Antibody analysis by microfluidic on a chip. Alessandro Mora, November 11th

, 2012

INTRODUCTION

Microfluidic based devices have been developed since early 90s although the technology reached its

popularity few years ago with application in the area of bioterapeutics monitoring and protein

characterization. This technology gradually replaced existing methods such as SDS-PAGE (sodium dodecyl

sulfate polyacrylamide gel electrophoresis) and Capillary Electrophoresis CE, being faster and increasing

limit of detection, accuracy and reproducibility of the experiment. The main component of a microfluidic

device is a microfabricated glass chip, where the protein assay is performed and a detector for LIF (Laser

induced fluorescence). The output is generally a software based electropherogram where further

investigation of the sample can be done. Another major advantage over existing methods is the high-

throughput capability and automation interface. Design of Experiment (DOE) studies can reveal the effect

of cell culture changes on homogeneity, purity, and post translational modifications of monoclonal

antibodies but produce large number of samples. This trend can exceed the capacity of modern analytical

laboratories which primarily depend on HPLC and CE based separations to monitor protein and bio-

therapeutic product quality. Microfludic on a chip offers new areas of development and bioterapeutic

studies. The leader in this field is Caliper Lifescience, recently

acquired by Perkin-Elmer. The most popular device using

microfluidic chip chemistry and technology is LabChip GXII,

which is a standard in most biomedical labs nowadays. This

report seeks to elucidate such technology, pointing out the bio-

mechanical engineering work behind it. A detailed evaluation of

existing literature will be addressed, focusing on the usage of

LabChip for Antibody process monitoring. Few examples (from

literature and recent personal experience) will be given and

areas of improvement evaluated. Figure 1. Caliper LabChip GXII device.

MATHERIALS AND METHODS

Monoclonal antibody products are inherently heterogeneous because of post-translation modification that

often occurs during the fermentation process. Therefore, thorough characterization of MAb products is

required in order to assess their quality attributes. In early stage process development, DOE or Factorial

Experimental Design (FED), experiments are performed to understand the relationships between the

process parameters and product quality in order to define a process design space around the variability in

the process parameters (Caliper Labchip GXII, Antibody analysis, white paper 2010) . The increased number

of variables to be examined in a DOE study of cell culture conditions or purification process produces large

number of samples that need to be processed.

Current method of monitoring protein product quality includes SDS-PAGE and HPLC. SDS-PAGE is both

labor intensive and difficult to automate and has been supplanted by CE-based separation in modern

analytical laboratories. CE and HPLC analysis require separation time of 15-60 minutes per sample, which

can limit DOE studies 2 . In the past several years, microfluidic-based assay for sizing, quantification, and

purity assessment of proteins are finding wide use because they address the limitations of SDS-PAGE, HPLC,

and CE methods 2,3 . Chen et. al. 2 studied the use of the LabChip technology for screening MAb

product quality attributes. They analyzed MAbs from crude CHO cell culture supernatants and purified

samples under reducing and non-reducing conditions. They showed similar resolution and sensitivity as

conventional CE-SDS on a 20 cm capillary but at time scale of approximately 70 times faster (41 sec versus

50 min per sample). The LabChip GXII (LCGXII) platform shown in Figure 1 provides an automated platform

for rapid analysis of proteins and nucleic acids. Sampling directly from a 96- or 384-well plate, the LCGXII

offers ease of use and walk away time. Each individual sample can be analyzed in 41 seconds or less,

enabling a 96-well plate to be analyzed in slightly longer than 1 hour.

The protein sizing on the microfluidic chip (see Figure 2) is achieved by integration of the main features of

one-dimensional SDS-PAGE: these include the separation, staining, destaining, and detection steps 4 .

Denatured proteins are brought onto the chip directly from a microtiter plate through a capillary sipper.

The samples are then electrokinetically loaded and injected into the 14mm long separation channel that

contains a low viscosity matrix of entangled polymer solution. The protein-SDS complex and free SDS

micelles in the sample plug are immediately stained by dye present in the sieving matrix. At the end of the

separation channel, the sample is diluted to reduce the SDS concentration below its critical micelle

concentration in order to reduce the background fluorescence so that protein-SDS-dye complexes can be

detected. The integrated microchip-based assay was used to separate and quantify monoclonal antibodies

and their fragments with minimal sample preparation artifacts, including crude samples.

Figure 2. Back side view of the chip Figure 3. Separation channels in the chip

Microchip design and fabrication. The chip were made from soda lime glass, with photolitopgraphy

defined wet-etched channels, fabricated using proprietary procedures (Bousse, Mouradian, et all. 2001).

The cannel width of the mask was 10 µm, which after the etch resulted in channels 13 µm deep and 36 µm

wide. A lid with 2 mm diameter holes drilled in it was thermally bonded to the lower plate to complete the

channels. After dicing, the size of each chip is ~ 17.5 mm square. Figure 2 shows the chip design with which

all the data presented here was obtained.

Sample Preparation. All reagents along with the chip used in this report come in a convenient kit

purchased at Caliper. Samples come from Abbott Bioresearch Center, Worcester, MA. One box includes the

following reagents: Protein Gel Matrix, Express Dye Solution, Protein Express Lower Marker, Protein

Express Sample Buffer, Protein Express Ladder, Protein Express Wash Buffer. Chip is good for up to 400

sample runs and has a lifetime of one month. Protein sample (coming from cell culture crude supernatant)

is incubated with Protein Express Sample Buffer, containing SDS. Concentration of SDS in the sample buffer

has been seen to dramatically affect the outcome of the assay (reasons will be explained in the next sub

chapter). A standard incubation time can be 75 degrees per 10 minute, though Antibodies can easily bear

100 degrees. Minor adjustments have to be made in order to prevent thiol formation which is a known

phenomena that cell culture supernatant can go into. Therefore samples have been treated with IAC (Iodo

Acetamide) that prevents thiol free S group to bind Lysin residues on the protein and spoil the size reading.

For reduced samples, DTT has been used as reducing agent. Samples are prepared on a 96 well format plate

and then the plate is spun down in order to precipitate eventual particulate or debris to the bottom and

prevent the sipper to clog.

Chip Preparation. Based on figure 3, two wells of the chip are loaded with the Gel-Dye solution made

fresh out of the Protein Gel Matrix and the Express Dye Solution. The destain solution is clean Protein Gel

Matrix and finally the marker is used as internal standard. Also the ladder and the wash buffer are loaded

outside of the chip and used to provide a molecular weight reference during the readings and to wash the

chip between a cycle and the next one.

Principle of operations. In the chip, each sample is sipped by negative pressure from a 96 well

format plate until a sufficient quantity is loaded in the chip. In the central channel the sample mixes up with

the Gel-Dye mixture and the Dye binds covalentley the SDS already bound to the protein in the sample. No

direct binding protein-dye appears to happen. Complete biding is delivered in 100 ms. A known drawback

is that dye can bind SDS micelles rather than protein sample and it depends upon CMC (critical micellar

concentration), this phenomenon is drastically reduced in the following destaining part. As the fragments

move down the central channel, they separate by size, under electro-osmotic flow (EOF) of 5-10 e6 cm2 /V

sec. Next the sample is carried away the separation channel by the action of the current. Next step is

separation in presence of SDS which occurs at field of 770 V/cm in the long 14 mm channel . High level of

SDS dilute the sample buffer and decrease the CMC, literally breaking all the micelles and releasing the dye

bound to them. This step decrease the background and is comparable to a destain procedure. Also the free

dye can bind the protein and provide a better resolution. Similar treatment happens to the ladder (in a

separate part of the instrument) which aims to be a reference peak throughout the run, normalizing peak

areas and correcting for fluctuations in the separation field. Finally the destained protein/dye complex

passes the laser and it is excited at 630 nm and emits at 700 nm which is detected by the instrument and

converted into a digital output. The software plots fluorescence intensity versus time and produces

electropherograms for each sample. Quantization of the concentration and accurate sizing of each

fragment are achieved by comparing against a sizing ladder and running internal standards or “markers”

with each sample. Internal standards of known concentration are mixed with the sample to aid in

quantization. The amount of sample sipped into the chip depends on pH, salt concentration, and buffer

additives. The internal standards normalize these factors so that the software can use the ratio of the area

of the curve of the standard to the unknown peak to determine concentration directly. The internal

standards lie slightly outside the assay range so they do not interfere with analysis.

Figure 4. Series of photograph showing the Figure 5. Photograph of the SDS

on-column staining process (a) and injection dilution process. Dilution of 4 (a) and

downward the separation channel. dilution of 12 (b) still show fluorescence

EXPERIMENTAL PART

Based upon the standard LabChip Protein Express Assay (Caliper Labchip GXII User Manual) described in

the Materials and Methods section an IgG1 reduced sample has been denatured and run (Figure 6).

Figure 6. H heavy chain, L light chain, NGH non glycosilated heavy chain, S system peak, LM lower marker. H and L peak

corresponds to 56KDa and 25KDa as reported by literature for IgG1.

Such graph is called electropherogram and offers the size of the molecule on the x axis (size is related with

the time of fluorescence through LIF and thus directly related) and the intensity of fluorescence on the y

axis that is related with the amount of such protein. An electropherogram is a generic output for a

microfluidic on a chip reading. Calculated size of the antibody in figure 6 are 56KDa for heavy chain H which

is in accordance with literature and orthogonal analysis being heavy chain (50KDa) plus glycoform species

(5-7 KDa) depending upon the cell line ability to generate such species. NGH peak therefore lays exactly on

50KDa lacking of the glycosilated part. Light chain as predicted and expected falls at 25KDa and other peaks

are system peaks and internal standards. Furthermore (not reported) the absolute amount of H and L

matches the generic 1:2 ratio. Moreover, the x axis shows an output in seconds that corresponds to the size

of the molecule after lining up the internal standards. Different assays can give slightly different responses

in terms of seconds and it depends upon the user, batch of reagents used and other variables. Following

electropherograms show the main antibody peak at 20sec or 21sec. The author doesn’t point out that such

value can differ between different runs though it must be taken into account.

25KDa

56KDa

50KDa

CE-based glycan type (CEGT) assay.

Microchip CE-SDS was evaluated as high-throughput alternative to conventional CE-SDS for monitoring

monoclonal antibody protein quality (Chen, Tang, Lee, Flynn 2008). A simple glycan typing assay was

developed for the reducing microchip CE-SDS format. Antibodies from crude supernatant (cell culture

process) or purified are treated with Endoglycosidase H which releases the high mannose glycan types.

An overview of the glycosylation process for IgG is offered in Figure 7 (Hristodorov at all, 2012) .

Figure 7. Overview of IgG glycosylation processing in mammalian cell lines producing antibodies.

Endopeptidase H cleaves at the intersection of Mannose (blue triangle) and GlcNac (green circle) and all the

oligosaccarides are released from the main purified antibody. Several species are originated after this and

the resulting product of digestion is usually run on a HPLC and following peptide mapping, along with novel

technologies, most of them time consuming. A time saving, high throughput alternative as Caliper Labchip

has been successfully introduced in early stage screening for candidate cell lines producing antibodies and

replacing HPLC and peptide mapping for very specific applications. Microchip offers a clear advantage for

time saving but it reduces the resolution, although for the screening purpose it still offers an acceptable

range of detection and good linearity. Several antibodies have been checked for quality (data shown) and

compared to conventional CE-SDS/ HPLC technology. The idea, in this second case, is not to screen for size

and amount of the heavy and light chain, but to screen for quality attributes such as high mannose and

hybrid type glycan percentage as reported in figure 8. Refer to Figure 7 for structures; insights about

glycoform go beyond the purpose of this report.

Figure 8. Microchip CEGT assay for measuring HM and hybrid type glycans. (A) Antibody sample treated with excess Endo

H; (B) untreated antibody sample. GlcNAc-H: heavy chain with the glycan reducing end Nacetylglucosamine residue remaining

attached after Endo H digestion.

In this second application, the sample not treated with Endo-H (B) was run and expected not to release the

high-mannose species along with hybrid kind of glycoform and leave the only NGH peak thus not

increasing the existing levels of non glycosilated species NGH. Treated sample (A) still show a peak

corresponding to the main antibody (56KDa) but it is smaller than peak in B (y axis difference is 1400 vs

1750 unit of fluorescence) and a minor peak corresponding to the NGH species (50KDa) as expected though

the NGH peak in this case is bigger since it includes high mannose species released via Endo-H activity.

Calculations of the increase of the NGH peak can be done by the software and the resulting value will

represent the high mannose species existing in the sample during cell culture process. Therefore such tool

can be valuable in order to give a quick estimation about the quality of immunoglobulin produced in a

standard process, prior MS/HPLC (the time consuming assays). Based on the output the scientist can have

an educated guess of which cell lines are potential good candidates as a matter of high mannose, which is a

critical attribute in IgG quality. Data about specific amount of NGH and hybrid species are not disclosed by

the author of the paper.

Immunoglobulin processing study monitoring

A relatively used feature of microfluidic on a chip was to develop a monitoring system to daily check

protein production in cell culture. Part of the work showed next has been done at Abbott Bioresearch

Center, Process Sciences Department. Samples from the same culture have been assayed at different days

to monitor the rate of protein production. Samples have been treated according to manufacture’s direction

and software elaborated. A serial dilution of a standard protein was needed in order to evaluate resolution,

limit of detection and linearity of the assay, figure 9 (4 electropherograms have been overlaid). The assay is

considered valid for protein concentration between 1 g/L and 0.06 g/L.

Figure 9. A serial dilution showing limit of detection, resolution and linearity of Caliper Labchip GXII.

Figure 10 shows the tendency in cell culture process. Particularly of interest, in this case is the main protein

peak at 30 seconds along with other peaks at 19 and 22 seconds, checked at different days. As cell grow

and Viable Cell Density increase throughout the culture time, the antibody concentration increase

accordingly. Early checking is shown in blue, middle checking is red and late checking is green.

Figure 10. Increased production of protein in a cell culture process. Early checking is in blue, middle checking is red and

late checking is green.

Glycoforms studies

A possibility in microfluidic on a chip is to generate different assays based upon different dyes and

molecule-dye interaction. The protein express assay works on a proprietary dye able to bind protein-SDS

complexes. Another possibility is to use a dye that binds sugars and have an understanding about sugar

processing and inclusion in glycosilated proteins. Caliper developed a new assay Called Rapid Analysis of N-

Glycans on the LabChip GXII Microchip-CE Platform (Caliper application note 403). Such assay is based

primarily on different chemistry for the sugar-dye binding (proprietary information, not disclosed). Another

major difference is the density of the gel sieving matrix which is higher than protein express thus allowing

small molecules as sugars to be retained in the separation channel long enough to be read apart. Such

assay can be valuable in cell culture screening and protein quality assessment. Glycosilated protein samples

have to be purified prior (protein A purification is ok for antibodies) and accurately quantified prior

incubation with PNG-ase F (N-Glycosidase F, also known as PNGase F, is an amidase that cleaves between

the innermost GlcNAc and asparagine residues of high mannose, hybrid, and complex oligosaccharides

from N-linked glycoproteins -New England BioLabs). Therefore the sugary part of the antibody is released

and incubated with the dye, sample reconstituted and finally separated on a chip. Details in Figure 11.

Figure 11. Overview of the sample treatment prior a glycan assay on Caliper Labchip GXII.

The electropherogram generated in this case has to be carefully lined up with a standard of the different

olygosaccaride species in order to give an estimate of the amount, as shown in figure 12, 13 and 14.

Figure 12. Overlaid electropherogram of 5 different glycan standards representing the major glycan species.

Figure 13. Overlaid electropherogram of glycan standars and a digested IgG sample profile.

Figure 14. IgG glycan profile with corresponding RSD values.

Glycan profiling through Caliper can be a cost effective and time saving alternative to HPLC/MS.

Detection of dimers in IgG processing

A known mechanism in IgG cell culture is the formation of dimers over time. Single antibody (monomers)

can stick to each other and generate molecular species of double molecular weight. This process yields a

decrease in titer and negatively affects the quality of the culture. Also generates an undesirable molecule

that needs to be further eliminated prior using the process in a FDA approved production. Detecting cell

lines producing dimers at an early stage can lead to a maximization of good candidates and decrease the

downstream purification work, along with minimizing arguments with regulatory agencies. Though no

information is provided by Caliper in order to detect molecules bigger than 200KDa, some groups tried to

modify the software script and play with different reagents in order to properly detect dimers.

The main differences versus the standard protein Express detection are the density of the sieving matrix

which is lower in this case and the pre-treatment of the molecule with a cross-linker agent. Ultimately

software engineering capability is needed in order to change the detection time of the instrument

switching from 40 seconds per reading up to 56 seconds.

Dimers are non-covalently bound monomers and if the sample is denatured, they will detach as monomers

and will lay on the main antibody peak. It will generate a false negative that later on will be detected by SEC

(size exclusion chromatography) which is time and money consuming. Also most labs are not equipped to

bear hundreds of samples for SEC. In this specific case microfluidic on a chip generates a decent resolution

output in a timely fashion. Samples covalently bound will be run on a customized protocol not existing in

the official Caliper literature and the reading will be used to discard cell lines producing high amount of

dimers. A study SEC vs Caliper (modified script/protocol) is shown in figure 15 (Chen, Flynn 2009).

Figure 15. a) Analysis of an aggregate enriched Mab B sample by SEC (a) and crosslinking/microchip CE-SDS (b). LMW, low

molecular weight species. Main antibody peak at 15 seconds in a) and 26 seconds in b).

In this study two orthogonal methods have been tested for checking dimers. SEC order of elution is

opposite than microchip CE-SDS starting with the highest molecular weight specie eluting first. Also no

capacity of detecting oligomers has been shown in Caliper (oligomers like 12H12L or 3mAbs species - triple

the size). Relative abundance is consistent using either method though for more sophisticated studies like

exact molecular weight or hybrid species, SEC is still the main technology. Caliper LabChip shows a decent

accordance and definitely can shorten timelines for screening high number of samples. Also can be used in

automation and high-throughput which makes the instrument highly desirable in biotechnology labs.

CONCLUSIONS

In this review of existing literature about Caliper Labchip GXII, insights about the instrument and principle

of operations have been given. Limited information about dye or chemical properties of reagents used are

available but existing efforts in the customization of protocols and chemistry of the microchip can lead to a

new branch of the area. Microfluidic on a chip, mimicking the CE-SDS electrophoresis can be a valid support

in any lab concerned with studies about protein size and quantity, quality attributes (high mannose,

glycoform and dimers producing IgG) along with optimization and screening of existing platforms. This

technology is still in an embryonic phase to properly replace HPLC, MS or SEC but differently from these

tools, has a high-throughput capability and can screen up to 400 samples a day versus tens or single digit

samples for the mentioned existing tools. The clear advantage shown in time saving is though balanced by a

variable loss in resolution and some difficulties in dealing with limit of detection but there are lots of

improvements in making this attributes better. Specific studies have been offered, including featured white

paper from Caliper ltd. along with independent group involved in microfluidic studies and also samples run

at Abbott Bioresearch Center.

REFERENCES

Cheng, Flynn, Journal of Chromatography (2009) 3012

Primack, Flynn, Pan, Electrophoresis (2011) 1129

Han, Chen, Electrophoresis (2012) 765

Bousse, Mouradian, Minalla, Yee, Williams, Dubrow, Anal. Chem. (2001) 1207

Chen, Tang, Lee, Flynn, Electrophoresis (2008) 4993

Swalley, Fulghum, Chambers, Anal. Biochem. (2006) 122

Caliper Labchip GXII, Technology for Vaccines and Biologics, white paper (2010)

Caliper Labchip GXII, Antibody Analysis GXII, white paper (2010)

Hristodorov, Fischer, Linden, Mol. Biotechnol (2012) DOI 10.1007/s12033-012-9612-x

Jefferis, Biotechnol. Prog (2005) 11