Protein Detection by DigitalQuantification of Amplified SingleMoleculesRui Huang

(1)

Protein Detection by Digital

Quantification of Amplified Single Molecules

Rui Huang

(2)

1

Summary

The digital quantification of amplified single- molecule detection (SMD) method is based on engineering of molecular recognition and amplification reactions that combines specific molecular recognition with background- free signal amplification. The aim of this project is to optimize the combined protocol of using proximity ligation assay (PLA), rolling circle amplification (RCA) and circle-to-circle amplification (C2CA) for the detection of protein biomarkers.

In this project, protein biomarkers (CSTB and NGF) were chosen as the targets and used for protocol optimization. The protocols of solution phase PLA, RCA and C2CA were optimized.

I mainly focused on the comparison of the reagents, the concentration of the reagents and the reaction time of different steps. The simplicity, consumption of time and cost of the protocol were improved and the precision was also improved comparing to qPCR. Furthermore, VEGF and IL-8 were chosen to validate the optimal protocol, demonstrating the low limits of detection (LOD) and broad dynamic range. In conclusion, amplified SMD method combined with PLA, RCA and C2CA was shown as a high-precision and highly sensitive protein detection approach, which is suitable for basic and clinical research as well as diagnostics where proteins need to be quantified at low concentrations.

(4)

3

Abbreviations

SMD Single- molecule detection PLA Proximity ligation assay RCA Rolling circle amplification C2CA Circle-to-circle amplification LOD Limits of detection

ASMs Amplified single molecules ssDNA Single-stranded DNA S/N Signal to noise ratio

qPCR Quantitative real-time polymerase chain reaction CV Coefficient of variation

RO Restriction oligonucleotide

ELISA Enzyme linked immunosorbent assay IL-8 Interleukin-8

VEGF Vascular endothelial growth factor NGF Nerve growth factor

CSTB Cystatin-B

(5)

Introduction

Proximity Ligation Assay (PLA)

Proximity Ligation Assay (PLA) is an immunoassay which converts specific protein detection into DNA analysis, offering high specificity and sensitivity (Fredriksson et al 2002) (Darmanis et al 2010). The antibody-based solution phase PLA was firstly introduced in 2004 (Gullberg et al 2004), which takes place in solution. The reaction includes 3 steps (Figure 1):

(a) Incubating samples with pairs of probes which conjugat e to the oligonucleotides. When two or more probes recognize the same target molecule or molecular complex, the free ends of the oligonucleotides will be brought into proximity. (b) A so-called splint or connector oligonucleotide, which hybridizes to free ends of the proximity probes, serves as a template for proximity probes ligation (Gustafsdottir et al 2005). Ligation happens with the presence of DNA ligase and splint. (c) The ligated products are amplified and detected by real-time PCR (Gustafsdottir et al 2006). Probes that are not in proximity result in nonspecific signals. The requirement of recognize the same target ensure the high degree of the specificity (Nilsson et al 2003).

(6)

5

Figure 1. The principle of solution phase proximity ligation assay. (a) Incubate target molecules (green) and

pairs of pro ximity probes (black antibodies) wh ich connected to the oligonucleotides (blue). The two free ends of

the two oligonucleotides are brought close to each other when they recognize the sa me target mo lecule. (b) A

DNA ligase (orange cross) is added together with the splint (orange line) to join in the ligation of two

oligonucleotides. (c) The DNA ligated products serve as the templates detected by real-t ime PCR.

In this project, instead of using real- time PCR for readout of PLA, a digital quantification method by counting amplified single molecules (ASMs) is used (Jarvius et al 2006). After producing the ligated products, restriction enzymes and restriction oligos are added. The ligated products are digested into numbers of single-stranded DNA (ssDNA). The ssDNA can be ligated and form a circular DNA at the presence of DNA ligase and a template which is complementary to the two ends of ssDNA. (Figure 2).

(7)

Figure 2. The ligated products are digested into ssDNA by presence of restriction enzy mes (dark green dashed)

and restriction oligos (purple lines). A DNA ligase (orange cross) is added with the template (red line) and the

ligated product will hybridize and form a DNA c irc le.

RCA and C2CA

The rolling circle amplification (RCA) was introduced in the mid 1990s. Because of the robustness and high simplicity, it is useful for probe, target and signal amplification (Baner et al 1998) (Lizardi et al 1998). RCA is a linear- isothermal process in the presence of certain DNA polymerases, using the ssDNA minicircle as a template (Fire et al 1995， Daubendiek et al 1995). DNA can be replicated about 500 to 1000 times, depending on the amplification time (Jarvius et al 2006). Briefly, in this project, a linear ligated product is formed in a target protein dependent manner which is then ligated to form a DNA circle by existence of DNA

(8)

7

Circle to circle amplification (C2CA) is based on RCA. It consists of 3 steps: replication, monomerization and ligation (Dahl et al 2004) (Figure 3 B). The original circular DNA is considered as the positive polarity and after one step of replication (RCA reaction) the product is converted into opposite polarity. RCA product is cut into monomers by presence of restriction enzymes and positive polarity of the restriction oligos (RO+). Then the monomers can be guided into ligation step and circularized. These circles serve as the templates for next RCA, primed by the RO+. The process is not product prevented and can be further repeated which can produce around 100- fold higher concentration of the monomers than PCR (Dahl et al 2004).

Figure 3. Sche matic dra wing of the mechanisms of RCA and C2CA. The (+) and ( –) polarit ies are shown in blue

and red respectively. (A) The origina l DNA circle is formed with DNA ligase (orange cross) and the prime r of (–)

polarity (red line with arro w). (B) Three steps of one cycle of C2CA. The orig inal DNA c irc le serves as the (+)

(9)

polarity and the primer of (–) polarity which can be e xtended and produce a long ssDNA with DNA poly me rase.

The ssDNA of RCA product of (–) po larity then be mono merized by restriction enzy mes (dark g reen cashed) and

restriction oligos (RO+) (short blue line). The RO+ (b lue line with arrow) serves as the primer for ne xt RCA.

Amplified single-molecule detection (SMD) method

The amplified single- molecule detection (SMD) method retains the discrete property of the molecules. It has highly precise quantification of molecules and also allows multiplex detection. It is more precise than quantitative PCR (qPCR). Because SMD allows the calculation of the direct number of ASMs but qPCR only allows measuring the ensemble average of the amplified short and diffusible DNA copies (Jarvius et al 2006).

After C2CA, the amplified long ssDNA spontaneously forms into a random coil of DNA, which contains numbers of repeated sequence. The amplified products are labeled by hybridizing fluorescent molecule-tagged probes. They are pumped through thermoplastic micro channel and detected using a confocal microscope operating in line-scanning mode, perpendicular to the direction of liquid flow (Jarvius et al 2006). The probes hybridized on one ssDNA resulting in a confined cluster of fluorophores, which are visible as a bright object.

The raw data files are threshold and output binary files with data points which can be counted.

(10)

9

Figure 4. The mechanis m of a mplified SMD. (a ) Hybrid ize the fluorescent probes (green and red stars) to the

repeated sequences of the amplified products. (b) Pu mp the labeled samp les through the micro channel,

line-scanning by confocal microscope. The signals are visible as the bright objects. (c) Output the raw data files

and calculate the single molecule objects.

Aim

The aim of this project was to optimize the protocol involved in our approach, which includes, solution phase PLA, RCA and C2CA, in order to increase the sensitivity, precision and the dynamic range. The optimization mainly focused on the comparison of reagents, the concentration of the reagents and the reaction time of several steps.

(11)

Results

Comparison of 2 RCA with 3 RCA; with UNG or without UNG

The number of ASMs is determined by the numbers of RCA and C2CA. In this project, 2 RCA included RCA for 1 hour and 1 cycle of C2CA for 1 hour while 3 RCA included RCA for 20 min and 2 cycles of C2CA, 20 min for each cycle. The function of the enzyme UNG was to eliminate the splint which contains dUTP in the sequence, so that the interference and competition from the splint could be prevented after forming the PLA ligated products. Figure 5 (A) and (B) showed the ASMs and signal to noise ratio (S/N) in a serial dilution of NGF with and without UNG (0.02 unit, Fermentas) of 2 RCA and 3 RCA. S/N equals to the number of molecule objects divided by the result of the negative control. The results indicated that 2 RCA with UNG performed equally well as comparing to 3 RCA without UNG.

However, considering to the simplicity of handling, 2 RCA with UNG was chosen to do the following improvements.

(A)

100 1000 10000 100000

amplified single cule objects

2 RCA without UNG 2 RCA with UNG 3 RCA without UNG

(12)

11

(B)

Figure 5. Detection of NGF with and without UNG of 2 RCA and 3 RCA. (A) Nu mber of A mp lified single

mo lecule objects . (B) Signal to noise ratio. The 0.1 pM on X -a xis is the negative control where there is no

antigen in the solution.

Optimization of UNG adding step

The splints containing dUTP can be eliminated by adding UNG. We were considering which step was more efficient to eliminate the splints by adding 0.02 unit UNG, either in the step of digesting the PLA ligated products into ssDNA (Digestion I step) or the step of original RCA circle formation (Ligation step). Adding UNG in Ligation and Digestion I step performed equally well (Figure 6 (B)), but considering to the number of ASMs, adding UNG in Ligation step was a little more than in Digestion I step from 0.1 pM to 1000 pM (Figure 6 (A)). So, UNG was chosen to be added in the Ligation step after the PLA products digesting into ssDNA.

0.1 1 10 100 1000

S/N

NGF (pM)

2 RCA without UNG 2 RCA with UNG 3 RCA without UNG 3 RCA with UNG

(13)

(A)

(B)

Figure 6. Optimization of the UNG adding step. (A) Nu mber of A mplified single mo lecule ob jects . (B) Signal to

noise ratio. Digestion I is the step in which the PLA ligated products were digested into ssDNA. Ligation is the

step in which the original RCA circles were fo rmed. The 0.01 pM on X-a xis is the negative control where there

is no antigen in the solution. Erro r bars along the Y-a xis indicates the standard deviation from the duplicate

measure ments.

1 10 100 1000 10000 100000

0.01 0.1 1 10 100 1000

Counted amplified single molecule objects

NGF (pM)

UNG in Digestion I UNG in Ligation

1 10 100 1000

0.01 0.1 1 10 100 1000

S/N

NGF (pM)

UNG in Digestion I UNG in Ligation

(14)

13

Buffers comparison

In the solution phase PLA, I compared three buffers as the antigen dilution buffer (FBB Buffer I, Strek Buffer, Buffer I) and probes incubation buffer (FBB Buffer II, Strek Buffer, Buffer I). The compositions of these buffers are explained in the Methods and materials section. To be simple, both FBB Buffer I and FBB Buffer II were considered as FBB Buffer.

Comparing with the Strek Buffer, the number of ASMs of CSTB from 0.1 pM to 1000 pM was more than using FBB Buffer (Figure 7). Then, I compared FBB Buffer with Buffer I. By using Buffer I, the differences of the number of ASMs could not be distinguished among each concentration of CSTB (Figure 8). Therefore, FBB buffer was the best among the three.

Figure 7. Co mparison of FBB Buffe r and Strek Buffe r. The co mpositions of the two buffe rs are e xp la ined in the

Methods and Materials section. The 0.01 pM on X-a xis is the negative control where there is no antigen in the

solution.

10 100 1000 10000

0.01 0.1 1 10 100 1000

CSTB (pM)

FBB Buffer Strek buffer

(15)

Figure 8. Co mparison of FBB Buffer and Buffe r I. The co mpositions of the two buffers are exp lained in the

Methods and Materials section. The 0.01 pM on X-a xis is the negative control where there is no antigen in the

solution. Error bars along the Y-a xis indicates the standard deviation fro m the duplicate measurements.

Optimization of probes incubation time and temperature

In the solution phase PLA, the target proteins can be recognized by the proximity probes at an appropriate temperature for a proper time. This reaction would greatly influence the following steps. Incubating at 37 ºC for 1 hour showed a much lower background as well as much higher S/N from 0.1 pM to 1000 pM comparing to the result of incubating at 4 ºC for 1.5 hours (Figure 9). So, 37 ºC for 1 hour was chosen as condition of incubating the target molecules and the proximity probes.

10 100 1000 10000 100000

0.01 0.1 1 10 100 1000

CSTB (pM)

FBB buffer Buffer I

(16)

15

(A)

(B)

Figure 9. Optimization of probes incubation time and te mperature. (A) Nu mber of A mplified single mo lecule

objects. (B) Signal to noise ratio. The 0.01 pM on X-a xis is the negative control where there is no antigen in the

solution. Error bars along the Y-a xis indicates the standard deviation fro m the duplicate measurements.

Optimization of the probe concentrations

After optimizing the probes incubation time and temperature, I compared two concentrations of probes: 35 pM and 70 pM. The curves were almost parallel (Figure 10 (A)) and the curves in Figure 10 (B) were almost superposed. These indicated that the performances were almost

1 10 100 1000 10000

0.01 0.1 1 10 100 1000

CSTB (pM)

37°C 1 hour 4°C 1.5 hours

1 10 100 1000

0.01 0.1 1 10 100 1000

S/N

CSTB (pM)

37°C 1 hour 4°C 1.5 hours

(17)

equally good by using both concentrations. So, 35 pM was enough to carry out the reaction.

The sequences of the probes are shown in Table 1.

(A)

(B)

Figure 10. Optimizat ion of the probe concentrations. (A) Nu mber of A mp lified single molecu le objects . (B)

Signal to noise ratio. The 0.01 pM on X-a xis is the negative control where there is no antigen in the solution.

Error ba rs along the Y-a xis indicates the standard deviation from the duplicate measurements.

100 1000 10000 100000

0.01 0.1 1 10 100 1000

CSTB (pM)

35 pM 70 pM

0.1 1 10 100

0.01 0.1 1 10 100 1000

S/N

CSTB (pM)

35 pM 70 pM

(18)

17

Optimization of the time in Digestion I

Digestion I was the step in which the PLA ligated products were digested into ssDNA at 37 ºC by existence of restriction enzymes and restriction oligos, followed by enzyme inactivation at 65 °C for 5 min. I compared digesting the PLA ligated products at 37 ºC for 10 min, 30 min and 60 min. The curves in Figure 11 (A) were almost parallel, but as seen in Figure 11 (B), digesting for 10 min had higher signal to noise ratio at each concentration of CSTB. Therefore, digesting at 37 ºC for 10 min was enough and demonstrated a good performance.

(A)

(B)

Figure 11. Optimization of the time in Digestion I. (A) Nu mber of A mp lified single mo lecule objects . (B) Signal 200

2000

0.1 1 10 100

CSTB (pM)

10min 30min 60min

1 10

0.1 1 10 100

S/N

CSTB (pM)

10 min 30 min 60 min

(19)

to noise ratio. The 0.1 pM on X-a xis is the negative control where there is no antigen in the solution. Error bars

along the Y-a xis indicates the standard deviation from the duplicate measurements.

Optimization of the concentrations of enzymes in Digestion I

The restriction enzymes recognized and cut the double-stranded DNA formed by the probes and the restiction oligos. AluI (New England Biolabs) and HpyCH4 IV (New England Biolabs) were used for CSTB while AluI (New England Biolabs) and MboI (Fermentas) were used for NGF. I compared three concentrations: 0.1 U/ul, 0.2 U/ul and 0.5 U/ul. The curves were almost parallel in Figure 12 (A), but the curve of 0.1 U/ul enzymes shows a bit lower signal to noise ratio in Figure 12 (B). The curves of 0.2 U/ul and 0.5 U/ul enzymes were almost superimposed, but considering to the cost, 0.2 U/ul AluI and 0.2 U/ul HpyCH4 IV were chosen in Digestion I. The same resluts were concluded for NGF : the concentrations of AluI and MboI were 0.2 U/ul, respectively (Figure 13).

(A)

1000 10000

ted amplified single molecule objects

0.1 U/ul 0.2 U/ul 0.5 U/ul

(20)

19

(B)

Figure 12. Optimizat ion of the concentrations of enzy mes in Digestion I by using CSTB. (A) Nu mber of

Amplified single mo lecule objects . (B) Signal to noise rat io. The 0.1 pM on X-a xis is the negative control where

there is no antigen in the solution. Error bars along the Y-a xis indicates the standard deviation fro m the duplicate

measure ments.

(A)

1 10

0.1 1 10 100

S/N

CSTB (pM)

0.1 U/ul 0.2 U/ul 0.5 U/ul

1 10 100 1000 10000 100000

0.1 1 10 100

NGF (pM)

0.2 U/ul 0.5 U/ul

(21)

(B)

Figure 13. Optimizat ion of the concentrations of enzy mes in Digestion I by using NGF. (A) Nu mber of

measure ments.

Comparison of T4 ligase and Ampligase

After digesting the PLA ligated products, the ssDNA would form the initial RCA circle at the presence of a certain DNA ligase. I compared 0.5 U/μl of T4 ligase (Fermentas) and 0.5 U/μ l of Ampligase (Fermentas) and the results were shown in Figure 14. Ampligase performed much better and the curves were more regulated than T4 ligase both in Figure 14 (A) and (B).

So, Ampligase was regarded as the choice of the certain DNA ligase.

1 10 100

0.1 1 10 100

S/N

NGF (pM)

0.2 U/ul 0.5 U/ul

(22)

21

(A)

(B)

Figure 14. Co mparison of T4 ligase and Ampligase. (A) Nu mber of A mplified single molecule objects . (B)

Optimization the concentration of Ampligase

After pointing out that Ampligase performed better comparing with T4 ligase, I focused on optimizing the concentration of Ampligase in the reaction. The concentrations of 0.2 U/μl, 0.5

100 1000 10000 100000

0.1 1 10 100 1000

NGF (pM)

Ampligase T4 ligase

1 10 100

0.1 1 10 100 1000

S/N

NGF (pM)

Ampligase T4 ligase

(23)

U/μ l and 1 U/μl were compared. Though using 0.2 U/μl Ampligase has higher S/N ratio than 0.5 U/μ l and 1 U/μl Ampligase, the standard deviation of the negative control and 1 pM were higher than the other two curves. Furthermore, the number of ASMs was smaller for each concentration (Figure 15). Comparing the results, 0.5 U/μ l should be chosen as a more appropriate concentration of Ampligase in the reaction.

(A)

(B)

100 1000 10000

0.1 1 10 100

CSTB (pM)

0.2 U/ul 0.5 U/ul 1 U/ul

10 100

S/N 0.2 U/ul

0.5 U/ul 1 U/ul

(24)

23

(B) Signal to noise ratio. The 0.1 pM on X-a xis is the negative control where there is no antigen in the solution.

Optimization of the concentration of restriction oligonucleotide in Digestion II

In Digestion II step, the RCA products were cut into monomers in the presence of restriction enzyme and restriction oligonucleotide. Restriction oligonucleotide III (RO III, the sequence is shown in Table 1) was used in this step and I co mpared two concentrations: 0.08 μM and 0.16 μM. By using 0.16 μM RO III, the number of ASMs and S/N ratio were higher than that of 0.08 μM RO III (Figure 16). Furthermore, the standard deviation of 0.16 μM was smaller of each concentration of CSTB, and the curves were more regulated as well. Therefore, the doubled concentration of RO III was chosen in Digestion II step.

(A)

100 1000 10000

0.1 1 10 100

CSTB (pM)

0.08 μM 0.16 μM

(25)

(B)

Figure 16. Optimization of the concentration of restriction oligonucleotide in Digestion II. (A) Nu mber of

measure ments.

Optimization of the time in Digestion II

In Digestion II step, the long ssDNA were digested into monomers at 37 ºC and followed by enzyme inactivation at 65 °C for 5 min. I compared at 37 ºC for 10 min with 30 min. The curves were almost parallel in Figure 17 (A), and were almost superposed in Figure 17 (B). It indicated that there was no need to digest for 30 min; 10 min was enough for monomerization.

1 10

0.1 1 10 100

S/N

CSTB (pM)

0.08 μM 0.16 μM

(26)

25

(A)

(B)

Figure 17. Optimization of the t ime in Digestion II. (A) Nu mber of A mplified single mo lecule objects . (B)

Optimization of the amount of PLA ligated products

The prepared PLA ligated products of each concentration were 100 μl and the amount taken into the following processes should be optimized. I compared 5 μl, 10 μl and 15 μl of PLA

100 1000 10000

0.1 1 10 100

CSTB (pM)

10min 30 min

1 10

0.1 1 10 100

S/N

CSTB (pM)

10 min 30 min

(27)

ligated products. The curves were almost parallel in Figure 18 (A) and were almost superposed in Figure 18 (B). The results illustrated that they were performed equally well.

However, the coefficient of variation (CV) should be taken into account (Figure 18 (C)). CV (%) is a measurement of dispersion of the distribution probability. In this experiment, it equaled to the standard deviation divided by the mean of ASMs. Theoretically, the bigger the mean is the smaller the CV (%) should be. Figure 18 (A) demonstrated that by using 15 μ l of products had more ASMs of each concentration comparing with 5 μ l and 10 μl. It indicated that the CV (%) would be smaller by using 15 μ l of PLA ligated products theoretically.

Therefore, 15 μl of ligated products was chosen.

(A)

100 1000 10000

0.1 1 10 100

CSTB (pM)

5 μl 10 μl 15 μl

(28)

27

(B)

(C)

Figure 18. Optimization of the a mount of PLA ligated products. (A) Nu mber of A mp lified single mo lecule

objects. (B) Signal to noise ratio. (C) Coeffic ient of variat ion. The 0.1 pM on X-a xis is the negative control

where there is no antigen in the solution. Error bars along the Y-a xis indicates the standard deviation fro m the

duplicate measurements.

Comparison of the restriction oligonucleotide in Digestion I

There were two restriction oligonucleotides (RO I and RO V) for NGF to digest the PLA ligated products. RO V had longer complementary sequence than RO I, so RO V had higher

1 10 100

0.1 1 10 100

S/N

CSTB (pM)

5 μl 10 μl 15 μl

1 10 100

0.1 1 10 100

CV (%)

CSTB (pM)

5 μl 10 μl 15 μl

(29)

specificity to the probes. The results indicated that the two restriction oligos were performed equally well (Figure 19). Therefore, both of the oligos could be used. The sequences of RO I and RO V are shown in Table 1.

(A)

(B)

Figure 19. Co mparison of the restriction oligonucleotide in Digestion I. (A) Nu mber of A mplified single 1

10 100 1000 10000 100000

0.1 1 10 100

NGF (pM)

RO I RO V

1 10 100 1000

0.1 1 10 100

S/N

NGF (pM)

RO I RO V

(30)

29

Standard curves of CSTB and NGF

After the optimization of the protocol, the standard curves of CSTB and NGF were shown in Figure 20 and Figure 21. The number of ASMs and the S/N ratio could be distinguished between the negative control and 1 pM. The CV (%) of CSTB was below 10% (Figure 20 (C)) and the CV (%) of NGF was around 10% (Figure 21 (C)).

(A)

(B)

100 1000 10000

0.01 0.1 1 10 100 1000

CSTB (pM)

CSTB

0.1 1 10 100

0.01 0.1 1 10 100 1000

S/N

CSTB (pM)

CSTB

(31)

(C)

Figure 20. Standard curve of CSTB. (A) Nu mber o f A mplified single mo lecule objects . (B) Signal to noise ratio.

(C) Coefficient of variation. The 0.01 pM on X-a xis is the negative control where there is no antigen in the

solution. Error bars along the Y-a xis indicates the standard deviation fro m the triplicate measure ments.

(A)

1 10 100

0.01 0.1 1 10 100 1000

CV (%)

CSTB (pM)

CSTB

100 1000 10000 100000

0.01 0.1 1 10 100 1000

NGF (pM)

NGF

(32)

31

(B)

(C)

Figure 21. Standard curve of NGF. (A) Nu mbe r of Amp lified single molecu le objects . (B) Signal to noise ratio.

solution. Error bars along the Y-a xis indicates the standard deviation fro m the triplicate measure ments.

Multiplex Detection

We combined the optimal protocol of NGF and CSTB to detect the two biomarkers at the same time (Figure 22). The number of ASMs and the S/N ratio could be distinguished

1 10 100 1000

0.01 0.1 1 10 100 1000

S/N

NGF (pM)

NGF

1 10 100

0.01 0.1 1 10 100 1000

CV (%)

NGF (pM)

NGF

(33)

between the negative control and 1 pM. The Figure 22 (D) shows the detected result (1000 pM) by confocal microscope, and stored in a 24-bit rgb-TIF file.

(A)

(B)

10 100 1000 10000 100000

0.01 0.1 1 10 100 1000

Counted amplified single molecule objects

NGF and CSTB (pM)

NGF CSTB

0.1 1 10 100 1000

0.01 0.1 1 10 100 1000

S/N

NGF CSTB

(34)

33

(C)

(D)

Figure 22. Standard curve of mult iple x detection. (A) Nu mbe r of A mp lified single molecu le objects . (B) Signal

to noise ratio. (C) Coeffic ient of variation. (D) The data of 1000 pM stored in a 24-b it rgb-TIF file . The 0.01 pM

on X-a xis is the negative control where there is no antigen in the solution. Error bars along the Y-a xis indicates

the standard deviation fro m the triplicate measure ments.

Detection of IL-8 and VEGF

The other two biomarkers: IL-8 and VEGF were detected by using the optimal protocols of CSTB and NGF, respectively. The results were shown in Figure 23 and Figure 24. The

1 10 100

0.01 0.1 1 10 100 1000

CV (%)

NGF CSTB

(35)

background was very small and the number of objects and the S/N ratio could be distinguished between the negative control and 0.1 pM.

(A)

(B)

(C)

1 10 100 1000 10000

0.01 0.1 1 10 100 1000

IL-8 (pM)

IL-8

1 10 100 1000 10000

0.01 0.1 1 10 100 1000

S/N

IL-8 (pM)

IL-8

10 100 1000

CV (%)

IL-8

(36)

35

Figure 23. Standard curve of IL-8. (A) Nu mber of A mp lified single mo lecule objects . (B) Signal to noise ratio.

solution. Error bars along the Y-a xis indicates the standard deviation fro m the duplicate measurements.

(A)

(B)

(C)

1 10 100 1000 10000 100000

0.01 0.1 1 10 100 1000

VEGF (pM)

VEGF

1 10 100 1000 10000

0.01 0.1 1 10 100 1000

S/N

VEGF (pM)

VEGF

0.1 1 10 100

0.01 0.1 1 10 100 1000

CV (%)

VEGF (pM)

VEGF

(37)

Figure 24. Standard curve of VEGF. (A) Nu mber of A mp lified single molecu le objects . (B) Signal to noise ratio.

solution. Error bars along the Y-a xis indicates the standard deviation fro m the duplicate measu rements.

(38)

37

Discussion

In order to increase the sensitivity, precision and dynamic range of our protocol, which includes solution phase PLA, RCA and C2CA, I compared the reagents, the concentration of the reagents and the reaction time of several steps. CSTB and NGF were chosen as the targets and used for protocol optimization while VEGF and IL-8 were used to validate our approach.

The protocol was simplified, like 3 RCA was changed to 2 RCA, so that one cycle of C2CA was reduced. Furthermore, the time consumption was reduced. For instance, the probes incubation time as well as the reaction times in Digestion I and II was optimized. In addition, the cost was saving as well by reducing concentrations of the enzymes, probes and restriction oligos, etc. Accordingly, the simplicity, consumption of time and cost of the protocol were improved.

The amplified SMD method is more precise than qPCR. It can be validated by CV (%), which is a measurement of dispersion of the distribution probability. The CV (%) was around 20% to 30% by using qPCR (Fredriksson et al 2002) (Darmanis et al 2010). However, the CV (%) for detection of all the four biomarkers was below 20% and mainly around 10% by using our approach. Obviously, the precision was improved comparing with qPCR. Therefore, proteins can be quantified more precisely by using our approach and detect small variations.

The precision of protein quantification was improved; however, we didn’t see the

(39)

improvement of LOD in our assay for detection of NGF and CSTB. We suspected that it could be due to the affinity and specificity of the proximity probes used in the experiment. We then chose to use VEGF and IL-8, which we know that good proximity probes for these two targets are available within the group, to validate the approach again. As expected, when detecting these two targets, much better sensitivity was achieved by using our approach.

Accordingly, our hypothesis is correct and the results can be improved by using better performed proximity probes. As a consequence of better proximity probes used in the detection of VEGF and IL-8, the dynamic range of the assay was also much better than the other two targets. Theoretically, the number of amplification objects should increase 10- fold as antigen concentration increases 10- fold. We therefore concluded that, to get a good performance in PLA, the quality of proximity probe is one of the dominant factors.

One of the important parameters in a typical immunoassay is the dynamic range. Generally, PLA provides a much broader dynamic range than ELISA, therefore is better for quantitative detection. The dynamic range of solution PLA in our approach depends on several factors, the concentration of proximity probes, the dynamic range of the instrument and the reaction components. The upper limit of the dynamic range of the digital quantification of amplified SMD method could be reached if too many ASMs are present in the sample. The reason for this is that too many ASMs present in the sample at the same time will cause the overlap between each ASM. Another reason is the reagents, such as dNTP and phi29 polymerase, are

(40)

39

In conclusion, amplified SMD method combined with PLA, RCA and C2CA has shown as a high-precision and highly sensitive protein detection approach. So, small variations of target proteins in the samples could be picked up more easily by our method. It is suitable for basic and clinical research as well as diagnostics where proteins needed to be quantified at low concentrations.

(41)

Materials and methods

Buffers, biomarkers and oligonucleotides

1 x PBS: 1 x PBS contained 137 mM NaCl, 2.7 mM KCl, 10 mM sodium phosphate dibasic, 2 mM potassium phosphate monobasic and the pH was 7.4.

Probes dilution buffers: (a) FBB Buffer I contained FBB buffer (Olink), Triton X-100, 1%

purified BSA (Bovine Serum Albumin, New England Biolabs); (b) Buffer II contained 0.1%

purified BSA, 2mM EDTA, 1x PBS ; (c) Strek Buffer contained 1 mM D-biotin (Invitrogen), 0.1% purified BSA, 0.05% Tween-20 (Sigma-Aldrich), 100nM goat IgG (Sigma-Aldrich), 0.1 μg/μl salmon sperm DNA (Invitrogen), 5 mM EDTA, 1xPBS.

Biomarke rs dilution buffers: (a) FBB Buffer II contained FBB buffer (Olink), Triton X-100, 0.1% purified BSA; (b) Buffer II; (c) Strek Buffer.

Ligase dilution buffer: 50% Glycerol, 10 mM Tris-HCl pH 8.0, 50 mM KCl, 20 mM DTT, 0.1 mM EDTA.

Proximity probes ligation buffe r: 20 mM Tris-HCl pH 8.0, 25 mM KCl, 2.5 mM MgCl2,2

(42)

41

Biomarke rs:

The mainly used antigens were Cystatin- B (CSTB, R&D Systems) and Nerve growth factor (NGF, R&D Systems). We also used the optimized protocol to detect vascular endothelial growth factor (VEGF, R&D Systems) and Interleukin-8 (IL-8, R&D Systems).

The oligonucleotide sequences are shown in Table 1.

Table 1. Oligonucleotides

Name Seque nce Modi fication

NGF Probe I 5'-A GTGA GCTA GA CTTATTGCGTCACGATGA GACTG GATGAA-3'

5' Th iol

Probe II 5'-TCACGGTA GCATAA GGTGCA GGCGATCCAATTATC AGTA C-3'

5' Phosphate;

3' Th iol

RO I 5'-GTCTA GCTCA CT-3' None

RO II 5'-GATAATTGGATCGCCT-3' None

RO V 5'-ATAAGTCTA GCTCA CT-3' None

Te mpl ate 5'-ACGCAATAAGTCTA GGCCTGCA CCTTATGC-3' None

DO– 5'-ACGCAATAAGTCTA GGCC-3' 5’ Cy 3

DO+ 5'-GGCCTA GACTTATTGCGT-3' 5’ Cy 3

CSTB Probe I 5'-GCCCAATGTCCAATAGCTTATCACGATGA GA CTGG ATGAA-3'

5' Th iol

Probe II 5'-TCACGGTA GCATAA GGTGCA GA GATCATACGTCCT ACAAC-3'

5'Phosphate;

3' Th iol

RO I 5'-GTTGTA GGA CGTATGA-3' None

RO II 5'-GATAAGCTATTGGA CATTGG-3' None

Te mpl ate 5'-TCTCATCGTGATAAGTATGATCTCTGCA CC-3' None

DO– 5'-CGTGATAA GTATGATCTCTGC-3' 5’ Cy 5

Splint 5'-UGA CCUACUUA GUGCCAUCGUA-3' None

RO III 5'-CTGGATGAATCACGGTA G-3' None

RO IV 5'-CTA CCGTGATTCATCCA G-3' None

Solution phase PLA

The whole procedure was based on the original protocol. The antigen that would be detected

(43)

in the assay was serially diluted from 1000 pM to 0.1 pM in antigen dilution buffer. A negative control with no antigen added was included. The antibody conjugated with probes (Olink) were mixed and diluted in probes dilution buffer. T4 ligase (Fermentas) was diluted to 0.004 unit by using ligase dilution buffer. Solution phase PLA, started with incubation of 2 μ l diluted antigen and 2 μl affinity probes. After incubation, the proximity probes were ligated by adding 96 μ l ligation mix which contained ligation buffer and diluted T4 ligase. It was incubated at 37 °C for 10 min, followed by the enzymes inactivation at 65 °C for 10 min.

I optimized (a) the probes dilution buffer and antigen dilution buffer, (b) the concentration of the probes, (c) the probes incubation time and temperature.

Digestion of PLA products and preparation of DNA circle

After PLA, the ligated products were digested by adding 0.2 μM restriction oligo I and 0.2 μM restriction oligo II of NGF or restriction oligo I and II of CSTB, 1 unit AluI (New England Biolabs), 1 unit MboI (Fermentas) for NGF or 1 unit HpyCH4 IV (New England Biolabs) for CSTB, 0.2 μg/μ l BSA and phi29 DNA polymerase buffer in a total volume of 5 μl. It was incubated at 37 °C for 10 min, followed by the enzymes inactivation at 65 °C for 5 min. After formation of the ssDNA, the DNA circle was formed by adding 0.5 unit Ampligase (Fermentas) and 0.1 μM template oligonucleotide in 3 x Ampligase buffer (Fermentas) in a

(44)

43

I optimized (a) which step to add UNG (Fermentas), (b) the amount of the PLA products to be tested, (c) the concentration of restriction enzymes, (d) digesting time, (e) if using T4 ligase or Ampligase, (f) the concentration of the ligase, (g) one of the restriction oligo of NGF comparing to another oligo with different sequence.

C2CA reaction

(a) The replication of the first cycle of C2CA was primed by the excessive template oligonucleotide in the step of preparation of DNA circle, 0.6 unit phi29 DNA polymerase, 0.25 mM dNTP, 0.2 μg/μl BSA and phi29 DNA polymerase buffer in a total volume of 5 μl. This was incubated at 37 °C for 20 min and terminated by 1 min incubation at 65 °C.

(b) The RCA products were monomerized by adding the reagent with 0.4 μM restriction oligonucleotide (RO III), 1 unit Hinf I (Fermentas) in phi29 DNA polymerase buffer in a total volume of 5 μl and incubate at 37 °C for 10 min, followed by the enzymes inactivation at 65 °C for 5 min.

(c) The ligation of the first cycle and the replication of the second cycle of C2CA were carried out by the excessive restriction oligonucleotide (RO III), 0.3 unit phi29 DNA polymerase , 0.05unit T4 ligase, 0.25 mM dNTP, 2 mM ATP, 0.2 μg/μl BSA and phi29 DNA polymerase buffer in a total volume of 10 μl. This was incubated at 37 °C for 20 min and terminated by 1 min incubation at 65 °C.

(d) The monomerization was directed by adding the reagent contained 0.8 μM restriction oligonucleotide (RO IV), 1 unit Hinf I and phi29 DNA polymerase buffer in a total

(45)

volume of 5 μl and incubated at 37 °C for 10 min. After monomerization, the enzyme was inactivated at 65 °C for 5 min.

(e) The monomerized products were circularized and further replicated using the same reagent mixture as in (c).

I optimized (a) the cycle number of C2CA, (b) the concentration of restriction oligonucleotide in Digestion II, (c) digesting time of Digestion II

Hybridization of the fluorescent based probes

The reaction of hybridization carried out by adding 10 nM detection oligo in 40 mM Tris-HCl pH 8.0, 40 mM EDTA, 0.2% Tween-20 and 2M NaCl. It was incubated at 80 °C for 1 min, followed by 65 °C for 10 min.

Multiplex detection

Both NGF and CSTB were diluted together from 1000 pM to 0.1 pM in antigen dilution buffer. The probes of NGF and CSTB were mixed and diluted in probes dilution buffer. Then the same protocol of solution phase PLA was used. The following steps were the combination of the two protocols of NGF and CSTB.

Protein Detection by DigitalQuantification of Amplified SingleMoleculesRui Huang