Implications of Vector Fragmentation and InitialProduction Levels in Production Cell LineDevelopment Using MTX

Vector Fragmentation and Initial Production Levelsin Cell Line Development

Say Kong Ng, Wenyu Lin, Rohit Sachdeva, Daniel I.C. Wang,and Miranda G.S. Yap

Abstract The Chinese Hamster Ovary (CHO) production cell line developmentprocess using methotrexate (MTX) amplification is well-studied and commonlyused for biopharmaceutical processes. However, successful MTX amplificationvaries from clone to clone and suggested reasons include vector fragmentationduring the transfection process and genomic rearrangement of the CHO chromo-somes. Here, we elucidated the vector integration patterns of 40 transfected singlecell clones by Southern blotting and showed that vector fragmentation occurs at asignificant level in our experiment. This concurs with MTX amplification studiesimplying that single cell cloning is necessary to ensure a successful amplificationprocess. The single cell clones were then subjected to MTX amplification, duringwhich their fluorescence levels were tracked using FACS. We showed that expres-sion levels upon amplification do not correlate to initial expression levels of theclones, and amplified high producing cells can be derived from initial low produc-ers. This raises the question of whether it is optimal to choose initial high producingclones for MTX amplification.

1 Materials and Methods

1.1 Vector Construction and Preparation

A methotrexate (MTX) amplifiable enhanced green fluorescent protein (EGFP)reporter vector was constructed previously (Ng et al. 2010). The vector contains

S.K. Ng (B)Bioprocessing Technology Institute, Agency for Science, Technology and Research(A∗STAR), 20 Biopolis Way, #06-01, Centros, Singapore 138668e-mail: ng_say_kong@bti.a-star.edu.sg

77N. Jenkins et al. (eds.), Proceedings of the 21st Annual Meeting of the European Societyfor Animal Cell Technology (ESACT), Dublin, Ireland, June 7–10, 2009, ESACT Proceedings 5,DOI 10.1007/978-94-007-0884-6_14, C© Springer Science+Business Media B.V. 2012

78 S.K. Ng et al.

Fig. 1 Constructed EGFP-DHFR reporter vector in circular and linearized forms. Vector mapsof the vector with restriction enzyme sites used for vector integration site analysis by Southernblotting are illustrated here. Agarose gel electrophoresis data of the linearized and circular vectoris inserted to show the quality of the vector used for transfection. Reproduced with permission ofJohn Wiley & Sons Inc. from Ng et al. (2010)

two BamHI restriction sites that flank the EGFP and DHFR genes, and a BglII sitebetween the two genes (Fig. 1) to facilitate Southern blot analysis of the vector inte-gration site. The vector was linearized with PvuI (New England Biolabs, Ipswich,MA), purified and characterized using agarose gel electrophoresis (Fig. 1).

1.2 Isolation and Manipulation of Single Cell Clones

CHO-DG44 cells (Urlaub et al. 1983) (Invitrogen, GibcoTM Catalog number 12609-012) were transfected with the EGFP reporter vector using NucleofectorTM kit V(Amaxa, Gaithersburg, MD). Single cell clones of the transfected cell pools werethen obtained as described previously (Ng et al. 2010). When the clones’ cellviabilities were greater than 90%, they were harvested for FACS analysis and stor-age in a −80◦C freezer for Southern blotting. The clones were then amplified bysequentially adapting the cells from 10 to 50 to 250 nM MTX (Sigma, St. Louis,MO). The amplified clones were also analyzed using FACS at each stage of theprocess.

Implications of Vector Fragmentation and Initial Production Levels in Production . . . 79

1.3 Analysis of Vector Integrity by Southern Blotting

For Southern blotting, genomic DNA was first extracted from 107 cells using agenomic DNA purification kit (Gentra Puregene, USA). 10 μg of the genomic DNAwas then restriction digested with BamHI or BglII, and resolved on a 0.7% agarosegel. The DNA was transferred to a positively-charged nylon membrane (Roche,Germany), and hybridization and detection were performed using the DIG HighPrime DNA Labeling and Detection Starter Kit II (Roche, Germany). PCR probesfor DHFR (360 bp) and EGFP (510 bp) (Ng et al. 2010) were used in sequen-tial hybridization on the same membrane to find overlapping bands. Sizes of bandsobtained were estimated with respect to DNA ladders (DIG-labeled DNA MolecularWeight Marker III, Roche, Germany) that were on the same blot.

1.4 Analysis of EGFP Expression Using FACS

GFP fluorescence level of 20,000 cells was measured using fluorescence-activatedcell sorting (Becton Dickinson FACSCalibur, USA). A FL1 setting of 180 V on alog scale was used to quantify the EGFP fluorescence level, which was taken as thegeometric mean fluorescence intensities of the cells. At this setting, the geometricmean fluorescence intensities of non-fluorescing cells will be 1.0, occupying thelowest channel of the distribution.

2 Results and Discussions

2.1 Isolation of Single Cell Clones

In order to study vector integrity after transfection, different single cell clones trans-fected with a reporter vector are essential. EGFP was chosen as the reporter proteinto allow the use of FACS for single cell sorting according to expression levels togenerate these clones. Forty one EGFP-positive single cell clones were thus isolatedfrom CHO-DG44 cell transfected with the EGFP reporter vectors. Thirteen clonesfrom Set 1 are sorted from selected high producing cells, while 25 clones from Set 2and 3 clones from Set 3 are sorted for high and low fluorescence levels respectively,from transiently transfected cells that have yet to undergo the selection process.

2.2 Analysis of Vector Integrity in Single Cell Clones

The isolated clones were first characterized by Southern blotting. Band sizes of thedifferently digested and probed genomic DNA, and bands containing one or bothof the probed regions were determined. Scaled models of these DNA bands wereconstructed and assembled to give the proposed integration pattern that accounts forthe presence of all probed regions in all the bands. These data are shown in Table 1.

80 S.K. Ng et al.Ta

ble

1Su

mm

ary

ofth

eba

ndsi

zes

dete

cted

usin

gSo

uthe

rnbl

ottin

gan

dth

epr

opos

edin

tegr

atio

npa

ttern

sof

the

fluor

esce

ntsi

ngle

cell

clon

es.S

ingl

ece

llcl

ones

are

num

bere

dby

“Set

num

ber–

Clo

nenu

mbe

r”.D

NA

band

sde

tect

edar

ecl

assi

fied

first

with

resp

ectt

oth

ere

stri

ctio

nen

zym

eus

edto

dige

stth

ege

nom

icD

NA

,i.e

.Bam

HI

orB

glII

,fol

low

edby

the

prob

esw

hich

the

band

hybr

idiz

edto

,i.e

.EG

FP+

DH

FR,E

GFP

,or

DH

FR.A

blan

kin

the

tabl

ein

dica

tes

that

noba

ndof

this

clas

sific

atio

nw

asde

tect

edfo

rth

ecl

one.

Prop

osed

inte

grat

ion

patte

rns

wer

ede

rive

dby

asse

mbl

ing

the

dete

cted

band

s.T

hese

data

wer

ede

rive

dfr

omce

llscu

lture

din

–HT

sele

ctio

nm

ediu

m,u

nles

sot

herw

ise

indi

cate

d.T

hein

tegr

atio

npa

ttern

sof

sing

lece

llcl

ones

culti

vate

din

+H

Tan

d–H

Tm

edia

wer

ege

nera

llysi

mila

r.R

epro

duce

dw

ithpe

rmis

sion

ofJo

hnW

iley

&So

nsIn

c.fr

omN

get

al.(

2010

)

Ban

dsi

ze(k

bp)

Bam

HI

dige

stB

g/II

dige

st

Clo

neN

umbe

rE

GFP

+D

HFR

EG

FPD

HFR

EG

FP+

DH

FRE

GFP

DH

FRPr

opos

edIn

tegr

atio

nPa

ttern

1–1

3.1

5.3,

5.0,

4.5

3.1

6.7

1–2∗

3.2

4.2

6.0

6.8,

4.6

1–3#

3.3

4.7

6.5

7.1,

5.1

1–4

3.0

4.4

6.1

7.2,

4.8

1–5∗

3.1

4.6

6.3

7.4,

5.0

1–6

3.2

4.7

6.5

5.2

1–7∗

3.2

4.3

6.7

6.7,

4.7

1–8∗

3.2

4.8

6.4

7.1,

4.5

1–9

3.2

8.0,

4.8

3.9,

3.3

1–10

∗3.

24.

15.

86.

7,4.

51–

11∗

3.2

5.2,

5.0,

4.5

3.3

6.6

1–12

∗3.

24.

83.

75.

41–

133.

02.

05.

05.

62.

02–

1∗3.

7,3.

24.

39.

1,5.

03.

9,3.

62–

2∗3.

25.

16.

02–

3∗3.

26.

9,5.

14.

32–

43.

22.

55.

3,4.

86.

1,3.

44.

32–

5∗3.

25.

03.

42.

52–

6∗3.

38.

74.

57.

72–

7∗3.

113

.53.

96.

02–

8∗3.

24.

7,2.

67.

2,4.

12–

9∗3.

24.

2,2.

74.

5,3.

52–

10∗

3.0

2.8

3.5

Implications of Vector Fragmentation and Initial Production Levels in Production . . . 81Ta

ble

1(c

ontin

ued)

Ban

dsi

ze(k

bp)

Bam

HI

dige

stB

g/II

dige

st

Clo

neN

umbe

rE

GFP

+D

HFR

EG

FPD

HFR

EG

FP+

DH

FRE

GFP

DH

FRPr

opos

edIn

tegr

atio

nPa

ttern

2–11

3.1

29.0

5.1

2–12

5.3

3.4

2.2,

4.0

2.3

2–13

∗3.

23.

04.

32–

14∗

3.1

6.0

3.8

2–15

∗3.

26.

7,5.

45.

02–

16#

3.1

6.2

4.1

2–17

∗3.

24.

13.

52–

18∗

3.1

3.1

3.5

2–19

∗3.

24.

82.

92–

20∗

3.3

5.1

2.6

2–21

3.3

3.0,

2.9

4.4,

3.5

2–22

#3.

15.

22.

52–

23#

1.5

15.7

2–24

#2.

55.

02–

25#

9.4

7.7

2.7

3–1#

3.2

3.3

4.5

3–2

3.3

5.2

2.9

3–3

No

data

Leg

ends

+E

xtra

frag

men

tsw

ithon

lyE

GFP

gene

dete

cted

from

cell

pelle

tsde

rive

dfr

om+

HT

cultu

res

#Pr

opos

edin

tegr

atio

npa

ttern

from

cell

pelle

tsde

rive

dfr

om+

HT

cultu

res

∗Id

entic

alpr

opos

edin

tegr

atio

npa

ttern

for

cell

pelle

tsde

rive

dfr

om+

HT

and

–HT

cultu

res

Bg/

IIsi

te|B

amH

Isi

teE

GFP

prob

edre

gion

DH

FRpr

obed

regi

onE

xpec

ted

Bam

/HI

frag

men

tV

aria

ble

leng

thre

gion

82 S.K. Ng et al.

Two clones (2–23 and 2–24) are shown to have the EGFP but not the DHFR gene.In addition, fragmentation of the EGFP transgene from the DHFR transgene wasalso observed in 7 other fluorescent clones in Set 2 (2–4, 2–6, 2–7, 2–9, 2–12, 2–15and 2–17). This confirms our postulate that the DHFR transgene have fragmentedfrom the EGFP transgene in some of these cells. From Set 2, 9/25 (36%) of theEGFP-positive clones contained fragmented transgenes. This suggests a rather highfrequency of transgene fragmentation in our experiment. As we have verified thequality of our transfected vector, it is likely that the reporter vector was shearedduring the transfection procedure, or cut in vivo by endonucleases such as DNase I.

One implication of this observation is the presence of cells having only the DHFRselection marker: Although no such clone was isolated in our study, this is likelydue to our cloning strategy of isolating fluorescing clones by FACS. Hence, DHFR-only cells are likely to be present in the original transfected cell pool. As thesecells survive and grow better than cells that are expressing another transgene inMTX-containing amplification medium, they will likely overtake the cell populationto result in a loss in protein productivity during MTX amplification, as observedpreviously (Ng et al. 2007). This can also explain the clone dependent success ofMTX amplification in other studies (Kim et al. 2001; Kaufman et al. 1985; Fannet al. 2000), implying that single cell cloning may be necessary to ensure successfulMTX amplification, even when a bicistronic expression vector is used.

2.3 Fluorescence Levels of Clones During MTX Amplification

The single cell clones were subjected to MTX amplification and their fluorescencelevels during the process were studied using FACS. All Set 1 clones survived MTXamplification, while 6 of the 25 Set 2 clones (2–1 to 2–6) survived. We compared thefluorescence levels of the single cell clones between different stages of MTX ampli-fication to study the predictability of production levels after amplification (Fig. 2).From Fig. 2, the correlation of the clones’ fluorescence level between each step ofMTX amplification was gradually lost with each amplification stage, suggesting thatthe predictability of final protein expression level becomes increasingly unfeasible.In addition, we noted that initially low-producing clones can become high produc-ing cells at 250 nM MTX. This suggests that the selection of initial high producingclones for MTX amplification may not be optimal for high productivity.

Implications of Vector Fragmentation and Initial Production Levels in Production . . . 83

Fig. 2 Comparison between fluorescence levels of single cell clones at different stages of theMTX amplification process. The fluorescence levels from clones that survived MTX amplificationare plotted on the graphs as dots, while a best-fit straight line is drawn on each graph as a gauge tothe spread of the points. The various stages of MTX amplification (or the medium) are denoted onthe left and bottom of the figure

Acknowledgement This work was supported by the Biomedical Research Council of A∗STAR(Agency for Science, Technology and Research), Singapore.

References

Fann CH, Guirgis F, Chen G, Lao MS, Piret JM. Limitations to the amplification and stabil-ity of human tissue-type plasminogen activator expression by Chinese hamster ovary cells.Biotechnol. Bioeng. 2000; 69(2): 204–212.

Kaufman RJ, Wasley LC, Spiliotes AJ, Gossels SD, Latt SA, Larsen GR, Kay RM. Coamplificationand coexpression of human tissue-type plasminogen activator and murine dihydrofolatereductase sequences in Chinese hamster ovary cells. Mol. Cell. Biol. 1985; 5(7): 1750–1759.

Kim NS, Byun TH, Lee GM. Key determinants in the occurrence of clonal variation in humanizedantibody expression of CHO cells during dihydrofolate reductase mediated gene amplification.Biotechnol. Prog. 2001; 17: 69–75.

Ng SK, Lin W, Sachdeva R, Wang DI, Yap MG. Vector fragmentation: characterizing vectorintegrity in transfected clones by Southern blotting. Biotechnol. Prog. 2010; 26(1): 11–20.

Ng SK, Wang DI, Yap MG. Application of destabilizing sequences on selection marker forimproved recombinant protein productivity in CHO-DG44. Metab. Eng. 2007; 9(3): 304–316.

Urlaub G, Kas E, Carothers AM, Chasin LA. Deletion of the diploid dihydrofolate reductase locusfrom cultured mammalian cells. Cell. 1983; 33: 405–412.