecost of recombinant glycoprotein production and thereforeshould be included in ongoing efforts to glycoengineer thebaculovirus-insect cell system. The results of this study alsocontributed new insights into the endogenous mechanismand potential mechanisms of CMP-sialic acid accumulationin the Golgi apparatus of lepidopteran insect cells.

Keywords: baculovirus / CMP-sialic acidtransport / glycoengineering / glycoprotein sialylation / insectcell glycosylation

Introduction

Insect cells produce N-glycoproteins with relatively simpleoligosaccharide side-chains, or glycans, which lack the ter-minal sialic acid residues often found on the N-glycans pro-duced by mammalian cells (Marz et al. 1995; Altmann et al.1999; Harrison and Jarvis 2006; Geisler and Jarvis 2009).One reason for this important structural difference between in-vertebrate and vertebrate N-glycoproteins is that insect cellsgenerally lack adequate levels of the glycosyltransferasesneeded to elongate trimmed N-glycan processing intermedi-ates and synthesize complex, terminally sialylated end-products (Stollar et al. 1976; Butters et al. 1981; Hooker et al.1999). Another is that insect cells have an additionalN-glycan trimming enzyme that is not found in mammaliancells and that specifically removes terminal β1,2-linkedN-acetylglucosamine residues from the lower branch of a keyN-glycan processing intermediate (Altmann et al. 1995;Geisler et al. 2008; Geisler and Jarvis 2010, 2012). Thus, themajor processed N-glycans found on insect cell glycoproteins,including recombinant glycoproteins produced by insect cellsinfected with baculovirus expression vectors, are highlytrimmed core structures consisting of Man3GlcNAc2(±Fuc).Substitution of the complex, terminally sialylated N-glycansnormally found on native mammalian glycoproteins withthese relatively simple, core N-glycans is a major problemassociated with the production of recombinant glycoproteinsin insect cells because terminal sialic acids influence manybiomedical applications of these products. For example, it iswell known that the circulatory half-life and, therefore, thepharmacokinetic behavior and clinical efficacy of therapeuticglycoproteins are strongly influenced by terminal sialic acidson their mammalian-type N-glycans (Byrne et al. 2007;Durocher and Butler 2009; Sola and Griebenow 2011).Previous research has revealed that glycoengineering

approaches can be used to address the biotechnological limita-tions imposed by the relatively simple nature of endogenousinsect cell protein glycosylation pathways. Most of this workhas involved genetically transforming insect cells with mam-malian genes encoding the glycosyltransferases needed toconvert trimmed N-glycans into complex, terminally sialylatedstructures (Breitbach and Jarvis 2001; Hollister and Jarvis2001; Hollister et al. 2002; Yun et al. 2005). These effortshave yielded various transgenic derivatives of established lepi-dopteran insect cell lines, including Spodoptera frugiperda

1To whom correspondence should be addressed: Tel: +1-307-766-4282; Fax:+1-307-766-5098; e-mail: [email protected]

Glycobiology vol. 23 no. 2 pp. 199–210, 2013doi:10.1093/glycob/cws143Advance Access publication on October 12, 2012

© The Author 2012. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: [email protected] 199

at University of W

yoming L

ibraries on Decem

ber 27, 2012http://glycob.oxfordjournals.org/

Dow

nloaded from

(Sf9) (Summers and Smith 1987) and BTI-Tn-5B1-4/HighFive™ (Wickham et al. 1992), which are the most commonlyused hosts for baculovirus expression vectors.While it was clear that these transgenic derivatives could

produce sialylated recombinant glycoproteins when infectedby baculovirus expression vectors, this newly acquired abilitywas initially surprising because other studies had shown thatSf9 and High Five™ cells lack the nucleotide sugar, cytosine-5'-monophospho (CMP)-sialic acid, required as the donor sub-strate for glycoprotein sialylation (Hooker et al. 1999; Tomiyaet al. 2001). In addition, given the absence of any detectableCMP-sialic acid, one would not expect these cells to have thetransporter required to specifically import this nucleotidesugar into the Golgi apparatus (Eckhardt et al. 1996), where itcould be used for glycoprotein sialylation. However, these in-consistencies were partially resolved when it was discoveredthat recombinant glycoprotein sialylation by these transgenicinsect cell lines required not only expression of the mamma-lian glycosyltransferase genes, but also an exogenous sourceof sialic acids, such as fetal bovine serum or purified fetuin(Hollister et al. 2003). This new finding suggested that theseinsect cells had an endogenous sialic acid salvaging pathwaythat led to the production and import of CMP-sialic acid forutilization by the mammalian sialyltransferases, which wasable to support recombinant glycoprotein sialylation.Considering that a salvaging pathway might be less efficient

than a de novo biosynthetic pathway, we super-transformed atransgenic insect cell line that had been previously glycoengi-neered to encode the requisite mammalian glycosyltransferaseswith mammalian sialic acid synthase (SAS) and CMP-sialicacid synthetase (CMAS) genes, which are required for sialicacid and CMP-sialic acid biosynthesis (Aumiller et al. 2003).The resulting cell line produced high levels of CMP-sialic acidwhen cultured in serum-free media supplemented with thesialic acid precursor, N-acetylmannosamine. In addition, thesecells were able to produce recombinant glycoproteins with ter-minally sialylated N-glycans after being infected by baculo-virus expression vectors. The ability of these glycoengineeredinsect cell lines to produce sialylated N-glycoproteins in theabsence of an exogenous source of sialic acids implied thatthey had an endogenous mechanism for the accumulation ofCMP-sialic acid in the Golgi apparatus, as this is strictlyrequired for glycoprotein sialylation. This suggested that itmight not be necessary to glycoengineer insect cells to expressa mammalian CMP-sialic acid transporter (abbreviated CSATfor clarity herein; however, the HUGO nomenclaturecommittee-approved gene symbol is SLC35A1) and, therefore,the potential impact of heterologous CSAT gene expressionwas never investigated. In the present study, we examined theimpact of CSAT function on sialylation in insect cells by iso-lating a matched pair of glycoengineered lepidopteran insectcell lines with and without a human CSAT (hCSAT) gene andcomparing their relative capacities for glycoprotein sialylation.We found that the insect cell line expressing the hCSAT genehad higher levels of cell surface sialylation and supportedhigher levels of recombinant glycoprotein sialylation, particu-larly when cultured in low concentrations of N-acetylmannosa-mine. These results showed that the addition of an hCSATgene enhances sialylation, that it can reduce the cost of sialo-

glycoprotein production in glycoengineered baculovirus–insectcell systems and, therefore, that CSAT function should beincluded in ongoing insect cell glycoengineering efforts. Theyalso provided new information on the endogenous capacityand potential mechanisms for the accumulation of CMP-sialicacid in the Golgi apparatus of lepidopteran insect cell lines.

ResultsIsolation of transgenic Sf9 cell lines glycoengineered withand without an hCSAT geneThe general approach used in this study was to isolate amatched pair of transgenic lepidopteran insect cell lines withand without an hCSAT gene, each capable of producing ter-minally sialylated N-glycans when cultured in serum-freemedia supplemented with N-acetylmannosamine. Once thesecell lines were isolated and characterized, we were able to usethem to examine the impact of hCSAT gene expression on re-combinant glycoprotein sialylation by glycoengineered insectcell lines. Based upon previous work, which had establishedthe minimal set of mammalian genes required for this purpose(Aumiller et al. 2003, 2012), we initially transformed Sf9 cellswith human N-acetylglucosaminyltransferase II (hMGAT2),bovine β1,4-galactosyltransferase I (bB4GALT1), mouseα2,3-sialyltransferase (mST3GAL3), rat α2,6-sialyltransferase(rST6GAL1), mouse SAS and mouse CMAS genes, togetherwith a hygromycin B-resistance marker. After selection ingrowth medium supplemented with hygromycin B, single-cellclones were isolated by limiting dilution, pre-screened using acell surface Sambucus nigra agglutinin (SNA) lectin stainingassay and re-screened by RNA dot-blot hybridization, asdescribed in Materials and methods. These steps identified aclone that stained intensely with SNA (data not shown) andscored positive for all the six genes in the dot-blots (data notshown), which was designated SfSWT-4 and used for the re-mainder of this study. We subsequently super-transformedSfSWT-4 cells with an hCSAT gene and neomycin-resistancemarker and, after selection in growth medium supplementedwith Geneticin®, we re-isolated single-cell clones by limitingdilution and screened them for hCSAT expression by RNAdot-blot hybridization. These steps identified a positive clone(data not shown) that was designated SfSWT-6 and used forthe remainder of this study.

Mammalian transgene expression in SfSWT-4and SfSWT-6 cellsReverse transcriptase–polymerase chain reaction (RT–PCR)assays with total RNA preparations and gene-specific primerswere performed to examine the expression of each mammaliantransgene in uninfected and baculovirus-infected SfSWT-4 andSfSWT-6 cells. Total RNA from the parental Sf9 cells wasused as a negative control and primers specific to an endogen-ous Sf9 ribosomal protein (L3) gene were used as positive con-trols. Duplicate assays were performed in parallel with andwithout the addition of RT to assess possible contamination ofthe total RNA preparations with DNA. The results showed thatnone of the RT–PCR assays yielded any detectable amplifica-tion products when performed without any RT, indicating that

H Mabashi-Asazuma et al.

200

at University of W

yoming L

ibraries on Decem

ber 27, 2012http://glycob.oxfordjournals.org/

Dow

nloaded from

the RNA preparations were not detectably contaminated withDNA (Figure 1, RT−). The results also showed that all of theRT–PCR assays yielded an amplification product of theexpected size when performed with RT and primers specific tothe endogenous Sf9 cell ribosomal protein gene, which vali-dated our RT–PCR method and provided an internal standardfor the assays (Figure 1, RT+; SfRPL3). None of the RT–PCRassays yielded an amplification product when they includedRT, total RNA from the parental Sf9 cells and primers specificto any of the mammalian transgenes, which further validatedour RT–PCR method by establishing its specificity (Figure 1,RT+; hMGAT2, bB4GalT1, mST3GAL3, rST6GAL1, mSAS,mCMAS, hCSAT). In contrast, all of the assays yielded anamplification product of the expected size when they were per-formed with RT, total RNA from either uninfected orbaculovirus-infected SfSWT-4 or -6 cells and primers specificto each mammalian transgene except hCSAT (Figure 1, RT+;hMGAT2, bB4GalT1, mST3GAL3, rST6GAL1, mSAS,mCMAS). Finally, the RT–PCR assay that included RT, totalRNA from uninfected or baculovirus-infected SfSWT-6 cellsand primers specific to the hCSAT gene also yielded an ampli-fication product of the expected size (Figure 1, RT+; hCSAT).Together, these results demonstrated that the two new transgen-ic insect cell lines produced for this study contained andexpressed a common set of mammalian glycogenes at approxi-mately equal levels. In addition, the SfSWT-6, but not theSfSWT-4 cells expressed the hCSAT gene, as expected.

Finally, the expression of each transgene appeared to beenhanced by baculovirus infection, as anticipated from thissame result in previous studies (Jarvis 1993).

Analysis of hCSAT function in glycoengineered insect cellsAvalid analysis of the impact of the hCSAT gene on glycoen-gineered insect cells required evidence that the gene productfunctions as a CMP-sialic acid transporter when expressed ininsect cells. Hence, we assayed hCSAT function in micro-somal membranes from SfSWT-4 and SfSWT-6 cells, withmembranes from Chinese hamster ovary (CHO) cells as thepositive and membranes from Lec 2, a CSAT mutant CHOcell line (Deutscher et al. 1984) with reduced CSAT activity,as the negative control. Upon incubation with radiolabeledCMP-sialic acid, we observed only low levels of transportwith the microsomal membranes isolated from SfSWT-4 cells(Figure 2). We were unable to determine whether the level ofuptake observed with the SfSWT-4 cell membranes wasabove background because there is no CSAT null mutant thatcould have been used to establish the background of the assayfor this purpose. Nevertheless, the results clearly demonstratedthat the microsomal membranes from SfSWT-6 cells trans-ported CMP-sialic acid at about 4-fold higher levels thanmicrosomes from SfSWT-4 or Lec 2 cells and at about thesame level as microsomes from CHO cells (Figure 2). Theseresults showed that hCSAT gene expression leads to the

Fig. 1. Transgene expression in uninfected and baculovirus-infected SfSWT-4 and SfSWT-6 cells. Total RNA isolated from uninfected or baculovirus-infectedSf9, SfSWT-4 and SfSWT-6 cells was used for RT–PCR, as described in Materials and methods. Each reaction was performed in the presence (RT+) or absence(RT−) of reverse transcriptase to assess DNA contamination of the RNA preparations. The primer pairs used for these RT–PCR were specific to an endogenousSf9 cell gene encoding ribosomal protein L3 or for each of the individual mammalian genes used to transform Sf9 cells and produce the transgenic derivatives,as indicated by the labels on the left-hand side of the figure. hMGAT2, human β1,2-N-acetylglucosaminyltransferase II; bB4GalT1, bovineβ1,4-galactosyltransferase I; rST6GAL1, rat α2,6-sialyltransferase I; mST3GAL3, murine α2,3-sialyltransferase III; mSAS, murine sialic acid synthase; mCMAS,murine CMP-sialic acid synthetase; hCSAT, human CMP-sialic acid transporter; SfRPL3, Spodoptera frugiperda ribosomal protein L3.

Human CMP-sialic acid transporter impact on insect sialylation

201

at University of W

yoming L

ibraries on Decem

ber 27, 2012http://glycob.oxfordjournals.org/

Dow

nloaded from

production of a functional CMP-sialic acid transporter in gly-coengineered insect cells.

Cell surface sialylation levels in SfSWT-4 and SfSWT-6 cellsSfSWT-4 and SfSWT-6 cells were infected with a wild-typebaculovirus to increase transgene expression (Jarvis 1993),cultured in serum-free media supplemented with 10 mMN-acetylmannosamine, which is a standard concentration usedfor this sialic acid precursor, and then cell surface glycosyla-tion patterns were examined in lectin staining assays with Sf9cells as controls, as described in Materials and methods. Theparental insect cells were not specifically stained, while theglycoengineered insect cells were stained with the sialic acid-specific lectin, SNA, as expected (Figure 3A). A quantitativeanalysis of the results revealed that SfSWT-6 expressed slight-ly, but significantly higher levels of cell surface sialic acidsthan SfSWT-4 cells (Figure 3B).To extend these results, we performed a cell surface SNA

staining assay on SfSWT-4 and -6 cells cultured in serum-freemedia supplemented with relatively lower (1 mM) and higher(50 mM) concentrations of N-acetylmannosamine. The SNAstaining intensities were about the same when these cells werecultured in high concentrations, but the SfSWT-6 cells weremore intensely stained when both were cultured in low con-centrations of this sialic acid precursor (Figure 4A and B).These results indicated that the hCSAT gene produced a func-tional CMP-sialic acid transporter, which had an impact oninsect cell sialylation by supporting higher levels of cell

surface sialylation, particularly when the cells were culturedin a low concentration of N-acetylmannosamine.

Impact of hCSAT expression on recombinantglycoprotein sialylationFinally, we examined the impact of hCSAT expression on sia-lylation of the biotechnologically relevant recombinant glyco-protein, human erythropoietin (hEPO), during baculovirusinfection of the two glycoengineered insect cell lines isolatedfor this study. SfSWT-4 and SfSWT-6 cells were infected witha recombinant baculovirus encoding a His-tagged version ofhEPO (hEPO-His) and then the infected cells were fed withserum-free media supplemented either with 1 or with 50 mMN-acetylmannosamine. The extracellular growth media wereharvested at 48 h after infection and the recombinanthEPO-His was affinity purified and analyzed by SNA lectinblotting. Commercial fetuin was used as the positive andneuraminidase-treated samples of hEPO-His and fetuin wereused as the negative controls. The results showed that SNAbound to untreated fetuin and hEPO-His, but not to either ofthe neuraminidase treated proteins, which validated the speci-ficity of the lectin blotting assay (Figure 5A). The results alsoshowed that SNA bound about equally well to the hEPO-Hispreparations from SfSWT-4 and -6 cells cultured with thehigh concentration of N-acetylmannosamine (Figure 5A).However, when both cell lines were cultured with low con-centrations of N-acetylmannosamine, SNA binding wasobserved only with the hEPO-His produced by SfSWT-6 cells(Figure 5A). These results were confirmed and extended by a

Fig. 2. In vitro CMP-sialic acid transport assays. Samples of microsomepreparations from SfSWT-4 and SfSWT-6 containing equal amounts of totalprotein were used for an in vitro assay of CMP-sialic acid transport, asdescribed in Materials and methods. Normalized samples of microsomes fromCHO and a CMP-sialic acid transport defective mutant CHO cell line (Lec2)were used as positive and negative controls, respectively. The results arepresented as the mean ± standard deviation obtained in four independentmeasurements. The statistical significance of the differences in CMP-sialicacid transport levels relative to the Lec2 negative control were determined byone-way ANOVA analysis.

Fig. 3. Cell surface sialylation. Sf9, SfSWT-4 and SfSWT-6 cells were seededinto culture plates, infected with a wild-type baculovirus, incubated for 36 hafter infection in growth media supplemented with 10 mMN-acetylmannosamine and then stained with SNA, as described in Materialsand methods. The cell surface staining patterns are shown in (A) and therelative fluorescence intensities measured using Image J, as described inMaterials and methods, are shown in (B). The error bars show the standarddeviations calculated using the results obtained with three independentsamples. One-way ANOVA analysis showed that there was a statisticallysignificant difference in the relative fluorescence intensities observed withSfSWT-4 and -6 cells (P < 0.005).

H Mabashi-Asazuma et al.

202

at University of W

yoming L

ibraries on Decem

ber 27, 2012http://glycob.oxfordjournals.org/

Dow

nloaded from

quantitative analysis of the sialic acids enzymatically releasedfrom the hEPO-His preparations purified from SfSWT-4 and-6 cells cultured in low (1 mM) or high (50 mM) concentra-tions of N-acetylmannosamine (Figure 5B).To more directly assess the impact of hCSAT expression on

hEPO sialylation, we enzymatically released and permethy-lated the N-glycans from each purified hEPO preparation anddetermined their structures by MALDI-TOF MS, as describedin Materials and methods. The results showed that thehEPO preparations from SfSWT-4 cells cultured with lowN-acetylmannosamine concentrations contained the lowest pro-portion, with only tiny peaks having m/z values correspondingto terminally sialylated N-glycans (Figure 6A). The propor-tions of sialylated N-glycans became progressively higher withhEPO preparations from SfSWT-4 cells cultured with 50 mMN-acetylmannosamine (Figure 6B), SfSWT-6 cells culturedwith 1 mM N-acetylmannosamine (Figure 6C) and fromSfSWT-6 cells cultured with 50 mM N-acetylmannosamine(Figure 6D), respectively. The proportions of terminally galac-tosylated (acceptor) and sialylated N-glycans found on each

hEPO preparation are shown in Figure 7. Together, the lectinblotting, quantitative sialic acid determinations and massspectrometric data all consistently demonstrated that hCSATexpression enhanced recombinant hEPO-His sialylation inglycoengineered insect cells, particularly when the cells werecultured in low concentrations of N-acetylmannosamine.

Discussion

The baculovirus-insect cell expression system is widely usedto produce recombinant glycoproteins, but it provides onlyrelatively simple protein glycosylation patterns that are inad-equate for some biomedical applications (Marz et al. 1995;Altmann et al. 1999; Harrison and Jarvis 2006; Geisler andJarvis 2009). This problem has been addressed, at least inpart, by engineering commonly used lepidopteran insect celllines to express mammalian glycosyltransferases (Breitbachand Jarvis 2001; Hollister and Jarvis 2001; Hollister et al.2002; Yun et al. 2005), as well as certain enzymes involvedin the de novo biosynthesis of CMP-sialic acid (Aumilleret al. 2003, 2012). This effort has yielded glycoengineeredinsect cell lines with modified glycosylation pathways that canproduce recombinant glycoproteins with authentic, human-type protein N-glycosylation patterns that include terminalsialic acids. These results imply that Sf9 and High Five™cells have the endogenous capacity to accumulate CMP-sialicacid in the Golgi apparatus, where it is required as the donorsubstrate for glycoprotein sialylation. However, neither theprecise nature of the infrastructure supporting this endogenouscapacity nor the impact of mammalian CSAT gene expressionon glycoprotein sialylation in insect cells had been examined.This study focused on the latter issue and provided newinsights into the former.Our basic approach was to isolate a matched set of glycoen-

gineered insect cell lines, both capable of glycoprotein sialyla-tion in the absence of an exogenous source of sialic acids,one with and the other without an hCSAT gene. We success-fully isolated both cell lines and demonstrated that they hadcomparable transgene expression patterns, except for the factthat the hCSAT gene was expressed only in SfWT-6 cells.Further analysis showed that expression of the hCSAT geneled to the production of a functional CSAT in SfSWT-6 cells.This set the stage for a direct comparison of the sialylationcapacity of these two matched cell lines in the presence andabsence of hCSAT function.Initially, we used an SNA staining assay to compare cell

surface sialylation levels and found that SfSWT-6 cellsstained more intensely than SfSWT-4 cells when both werecultured in serum-free growth media containing a standardconcentration (10 mM) of N-acetylmannosamine. This newresult provided our first indication that hCSAT gene expres-sion had an impact in glycoengineered insect cell lines,leading to a higher level of cell surface sialylation. It is in-structive to note that hCSAT expression induced only a rela-tively small (12%) increase, which was consistent with therelatively small (14%) increase in hEPO sialylation observedwhen CHO cells were glycoengineered to express higherlevels of hCSAT (Son et al. 2011). This small increase waspredictable in CHO cells, as it is clear that they have

Fig. 4. Impact of N-acetylmannosamine concentration on cell surfacesialylation. Sf9, SfSWT-4 and SfSWT-6 cells were seeded into culture plates,infected with a wild-type baculovirus, incubated for 36 h after infection ingrowth media supplemented with either 1 or 50 mM N-acetylmannosamineand then stained with SNA to detect cell surface sialylation, as described inMaterials and methods. The cell surface staining patterns are shown in (A)and the relative fluorescence intensities, determined as described in Materialsand methods, are shown in (B). The error bars show the standard deviationscalculated using the results obtained with three independent samples.One-way ANOVA analysis showed that the only statistically significantdifference (P < 0.0001) in relative fluorescence intensities was observed withSfSWT-4 and -6 cells cultured in 1 mM N-acetylmannosamine.

Human CMP-sialic acid transporter impact on insect sialylation

203

at University of W

yoming L

ibraries on Decem

ber 27, 2012http://glycob.oxfordjournals.org/

Dow

nloaded from

endogenous CSAT activity, but not in Sf9 cells, as the mech-anism used to accumulate CMP-sialic acid in the Golgi appar-atus of insect cells remains to be determined.Subsequently, we examined cell surface sialylation in

SfSWT-4 and -6 cells cultured in lower (1 mM) and higher(50 mM) concentrations of N-acetylmannosamine and foundthat the hCSAT gene had little impact (5% increase) when thecells were cultured in high concentrations and a much greaterimpact (59% increase) when the cells were cultured in lowconcentrations of this sialic acid precursor. We obtainedsimilar results when we examined how hCSAT gene expres-sion influenced the sialylation of a His-tagged version of thebiotechnologically relevant glycoprotein, hEPO, which wasexpressed by infecting the glycoengineered insect cells with abaculovirus vector. In these experiments, SfSWT-6 cells sialy-lated hEPO-His at higher levels (35%) than SfSWT-4 whenthe cells were cultured in 50 mM N-acetylmannosamine, butat significantly higher levels (67.5%) when they were culturedin 1 mM N-acetylmannosamine.Together, the results obtained in this study demonstrated

that hCSAT gene expression has an impact on cell surface andrecombinant glycoprotein sialylation in glycoengineered

lepidopteran insect cells, particularly when the cells are cul-tured in growth media supplemented with low concentrationsof N-acetylmannosamine. Indeed, SfSWT-6 cells providedabout the same levels of cell surface and recombinant glyco-protein sialylation when cultured in growth media containingeither 1 mM or 50 mM N-acetylmannosamine. Thus, thisstudy also demonstrated that the addition of an hCSAT geneis an approach that can be used to reduce the cost of recom-binant glycoprotein production by glycoengineered insect cellsystems by reducing the N-acetylmannosamine concentrationneeded to support sialylation. The addition of 10 mMN-acetylmannosamine adds �$14/L to the cost of the insectcell growth medium even when the sialic acid precursor ispurchased at the lowest available, bulk-rate price. Our resultsshowed that the addition of an hCSAT gene reduced the ef-fective concentration of N-acetylmannosamine to at least 1mM, thereby reducing the cost of media supplementation byabout 10-fold.We should note that the sialylation efficiencies obtained in

this study (�0.1–1.0%) were the lowest we have observed todate. In MS analyses of the N-glycan profiles of various re-combinant glycoprotein (Toth et al. unpublished) or total

Fig. 5. Sialylation of hEPO-His. SfSWT-4 and SfSWT-6 cells were infected with a recombinant baculovirus encoding hEPO-His and cultured in serum-freemedia supplemented with 1 or 50 mM N-acetylmannosamine, and hEPO-His was harvested and affinity-purified at 48 h after infection. The results of SNA lectinblotting and immunoblotting assays of samples treated with neuraminidase buffer alone (−) or buffer plus neuraminidase (+) are shown in (A). Fetuin was usedas a positive control. The amounts of sialic acid released by neuraminidase treatment of the purified hEPO-His samples were also determined, as described inMaterials and methods, with the results shown in (B). The error bars show the standard deviations calculated using the results obtained with three independentsamples.

H Mabashi-Asazuma et al.

204

at University of W

yoming L

ibraries on Decem

ber 27, 2012http://glycob.oxfordjournals.org/

Dow

nloaded from

glycoprotein (Jarvis et al. unpublished) preparations fromseveral different glycoengineered insect cell lines, we haveobserved sialylation efficiencies ranging from �15 to �40%.Potential reasons for the relatively low sialylation efficienciesobserved in this study include the nature of the model glyco-protein, the specific glycoengineered insect cell lines and/ortheir metabolic condition during the production runs and/orthe growth conditions used for those production runs.Nevertheless, we were still able to evaluate the impact ofhCSAT expression by using carefully matched cell lines inparallel expression runs, as indicated by the consistency of theresults obtained using complementary assays, and to draw aclear conclusion regarding the need to incorporate hCSAT ex-pression into ongoing insect cell glycoengineering efforts.

The results obtained in this study also provide new insightsinto the endogenous mechanism used to accumulateCMP-sialic acid in the Golgi apparatus of Sf9 cells. The rela-tively lower impact of hCSAT expression on sialylation byglycoengineered Sf9 cells cultured in high concentrations andthe higher impact observed in low concentrations ofN-acetylmannosamine indicates (i) that there is an endogenousCMP-sialic acid accumulation process in Sf9 cells and (ii)that it is inefficient. One possible model of the endogenousprocess is that it involves import of CMP-sialic acid into theGolgi apparatus by an insect CSAT with a low substrate affin-ity, which therefore requires a higher sialic acid precursor con-centration to generate adequate intra-organelle CMP-sialicacid concentrations. In this model, the hCSAT, with its higher

Fig. 6. N-glycan profiles of hEPO-His produced under various conditions. hEPO-His preparations were isolated from SfSWT-4 and SfSWT-6 cells cultured inserum-free media containing 1 or 50 mM N-acetylmannosamine, as described in the legend to Figure 5, and samples were used for MALDI-TOF MS analysis ofenzymatically released, permethylated N-glycans, as described in Materials and methods. The figure shows the N-glycan profiles obtained with hEPO-His fromSfSWT-4 cells grown in 1 mM (A) or 50 mM (B) and SfSWT-6 cells grown in 1 mM (C) or 50 mM (D) N-acetylmannosamine. The insets are magnifications ofthe sialylated N-glycan profiles observed from m/z 1900–2500. All molecular ions were detected as [M +Na]+ and assigned and annotated accordingly using thestandard cartoon symbolic representations.

Human CMP-sialic acid transporter impact on insect sialylation

205

at University of W

yoming L

ibraries on Decem

ber 27, 2012http://glycob.oxfordjournals.org/

Dow

nloaded from

substrate affinity, would amplify CMP-sialic acid import inthe presence of the relatively low CMP-sialic acid concentra-tions produced when the cells are cultured in low concentra-tions of N-acetylmannosamine. Another possible model of theendogenous process is that it actually involves importingsialic acid, rather than CMP-sialic acid, into the Golgi appar-atus. This would require a sialic acid, rather than aCMP-sialic acid transporter, as previously suggested by Koleset al. (2009), and would be followed by the conversion ofsialic acid into CMP-sialic acid within the Golgi compartmentby an endogenous, Golgi-localized CMAS, as previously sug-gested by Viswanathan et al. (2006). In this model, hCSATwould amplify the levels of CMP-sialic acid available in theGolgi apparatus by importing the CMP-sialic acid producedby the mCMAS expressed in glycoengineered (SfSWT-6)insect cells. At this time, it is not possible to distinguishbetween these two models of the endogenous mechanism forCMP-sialic acid accumulation in the Golgi apparatus of Sf9cells. The first model lacks experimental support for the ideathat any insect actually encodes a CSAT. In fact, when theDrosophila ortholog of the human CSAT gene was expressedand the gene product was functionally characterized, theresults showed that it was a UDP-galactose/UDP-N-acetylga-lactosamine transporter (Aumiller and Jarvis 2002; Segawaet al. 2002). The second model lacks experimental support forthe presence of a Golgi sialic acid transporter in any systemand is weakened by the inconclusive nature of published datasuggesting that the Drosophila CMAS gene product is loca-lized in the Golgi apparatus (Viswanathan et al. 2006). Futurestudies will be required to address the question of the en-dogenous CMP-sialic acid accumulation pathway. For now,the addition of an hCSAT gene will provide an effective wayof improving glycoprotein sialylation and reducing the cost ofsialoglycoprotein production using glycoengineered insect cellsystems.

Materials and methodsPlasmid constructionsA cDNA encoding the hCSAT transcript variant 2 (accessionno. NM_001168398) was purchased from OriGeneTechnologies (Rockville, MD). This cDNAwas used as a tem-plate to amplify the hCSAT open reading frame by PCR withthe forward primer 5′-AGATCTATGGCTGCCCCGAGAGAC-3′, the reverse primer, 5′-AGGCCTTCACACACCAATAACTCTCTCC-3′ and Phusion High-Fidelity DNA Polymerase(New England BioLabs, Ipswich, MA). The amplified DNAfragment was cloned into pCR®4TOPO® (Life Technologies,Gaithersburg, MD), an error-free clone was identified by re-striction mapping and sequencing and then the fragment wassubcloned into the NruI and BglII sites of pIE1HR3 (Jarviset al. 1996) to create pIE1-hCSAT. pIE1-Hygro, pIE1-Neo andthe three piggyBac vectors used in this study, pXLBacII-GnTII/GalT-DsRed1-LTR, pXLBacII-ST6.1/ST3.3-ECFP-LTRand pXLBacII-SAS/CMPSAS-EYFP-LTR, have beendescribed previously (Jarvis et al. 1990; Hollister and Jarvis2001; Shi et al. 2007).

Cells and virusesSf9, SfSWT-4 and SfSWT-6 cells were routinely maintained asshake-flask cultures in ESF 921 medium (Expression Systems,Woodland, CA) at 28°C. SfSWT-4 is a new cell line that wasisolated for this study by transforming Sf9 cells withpXLBacII-GnTII/GalT-DsRed1-LTR, pXLBacII-ST6.1/ST3.3-ECFP-LTR, pXLBacII-SAS/CMPSAS-EYFP-LTR andpIE1-Hygro, using previously described transfection and selec-tion methods (Jarvis and Guarino 1995; Harrison and Jarvis2007a, b). After selection, single-cell clones were isolated bylimiting dilution and screened by cell surface staining withSNA, as described below. SNA-positive clones were furtherscreened by RNA dot-blot hybridization using probes specific

Fig. 7. Impact of hCSAT expression on hEPO sialylation. The bar graph shows the proportions of galactosylated (A) and sialylated (B) N-glycans found on thehEPO preparations isolated from SfSWT-4 and SfSWT-6 cells cultured in serum-free media containing 1 or 50 mM N-acetylmannosamine, as calculated from theN-glycan profiles shown in Figure 6.

H Mabashi-Asazuma et al.

206

at University of W

yoming L

ibraries on Decem

ber 27, 2012http://glycob.oxfordjournals.org/

Dow

nloaded from

to each mammalian transgene, as described previously (Jarviset al. 1996), and a clone expressing all the six transgenes wasdesignated SfSWT-4 and used for the remainder of this study.SfSWT-6 is another new cell line isolated for this study bysuper-transforming SfSWT-4 cells with pIE1-hCSAT andpIE1-Neo. After selection in medium containing Geneticin®(Life Technologies), single-cell clones were once again iso-lated by limiting dilution and screened by RNA dot-blot hy-bridization with an hCSAT probe and a positive clone wasdesignated SfSWT-6 and used for the remainder of this study.CHO and Lec2 (Deutscher et al. 1984) cells were obtainedfrom the American Type Culture Collection and maintained inα-MEM (Life Technologies) containing 10% fetal bovineserum (Thermo Scientific, Logan, UT) at 37°C with 5% CO2.The E2 strain of Autographa californica multicapsid nucleo-

polyhedrovirus (AcMNPV; Smith and Summers 1978) wasused as the wild-type baculovirus. A new recombinant baculo-virus expression vector encoding a C-terminally His-taggedversion of human erythropoietin (AchEPO-His) was isolatedfor this study. Briefly, an hEPO cDNA (Accession No.BC093628) was obtained from Life Technologies and used to-gether with forward (5′-CACCATGGGGGTGCACGAATGTCC-3′) and reverse (5′-TCTGTCCCCTGTCCTGCAGG-3′)primers to amplify the open reading frame without a stopcodon. The product was cloned into pENTR™/D-TOPO®(Life Technologies) and an error-free clone was identified byrestriction mapping and sequencing and designated pENTR™/D-TOPO®-hEPO(-stop). The hEPO sequence was then trans-ferred from pENTR™/D-TOPO®-hEPO(-stop) to a baculo-virus destination vector (BaculoDirect™ C-term linear DNA,Life Technologies) in a Gateway™ LR reaction. The spent LRreaction was used to transfect Sf9 cells with Cellfectin® (LifeTechnologies) and, after being cultured for 120 h in mediumsupplemented with gancyclovir to select against the parentalvirus, the transfected cell-free medium was harvested andprogeny baculoviruses were plaque-purified. Recombinantviral clones were presumptively identified by their whiteplaque phenotypes, several were tested for hEPO-His expres-sion by western blotting, and a positive clone was designatedAchEPO-His, amplified, titered and used for the remainder ofthis study.

RT-PCR assaysTotal RNA was extracted from 1 × 106 Sf9, SfSWT-4 orSfSWT-6 cells using the TRI Reagent (Life Technologies)according to the manufacturer’s instructions. The RNA pre-parations were then treated with DNaseI Amplification Grade(Life Technologies) and 3 μg of each was reverse transcribed at50°C for 90 min with ThermoScript™ Reverse Transcriptase(Life Technologies) and oligo(dT)31-VN (5′-TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTVN-3′) as the primer. The result-ing cDNA preparations were treated with RNase-H (LifeTechnologies) and then used for PCRs with Crimson TaqDNA Polymerase (New England BioLabs). The PCR condi-tions included an initial denaturation step at 95°C for 30 s, fol-lowed by 32 cycles of denaturation at 95°C for 15 s, annealingat 50°C for 20 s and extension at 68°C for 30 s, except that theannealing temperatures used for mST3GAL3 and SfRPL3

were 52 and 56°C, respectively. The sequences of all theprimers used for the RT-PCR assays are given in Table I.

SNA cell surface staining assaySf9, SfSWT-4 or SfSWT-6 cells were seeded into 6-wellplates at a density of 1 × 106 cells per well and allowed toadhere for 1 h. The media were drained, wild-type baculoviruswas added at a multiplicity of infection of 2 plaque formingunits per cell and the virus was allowed to adsorb for 1 h. Theinocula were then removed and the infected cells were fedwith ESF 921 supplemented with 10 mM N-acetylmannosa-mine (New Zealand Pharmaceuticals, Palmerston North, NewZealand) and incubated for 36 h after infection at 28°C. Atthat time, the growth media were removed and the cells wererinsed with lectin buffer (10 mM HEPES, pH 7.5, containing150 mM NaCl, 1 mM CaCl2, 1 mM MgCl2, 1 mM MnCl2,and 0.08% NaN3) and incubated with fresh lectin buffer at4°C for 5 min. The buffer was then removed and biotinylatedSNA (Vector Laboratories, Burlingame, CA) was then addedto the cells at a concentration of 10 μg/mL in lectin buffer.After another 5 min at 4°C, the lectin was removed and thecells were washed once with lectin buffer and incubated at4°C for 5 min with fresh lectin buffer containing 5 μg/mL ofTexas Red-Streptavidin (Vector Laboratories). Finally, thecells were washed twice with lectin buffer and examinedunder an Olympus FSX100 microscope (Tokyo, Japan).ImageJ software (Abramoff et al. 2004) was used to measurethe average staining intensities of all cells observed in threeindependent micrographs. The numbers of cells in eachmicrograph were counted manually and the data were used tocalculate the average staining intensity/cell in three independ-ent micrographs, with averages presented as “Relative fluores-cence” in the bar graphs shown in Figures 3 and 4.

CMP-sialic acid transport assaysThe CMP-sialic acid transport assays used in this study havebeen described previously (Aumiller and Jarvis 2002). Briefly,microsomal fractions were isolated from SfSWT-4, SfSWT-6,CHO and Lec 2 cells and total protein content was deter-mined using a commercial BCA assay kit (Thermo Scientific

Table I. Primer sequences used for RT–PCR assays

Primer name Sequence (5′–3′)

hMGAT2-Fw GGTGCATCAATGCTGAGThMGAT2-Rv TGCATACCACAGTCTCCAbB4GalT1-Fw TGAGTTTAACATACCTGTGGACbB4GalT1-Rv CCCCAGTAGTTATTAGGAAATCrST6GAL1-Fw TTCCAATCCTCAGTTACCACrST6GAL1-Rv CCATTAAACCTCAGAACTGCmST3GAL3-Fw GTATTATTCTGCCTGGAAGCmST3GAL3-Rv TCTCAAAGCCCTTCACAGmSAS-Fw GGATGCAGTCAATGGACAmSAS-Rv GCTAGGTTGAAGATGTCTTCmCMAS-Fw CGACAAGACTGGGATGGAmCMAS-Rv GACACTTCATTGCCGAGAhCSAT-Fw ACTCTACTTTTCAACGACAGChCSAT-Rv CTGAGCACAATACAGCAATAGSfRPL3-Fw ACATCGAAACTCCTCATGGTCTSfRPL3-Rv TCTTGATAACCTTGCCATCCTT

Human CMP-sialic acid transporter impact on insect sialylation

207

at University of W

yoming L

ibraries on Decem

ber 27, 2012http://glycob.oxfordjournals.org/

Dow

nloaded from

Pierce, Rockford, IL). Samples of each microsome preparationcontaining 100 μg of total protein were then incubated with0.03 μCi of CMP-sialic acid ([Sialic-6-3H]; AmericanRadiolabeled Chemicals, St. Louis, MO) at 37°C for 10 min.After the incubation period, the reaction mixtures were filteredthrough Type HA 0.45 μm nitrocellulose disks (Millipore,Bedford, MA) and the disks were washed 3× with ice-coldphosphate buffered saline (10 mM Na3PO4, pH 7.4, 0.14 MNaCl), dried, placed in scintillation vials containing 7 mL ofUltima Gold F liquid scintillation cocktail (PackardInstrument Company, Meriden, CT) and counted in a ModelLS-6500 liquid scintillation spectrometer (Beckman CoulterInstruments, Fullerton, CA).

Expression and purification of recombinant hEPO-HisSf9, SfSWT-4 and SfSWT-6 cells were seeded into 50 mLshake flask cultures in ESF 921 media at a density of 2 × 106

cells/mL, and then infected with AchEPO-His at a multiplicityof about 5 plaque forming units/cell. The virus was allowedto adsorb for 1 h, then the infected cells were gently pelleted,resuspended in 50 mL of ESF 921 containing 1, 10 or 50 mMN-acetylmannosamine, returned to shake flasks and incubatedat 28°C to 48 h after infection. The cells and debris were pel-leted by centrifugation at 1000 × g at 4°C for 10 min, thesupernatant was harvested and the budded virus particles wereremoved by centrifugation at 70,000 × g at 4°C for 30 min.One Complete Protease Inhibitor Cocktail tablet (RocheDiagnostics, Indianapolis, IN) was added to this final super-natant and it was dialyzed against 0.1 M NaCl for 6 h andthen against 0.5 M NaCl for another 6 h in dialysis mem-branes with a 12–14,000 molecular weight cut-off (SpectrumLabs, Rancho-Dominguez, CA). hEPO-His was affinity puri-fied from the dialysates using ProBond nickel affinity resin(Life Technologies) according to the manufacturer’s instruc-tions and finally desalted using a PD10 desalting column (GEHealthcare, Piscataway, NJ) equilibrated with 10 mM Tris–HCl, pH 7.5.

Lectin blotting analysis of hEPO-His sialylationPurified hEPO-His and bovine fetuin (Sigma-Aldrich,St. Louis, MO) were treated with neuraminidase in neuramin-idase reaction buffer or neuraminidase reaction buffer aloneaccording to the manufacturer’s instructions (New EnglandBioLabs). The proteins were then resolved by sodium dodecylsulfate-polyacrylamide gel electrophoresis on 12.5% poly-acrylamide gels and electrophoretically transferred toImmobilon-P membranes (Millipore), which were thenblocked for 1 h at room temperature with 5% bovine serumalbumin (Sigma-Aldrich) in Tris-buffered saline (150 mMNaCl in 50 mM Tris–HCl, pH. 7.5) containing 0.5% Tween20 for immunoblotting or in Tris-buffered saline containing1% Tween 20 for lectin blotting assays. The immunoblottingassays were completed using rabbit anti-hEPO (U-CyTech,Utrecht, The Netherlands) as the primary antibody, goat anti-rabbit IgG conjugated to alkaline phosphatase(Sigma-Aldrich) as the secondary antibody and a standardchromogenic assay for alkaline phosphatase activity (Blakeet al. 1984). The lectin blotting assays were completed usingbiotinylated SNA (Vector Laboratories) as the primary probe

for terminal sialic acids, streptavidin-alkaline phosphatase(Sigma-Aldrich) as the secondary reagent and the samechromogenic assay for alkaline phosphatase activity, asdescribed previously (Geisler and Jarvis 2011).

Chemical analysis of hEPO-His sialylationPurified hEPO-His was treated with neuraminidase accordingto the manufacturer’s recommendations (New EnglandBioLabs) overnight at 37°C and the released sialic acid wasderivatized with 12 mM malononitrile in 0.15 M boratebuffer, pH 9.5, at 80°C for 5 min, as described previously(Honda et al. 1987; Markely et al. 2010). The derivatizedsialic acids were then quantified in a Wallac Model 1420fluorescent plate reader (Turku, Finland) using an excitationwavelength of 355 nm and an emission wavelength of 460nm. The resulting relative fluorescence values were normal-ized on a total protein basis, as determined with a commercialBCA assay kit (Thermo Scientific Pierce).

Mass spectrometryN-glycans were enzymatically released from the varioushEPO preparations produced and purified under the conditionsdescribed above by exhaustive digestion with PNGase-F (NewEngland Biolabs). The spent reactions were applied to pre-conditioned C18 SepPak cartridges (Waters Corp., Milford,MA) and the flow-through and a 5% (v/v) aqueous acetic acidwash were pooled, evaporated and permethylated, asdescribed previously (Dell et al. 1994). The permethylatedN-glycan derivatives extracted into chloroform with severalaqueous washes were re-evaporated, then resuspended inacetonitrile and mixed 1:1 with 2,5-dihydroxybenzoic acidmatrix (10 mg/mL in 50% acetonitrile in water), and thesamples were spotted onto the MALDI-TOF target plate. Dataacquisition was performed manually on a Model 4700Proteomics Analyzer equipped with an Nd:YAG laser(Applied Biosystems, Framingham, MA) and 1000 shots wereaccumulated in the reflectron positive ion mode.

Funding

This work was supported by Award Number R01GM49734from the National Institute of General Medical Sciences. Thecontent is solely the responsibility of the authors and does notnecessarily represent the official views of the NationalInstitute of General Medical Sciences or the NationalInstitutes of Health. The MS data were acquired at the CoreFacilities for Protein Structural Analysis at Academia Sinica,supported under the Taiwan National Core Facility Programfor Biotechnology, NSC Grant Number 100-2325-B-001-029.

Acknowledgements

The authors thank Jared Aumiller for isolating the recombin-ant baculovirus encoding human erythropoietin, which wasused in this study.

H Mabashi-Asazuma et al.

208

at University of W

yoming L

ibraries on Decem

ber 27, 2012http://glycob.oxfordjournals.org/

Dow

nloaded from

Conflict of interest

Donald L. Jarvis is the founder of a new biotech company,GlycoBac, LLC, that will focus on the glycoengineered insectcell platform as a tool for recombinant glycoprotein produc-tion. The other authors have no potential conflicts of interest.

Abbreviations

bB4GALT1, bovine β1,4-galactosyltransferase I; CHO,Chinese hamster ovary; CSAT, CMP-sialic acid transporter;hCSAT, human CMP-sialic acid transporter; hEPO, humanerythropoietin; hMGAT2, human β1,2-N-acetylglucosaminyl-transferase II; mCMAS, murine CMP-sialic acid synthetase;mSAS, murine sialic acid synthase; mST3GAL3, murineα2,3-sialyltransferase III; rST6GAL1, rat α2,6-sialyltransferaseI; RT–PCR, reverse transcription polymerase chain reaction;SNA, Sambucus nigra agglutinin.

References

Abramoff MD, Magalhaes PJ, Ram SJ. 2004. Image processing with ImageJ.Biophot Intl. 11:36–42.

Altmann F, Schwihla H, Staudacher E, Glossl J, Marz L. 1995. Insect cellscontain an unusual, membrane-bound β-N-acetylglucosaminidase probablyinvolved in the processing of protein N-glycans. J Biol Chem.270:17344–17349.

Altmann F, Staudacher E, Wilson IB, Marz L. 1999. Insect cells as hosts forthe expression of recombinant glycoproteins. Glycoconj J. 16:109–123.

Aumiller JJ, Hollister JR, Jarvis DL. 2003. A transgenic lepidopteran insectcell line engineered to produce CMP-sialic acid and sialoglycoproteins.Glycobiology. 13:497–507.

Aumiller JJ, Jarvis DL. 2002. Expression and functional characterization of anucleotide sugar transporter from Drosophila melanogaster: Relevance toprotein glycosylation in insect cell expression systems. Prot Expr Purif.26:438–448.

Aumiller JJ, Mabashi-Asazuma H, Hillar A, Shi X, Jarvis DL. 2012. A newglycoengineered insect cell line with an inducibly-mammalianized proteinN-glycosylation pathway. Glycobiology. 22:417–428.

Blake MS, Johnston KH, Russell-Jones GJ, Gotschlich EC. 1984. A rapid,sensitive method for detection of alkaline phosphatase conjugated anti-antibody on western blot. Analyt Biochem. 36:175–179.

Breitbach K, Jarvis DL. 2001. Improved glycosylation of a foreign protein byTn-5B1–4 cells engineered to express mammalian glycosyltransferases.Biotechnol Bioengr. 74:230–239.

Butters TD, Hughes RC, Vischer P. 1981. Steps in the biosynthesis of mos-quito cell membrane glycoproteins and the effects of tunicamycin. BiochimBiophys Acta. 640:672–686.

Byrne B, Donohoe GG, O’Kennedy R. 2007. Sialic acids: Carbohydrate moi-eties that influence the biological and physical properties of biopharmaceut-ical proteins and living cells. Drug Discov Today. 12:319–326.

Dell A, Reason AJ, Khoo KH, Panico M, McDowell RA, Morris HR. 1994.Mass spectrometry of carbohydrate-containing biopolymers. MethEnzymol. 230:108–132.

Deutscher SL, Nuwayhid N, Stanley P, Briles EI, Hirschberg CB. 1984.Translocation across Golgi vesicle membranes: A CHO glycosylationmutant deficient in CMP-sialic acid transport. Cell. 39:295–299.

Durocher Y, Butler M. 2009. Expression systems for therapeutic glycoproteinproduction. Curr Op Biotechnol. 20:700–707.

Eckhardt M, Muhlenhoff M, Bethe A, Gerardy-Schahn R. 1996. Expressioncloning of the Golgi CMP-sialic acid transporter. Proc Natl Acad Sci USA.93:7572–7576.

Geisler C, Aumiller JJ, Jarvis DL. 2008. A fused lobes gene encodes the pro-cessing beta-N-acetylglucosaminidase in Sf9 cells. J Biol Chem.283:11330–11339.

Geisler C, Jarvis DL 2009. Insect cell glycosylation patterns in the context ofbiopharmaceuticals. In: Walsh B, editor. Post-translational Modification ofProtein Biopharmaceuticals. Weinheim: Wiley-Blackwell. p. 165–191.

Geisler C, Jarvis DL. 2010. Identification of genes encoding N-glycan pro-cessing beta-N-acetylglucosaminidases in Trichoplusia ni and Bombyxmori: Implications for glycoengineering of baculovirus expression systems.Biotechnol Prog. 26:34–44.

Geisler C, Jarvis DL. 2011. Effective glycoanalysis with Maackia amurensislectins requires a clear understanding of their binding specificities.Glycobiology. 21:988–993.

Geisler C, Jarvis DL. 2012. Substrate specificities and intracellular distribu-tions of three N-glycan processing enzymes functioning at a key branchpoint in the insect N-glycosylation pathway. J Biol Chem. 287:7084–7097.

Harrison RL, Jarvis DL. 2006. Protein N-glycosylation in thebaculovirus-insect cell expression system and engineering of insect cells toproduce “mammalianized” recombinant glycoproteins. Adv Virus Res.68:159–191.

Harrison RL, Jarvis DL. 2007a. Transforming lepidopteran insect cells forcontinuous recombinant protein expression. Meth Mol Biol. 388:299–316.

Harrison RL, Jarvis DL. 2007b. Transforming lepidopteran insect cells forimproved protein processing. Meth Mol Biol. 388:341–356.

Hollister JR, Conradt HO, Jarvis DL. 2003. Evidence for a sialic acid salva-ging pathway in lepidopteran insect cell lines. Glycobiology. 13:487–495.

Hollister JR, Grabenhorst E, Nimtz M, Conradt HO, Jarvis DL. 2002.Engineering the protein N-glycosylation pathway in insect cells for produc-tion of biantennary, complex N-glycans. Biochemistry. 41:15093–15104.

Hollister J, Jarvis DL. 2001. Engineering lepidopteran insect cells for sialo-glycoprotein production by genetic transformation with mammalianß1,4-galactosyltransferase and a2,6-sialyltransferase genes. Glycobiology.11:1–9.

Honda S, Iwase S, Suzuki S, Kakehi K. 1987. Fluorometric determination ofsialic acids using malononitrile in weakly alkaline media and its applica-tion to postcolumn labeling in high-performance liquid chromatography.Anal Biochem. 160:455–461.

Hooker AD, Green NH, Baines AJ, Bull AT, Jenkins N, Strange PG, JamesDC. 1999. Constraints on the transport and glycosylation of recombinantIFN-gamma in Chinese hamster ovary and insect cells. Biotechnol Bioengr.63:559–572.

Jarvis DL. 1993. Effects of baculovirus infection on IE1-mediatedforeign gene expression in stably transformed insect cells. J Virol. 67:2583–2591.

Jarvis DL, Fleming JA, Kovacs GR, Summers MD, Guarino LA. 1990. Useof early baculovirus promoters for continuous expression and efficient pro-cessing of foreign gene products in stably transformed lepidopteran cells.Nat Biotechnol. 8:950–955.

Jarvis DL, Guarino LA. 1995. Continuous foreign gene expression in trans-formed lepidopteran insect cells. Meth Mol Biol. 39:187–202.

Jarvis DL, Weinkauf C, Guarino LA. 1996. Immediate-early baculovirusvectors for foreign gene expression in transformed or infected insect cells.Prot Expr Purif. 8:191–203.

Koles K, Repnikova E, Pavlova G, Korochkin LI, Panin VM. 2009.Sialylation in protostomes: a perspective from Drosophila genetics and bio-chemistry. Glycoconj J. 26:313–324.

Markely LR, Ong BT, Hoi KM, Teo G, Lu MY, Wang DI. 2010. A high-throughput method for quantification of glycoprotein sialylation. AnalytBiochem. 407:128–133.

Marz L, Altmann F, Staudacher E, Kubelka V. 1995. Protein glycosylation ininsects. In: Montreuil J, Vliegenthart JFG, et al. editors. Glycoproteins.Amsterdam: Elsevier. p. 543–563.

Segawa H, Kawakita M, Ishida N. 2002. Human and DrosophilaUDP-galactose transporters transport UDP-N-acetylgalactosamine in add-ition to UDP-galactose. Eur J Biochem. 269:128–138.

Shi X, Harrison RL, Hollister JR, Mohammed A, Fraser MJ, Jr, Jarvis DL.2007. Construction and characterization of new piggyBac vectors for con-stitutive or inducible expression of heterologous gene pairs and the identifi-cation of a previously unrecognized activator sequence in piggyBac. BMCBiotechnol. 7:5.

Smith GE, Summers MD. 1978. Analysis of baculovirus genomes with re-striction endonucleases. Virology. 89:517–527.

Sola RJ, Griebenow K. 2011. Glycosylation of therapeutic proteins: An effect-ive strategy to optimize efficacy. BioDrugs. 24:9–21.

Son YD, Jeong YT, Park SY, Kim JH. 2011. Enhanced sialylation of recom-binant human erythropoietin in Chinese hamster ovary cells by combinator-ial engineering of selected genes. Glycobiology. 21:1019–1028.

Stollar V, Stollar BD, Koo R, Harrap KA, Schlesinger RW. 1976. Sialicacid contents of sindbis virus from vertebrate and mosquito cells.

Human CMP-sialic acid transporter impact on insect sialylation

209

at University of W

yoming L

ibraries on Decem

ber 27, 2012http://glycob.oxfordjournals.org/

Dow

nloaded from

Equivalence of biological and immunological viral properties. Virology.69:104–115.

Summers MD, Smith GE. 1987. A manual of methods for baculovirusvectors and insect cell culture procedures. Tx Ag Expt Stn Bull No 1555.

Tomiya N, Ailor E, Lawrence SM, Betenbaugh MJ, Lee YC. 2001. Determinationof nucleotides and sugar nucleotides involved in protein glycosylation byhigh-performance anion-exchange chromatography: Sugar nucleotide contentsin cultured insect cells and mammalian cells. Anal Biochem. 293:129–137.

Viswanathan K, Tomiya N, Singh S, Park J, Lee YC, Palter K, BetenbaughMJ. 2006. Expression of a functional Drosophila melanogaster CMP-sialic

acid synthetase: Differential localization of the Drosophila and humanenzymes. J Biol Chem. 281:15929–15940.

Wickham TJ, Davis T, Granados RR, Shuler ML, Wood HA. 1992. Screeningof insect cell lines for the production of recombinant proteins and infec-tious virus in the baculovirus expression system. Biotechnol Progr.8:391–396.

Yun EY, Goo TW, Kim SW, Choi KH, Hwang JS, Kang SW, Kwon OY.2005. Galactosylation and sialylation of mammalian glycoproteins pro-duced by baculovirus-mediated gene expression in insect cells. BiotechnolLett. 27:1035–1039.

H Mabashi-Asazuma et al.

210

at University of W

yoming L

ibraries on Decem

ber 27, 2012http://glycob.oxfordjournals.org/

Dow

nloaded from