應用螢光顯微術比較分析兩肺腺癌細胞株CL1-0及CL1-5細胞核染色質

(1)

國立臺灣大學理學院物理學系碩士論文

Department of Physics College of Science

National Taiwan University Master Thesis

應用螢光顯微術比較分析兩肺腺癌細胞株 CL1-0 及 CL1-5 細胞核染色質

Comparative analysis of nucleus chromosome of lung

adenocarcinoma cell lines, CL1-0 and CL1-5, by fluorescence microscopy

傅學聖 Xue-Sheng Fu

指導教授：董成淵博士

Advisor: Chen-Yuan Dong, Ph.D.

中華民國 105 年 7 月

(2)

Abstract

Lung cancer is the most lethal cancer. In Taiwan, the mortality rate of lung cancer ranks first among the cancer-related mortality rate in recent years. To reduce the morality, the early detection of the lung cancer may help the patient get the proper treatment in time. The DNA content abnormality is one of the characteristics which are observed in the tumor cells. Some researches show the potential prognostic value of it. In this thesis, to find the relation between invasive ability and DNA content abnormality, we use the two lung adenocarcinoma cell lines, CL1-0 and CL-5 which have different invasive ability and stain their nuclei with DAPI. By the fluorescence microscopy technique, quantify the DNA content of nuclei with the fluorescence and get nuclei DNA content distribution histogram. From the nuclei DNA content data, we find that the cell line with higher invasive ability has larger DNA content abnormality and the standard deviation of the DNA content of the cells may have potential to be the invasive ability index.

Keywords: lung adenocarcinoma, CL1-0, CL1-5, two-photon microscopy, DAPI, nucleus, DNA content abnormality

(3)

摘要

在所有癌症中肺癌是最致命的一種，近幾年在臺灣肺癌時常居於癌症十大死因之首。為了降低肺癌的死亡率，在肺癌早期時發現能幫助病人即時的得到適當的治療。在腫瘤細胞中的 DNA 含量異常是一個被觀察到的特徵，有些研究顯示這有用來預測病情的潛力。在本論文中

為了找出轉移能力和DNA 含量異常的關係，使用了兩種不同轉移能力的肺腺癌細胞株 CL1-0

及CL1-5 並使用 DAPI 這種染劑對細胞核染色。透過螢光顯微術，將 DNA 含量以螢光訊號量

化，並得到細胞核DNA 含量的分布圖。從這些 DNA 含量的資料發現，DNA 含量異常越顯

著的細胞株，其轉移能力亦越強。此外，計算一群細胞的DNA 含量的標準差，有做為轉移

能力指標的潛力。

關鍵字: 肺腺癌、CL1-0、CL1-5、雙光子顯微術、DAPI、細胞核、DNA 含量異常

(4)

Lung cancer is the second most common cancer in both men and women, and the most lethal among all kinds of cancers[3]. In Taiwan, the mortality rate of lung cancer is 39.2 and ranks first among the cancer-related mortality rate, 197.0 in 2014[1]. The five-year survival rate is about 17% in the United States[2].

Lung cancer can be classified into two main types, non-small cell lung cancer (NSCLC) and small cell lung cancer (SCLC). There are about 80% to 85% of lung cancers which are NSCLC and 10%

to 15% which are SCLC. NSCLC can be further divided into several subtypes which start from different types of lung cells. The subtypes are adenocarcinoma, squamous cell (epidermoid)

carcinoma, large cell (undifferentiated) carcinoma, and other subtypes which is much less common.

Among the subtypes, adenocarcinoma is the most common that about 40% of lung cancers are adenocarcinomas. Adenocarcinoma occurs mainly in current or former smokers, but it is also the most common type of lung cancer seen in non-smokers[3].

The early detection of lung cancer is challengeable. The symptoms usually do not appear until the disease is at an advanced stage. However, there are still some methods that have potential to achieve early detection. In clinical, low-dose computed tomography (LDCT) is a technique to find lung cancer before the symptoms appear. And it has been shown by a study that the screening can lower the mortality from lung cancer[4].

DNA content abnormality or the common term of aneuploidy probably reflects a high genotypic instability in lung cancer. There are some reports showing that DNA content abnormality measured by flow cytometry might be a prognostic factor[5][6], however the conflict results also exists[7].

Despite the heterogeneity across the studies, there is a meta-analysis conclude that patients who benefit from a surgical resection for NSCLC with aneuploid DNA content have a higher risk of death. [8].

(8)

5.1 5.3 7.1

8.3 8.8

11.0 12.0

17.2

31.2 31.6

5.5 7.7 8.1 10.0 10.4 11.6 17.7 23.9 34.9

39.2

胰臟癌食道癌

攝護腺癌子宮頸癌

口腔癌胃癌

女性乳癌結腸直腸癌

肝癌肺癌

子宮頸及部位未明示子宮癌食道癌胰臟癌胃癌前列腺(攝護腺)癌

口腔癌女性乳房癌結腸、直腸和肛門癌肝和肝內膽管癌

氣管、支氣管和肺癌

160.5 197.0

50.0 40.0 30.0 20.0 10.0 0.0 10.0 20.0 30.0 40.0 50.0

103年 93年

圖20 十大癌症死亡率--103年 vs.93年

所有癌症死因死亡率所有癌症死因死亡率

每十萬人口

In this thesis, the method to measure DNA content of cells is imaging by fluorescence microscopy.

By comparing the two cell lines CL1-0 and CL1-5 which have different invasive ability, try to find the judgment of invasive ability with DNA content abnormality.

Figure 1.1 the ranking and population of different cancer cause death in 2014

(9)

Chapter 2. Basic principles

2.1 Two –Photon Excitation[9]

To understand two-photon excitation, time-dependent Schrödinger equation is used to calculate the photon absorption rate. By perturbation theory, we derive the standard result of the single-photon absorption rate and generalize the result to higher order which represent the multiphoton absorption processes.

We assume that the atomic wave function |𝜓(𝒓, 𝑡)〉obeys the time-dependent Schrödinger equation

𝑖ħ 𝜕

𝜕𝑡│𝜓(𝑡)〉 = 𝐻̂│𝜓(𝑡)〉, (2.1.1)

where the Hamiltonian 𝐻̂ is represented

𝐻̂ = 𝐻̂₀+ 𝑉̂(𝑡), (2.1.2)

Here 𝐻̂₀ is the time-independent Hamiltonian for free atom and

𝑉̂ = −𝜇̂𝐸̃(𝑡), (2.1.3)

where 𝜇̂ = −e𝑟̂, is the interaction energy with the electrified of the incoming light. For simplicity we take the monochromatic wave of which the electric field takes the form

𝐸̃(𝑡) = 𝐸𝑒^{−𝑖𝜔𝑡}+ 𝐸^∗𝑒^𝑖𝜔𝑡 . (2.1.4)

We assume the solutions of 𝐻̂₀ are known as the eigenstates │𝑛〉and eigenvalues 𝐸_𝑛, then the time dependent wave, │𝜓(𝑡)〉, can be represented as

(10)

│𝜓(𝑡)〉 = ∑ 𝑎_𝑛(𝑡)𝑒^−𝑖𝜔^𝑛^𝑡│𝑛〉

𝑛

. (2.1.5)

where 𝜔_𝑛 = 𝐸_𝑛/ħ. Substitute Eq. (2.1.5) into Eq. (2.1.1), then the equation becomes

𝑖ħ ∑[𝑎_𝑛̇ (𝑡)−𝑖𝜔_𝑛𝑎_𝑛(𝑡)]𝑒^−𝑖𝜔^𝑛^𝑡│𝑛〉

𝑛

= (𝐻̂₀+ 𝑉̂) ∑ 𝑎_𝑛(𝑡)𝑒^−𝑖𝜔^𝑛^𝑡│𝑛〉

𝑛

(2.1.6)

The two terms corresponding to the free atom solution can be canceled. Therefore we get 𝑖ħ ∑ 𝑎_𝑛̇ (𝑡)𝑒^−𝑖𝜔^𝑛^𝑡│𝑛〉

𝑛

= 𝑉̂ ∑ 𝑎_𝑛(𝑡)𝑒^−𝑖𝜔^𝑛^𝑡│𝑛〉

𝑛

. (2.1.7)

To simplify the Eq. (2.1.7), we use the orthonormality condition,〈m│𝑛〉 = δ

𝑚𝑛. multiply

〈𝑚│ on both sides of equation. We obtain

𝑖ħ𝑑𝑎_𝑚

𝑑𝑡 = ∑ 𝑎_𝑛(𝑡)𝑒^−𝑖𝜔^𝑛𝑚^𝑡𝑉_𝑚𝑛

𝑛

(2.1.8)

where 𝑉_𝑚𝑛 = 〈m│𝑉̂│n〉and 𝜔_𝑛𝑚 = 𝜔_𝑛 − 𝜔_𝑚.Using perturbation techniques, we introduce a parameter λ, replace 𝑉_𝑚𝑛 by λ𝑉_𝑚𝑛 and expand 𝑎_𝑛(t) in powers of the interaction as

𝑎_𝑚(t) = 𝑎_𝑚⁽⁰⁾(𝑡) + λ𝑎_𝑚⁽¹⁾(𝑡) + λ²𝑎_𝑚⁽²⁾(𝑡) + ⋯ . (2.1.9)

By equating the powers of λ on both sides of Eq. (2.1.8), we get 𝑑𝑎_𝑚^(𝑁)

𝑑𝑡 = (𝑖ħ)⁻¹∑ 𝑎_𝑛^(𝑁−1)𝑒^−𝑖𝜔^𝑛𝑚^𝑡𝑉_𝑚𝑛

𝑛

. 𝑁 = 1,2,3, … (2.1.10)

Assume that the atom is originally in the state g without the applied laser field so that

𝑎_𝑔⁽⁰⁾(𝑡) = 1, and 𝑎_𝑛⁽⁰⁾(𝑡) = 0 for n ≠ g . (2.1.11)

for all time t.

(11)

By using Eq. (2.1.3) and Eq. (2.1.4), represent 𝑉_𝑚𝑛 as −𝜇_𝑚𝑛 (𝐸𝑒^{−𝑖𝜔𝑡}+ 𝐸^∗𝑒^𝑖𝜔𝑡) and consider the first order of Eq. (2.1.10) which is

𝑑𝑎_𝑚⁽¹⁾

𝑑𝑡 = −(𝑖ħ)⁻¹𝜇_𝑚𝑔 [𝐸𝑒^𝑖(𝜔^𝑚𝑔^−𝜔)𝑡+ 𝐸^∗𝑒^𝑖(𝜔^𝑚𝑔^+𝜔)𝑡]. (2.1.12) The equation above can be integrated to solve 𝑎_𝑚⁽¹⁾ as

𝑎_𝑚⁽¹⁾(𝑡) = −(𝑖ħ)⁻¹𝜇_𝑚𝑔 ∫ 𝑑𝑡^′

𝑡 0

[𝐸𝑒^𝑖(𝜔^𝑚𝑔^{−𝜔)𝑡′}+ 𝐸^∗𝑒^𝑖(𝜔^𝑚𝑔^{+𝜔)𝑡′}]

= 𝜇_𝑚𝑔𝐸

ħ(𝜔_𝑚𝑔− 𝜔)[𝑒^𝑖(𝜔^𝑚𝑔^−𝜔)𝑡− 1] + 𝜇_𝑚𝑔𝐸^∗

ħ(𝜔_𝑚𝑔+ 𝜔)[𝑒^𝑖(𝜔^𝑚𝑔^+𝜔)𝑡− 1]

(2.1.13)

The first term of Eq. (2.1.13) represents the process of the one photon absorption and the second term correspond to the one photon emission. Since our interest is photon absorption, we drop the second term. The probability 𝑝_𝑚⁽¹⁾ that the atom is in state m at time t is given by

𝑝_𝑚⁽¹⁾= |𝑎_𝑚⁽¹⁾(𝑡)|² = |𝜇_𝑚𝑔𝐸|² ħ²

|𝑒^𝑖(𝜔^𝑚𝑔^−𝜔)𝑡− 1|² (𝜔_𝑚𝑔− 𝜔)²

=|𝜇_𝑚𝑔𝐸|² ħ²

4 sin²[(𝜔_𝑚𝑔− 𝜔) 𝑡/2]

(𝜔_𝑚𝑔− 𝜔)² =|𝜇_𝑚𝑔𝐸|² ħ² 𝑓(𝑡)

(2.1.14)

where

𝑓(𝑡) =4 sin²[(𝜔_𝑚𝑔− 𝜔) 𝑡/2]

(𝜔_𝑚𝑔− 𝜔)² 𝑎𝑛𝑑 lim

𝑡→∞𝑓(𝑡) = 2πtδ(ω_mg− ω) (2.1.15) With the Eq. (2.1.14), the Eq. (2.1.15) at large t becomes

𝑝_𝑚⁽¹⁾ =|𝜇_𝑚𝑔𝐸|²t

ħ² 2πδ(ω_mg− ω). (2.1.16)

The infinity of the delta function which appears in Eq. (2.1.16) is somewhat unphysical in the real situation. To deal with that, we should consider that the final state m would spread into a density of

(12)

state ρ(ω_mg) and the ρ(ω_mg)𝑑𝜔_𝑚𝑔 is the probability of transition frequency lying between 𝜔_𝑚𝑔 and 𝜔_𝑚𝑔+ 𝑑𝜔_𝑚𝑔

The density of the final state is normalized such that

∫ 𝑑ω_mg

∞ 0

ρ(ω_mg) = 1. (2.1.17)

Using the density function of continuous state m, the transition probability 𝑝_𝑚⁽¹⁾ that the atom transit from state g to state m is given by

𝑝_𝑚⁽¹⁾ =2𝜋|𝜇_𝑚𝑔𝐸|²𝑡

ħ² ∫ 𝑑ω_mg

∞ 0

ρ(ω_mg)𝛿(𝜔_𝑚𝑔 − 𝜔)

= 2𝜋|𝜇_𝑚𝑔𝐸|²𝑡

ħ² ρ(ω_mg= 𝜔).

(2.1.18)

The transition probability above is linearly increase with the time, hence the transition can rate can be defined as

𝑅_𝑚𝑔⁽¹⁾ =𝑝_𝑚⁽¹⁾

𝑡 = 2𝜋|𝜇_𝑚𝑔𝐸|²

ħ² ρ(ω_mg = 𝜔). (2.1.19)

Now we get the result of linear absorption, the transition rate of the single photon absorption with the applied laser field of which the frequency is 𝜔.

Figure 2.1.1 State m as the density of states described by

m

g

ρ(ω_mg)

ħω

ρ

(13)

Next we will calculate the second order term of the Eq. (2.1.10) which describes the two-photon absorption process (Fig 2.1.2).

As the previous discussion, we only keep the term that represent absorption process in 𝑉_𝑛𝑔 which becomes 𝑉_𝑛𝑔 = −𝜇_𝑛𝑔𝐸𝑒^{−𝑖𝜔𝑡}. With the Eq. (2.1.13), the second order term of the Eq. (2.1.10) can be written as

𝑑𝑎_𝑛⁽²⁾

𝑑𝑡 = (𝑖ħ)⁻¹∑ 𝑎_𝑚⁽¹⁾𝑒^−𝑖𝜔^𝑚𝑛^𝑡𝑉_𝑛𝑚

𝑚

= −(𝑖ħ)⁻¹∑ 𝜇_𝑛𝑔𝜇_𝑚𝑔𝐸²

ħ(𝜔_𝑚𝑔− 𝜔)𝑒^𝑖(𝜔^𝑛𝑔^{−2𝜔)𝑡}.

𝑚

(2.1.20)

Integrate Eq. (2.1.20)

= ∑ 𝜇_𝑛𝑔𝜇_𝑚𝑔𝐸²

ħ²(𝜔_𝑚𝑔 − 𝜔)[𝑒^𝑖(𝜔^𝑛𝑔^{−2𝜔)𝑡}− 1 𝜔_𝑛𝑔− 2𝜔 ]

𝑚

.. (2.1.21)

The probability of the atom being state n at time t is given by

𝑝_𝑚⁽²⁾= |𝑎_𝑛⁽²⁾|² = |∑ 𝜇_𝑛𝑔𝜇_𝑚𝑔𝐸² ħ²(𝜔_𝑚𝑔− 𝜔)

𝑚

|

2

|𝑒^𝑖(𝜔^𝑛𝑔^{−2𝜔)𝑡}− 1 𝜔_𝑛𝑔− 2𝜔 |

2

.. (2.1.22)

The last term of the Eq. (2.1.22) has the same form as the Eq. (2.1.14).

Virtual state

ħω

n m

g

Figure 2.1.2 two photon absorption process

(14)

While t → ∞, Eq. (2.1.22) would becomes

𝑝_𝑛⁽²⁾ = |𝑎_𝑛⁽²⁾|² = |∑ 𝜇_𝑛𝑔𝜇_𝑚𝑔𝐸² ħ²(𝜔_𝑚𝑔− 𝜔)

𝑚

|

2

2𝜋𝑡𝛿(𝜔_𝑛𝑔− 𝜔). (2.1.23)

By the same argument, introduce the density of state n, ρ(ω_ng), again. we get

𝑝_𝑛⁽²⁾ = |∑ 𝜇_𝑛𝑔𝜇_𝑚𝑔𝐸² ħ²(𝜔_𝑚𝑔− 𝜔)

𝑚

|

2

2𝜋𝑡𝜌(𝜔_𝑛𝑔= 2𝜔). (2.1.24)

Thus the transition rate of two photon absorption process is

𝑅_𝑛𝑔⁽²⁾ =𝑝_𝑚⁽¹⁾

𝑡 = |∑ 𝜇_𝑛𝑔𝜇_𝑚𝑔𝐸² ħ²(𝜔_𝑚𝑔 − 𝜔)

𝑚

|

2

2𝜋ρ(ω_ng= 2𝜔). (2.1.25)

From Eq. (2.1.19), the transition rate of the single photon absorption is proportional to |𝐸|², the intensity of the excitation light source 𝐼. While the Eq. (2.1.25) shows that the transition rate of the two-photon absorption is proportional to |𝐸|⁴ , the square of the light source intensity 𝐼². When the excitation light is focused by the objectives, the two photon excitation occurs only in the vicinity of the focal point. Since the fast decay of the term 𝐼², outside the focal volume the intensity of the excitation light is not high enough for the two photon absorption. In contrast, the single photon absorption will happen in the excitation light cone as showed in Figure (2.1.3).

(15)

To better understand the excitation volume, we examine the spatial distribution of the laser intensity in the vicinity of the focal point of the beam. There are four equations below that we will use[10].

First we assume our laser beam is a Gaussian beam which propagates along z axis and is cylindrically symmetric about z axis. The intensity of the laser is given by

𝐼(𝑟, 𝑧) = 𝐼₀(𝑧)𝑒

−2𝑟²

𝑤(𝑧)². (2.1.26)

where 𝑟 is the distance from z axis, 𝐼₀(𝑧) is the intensity along the z axis, and w(z) is the beam width along the z axis. The intensity is related to the power P per unit area by,

𝐼₀(𝑧) = 2𝑃

𝜋𝑤(𝑧)². (2.1.27)

The factor of two in equation above comes from the integration of Eq. (2.1.26) over all angles and from zero to infinity in 𝒓 to ensure the conservation of photon flux as the beam reshape. The beam radius is related to the focal location as,

𝑤(𝑧)² = 𝑤₀²[1 + ( 𝜆𝑧

𝜋𝑤₀²)²] (2.1.28)

where 𝜆 is the wave length of the laser. 𝑤₀ is the beam radius at the focal waist and it can be related to the system as

𝑤₀ = 𝜆𝑓

𝜋𝑤_𝐿 (2.1.29)

where 𝑤_𝐿 is the beam radius at the system lens and 𝑓 is the focal length of the lens.

By combining the Eq. (2.1.26), Eq. (2.1.27) and (2.1.28), we get

𝐼(𝑟, 𝑧) 𝐼_𝑚𝑎𝑥 =

𝑒𝑥𝑝(−2𝑟²/𝑤₀²[1 + ( 𝜆𝑧 𝜋𝑤₀²)²]) [1 + ( 𝜆𝑧

𝜋𝑤₀²)²]

(2.1.27)

(16)

z r

where 𝐼_𝑚𝑎𝑥 is 𝐼₀(0). Then we can plot the contours of ^{𝐼(𝑟,𝑧)}

𝐼_𝑚𝑎𝑥 and (^{𝐼(𝑟,𝑧)}

𝐼_𝑚𝑎𝑥)², Figure (2.1.4), (2.1.5)

Figure 2.1.5 the contour of (^{𝐼(𝑟,𝑧)}

𝐼𝑚𝑎𝑥)² z r

Figure 2.1.4 the contour of ^{𝐼(𝑟,𝑧)} r

𝐼𝑚𝑎𝑥

(17)

Due to the small excitation volume, the photo-bleaching is confined to the vicinity of the focal point in the two-photon excitation process. In addition, typically the two-photon microscopy uses the near infrared light which avoids the main absorption spectrum of biological tissue and hence will has better penetration to the deeper location of the tissue

2.2 principles of optical microscopy

2.2.1 mode-locking

To achieve two-photon excitation which is used with sensible biological sample, the ultrashort laser pulses are needed to provide enough intensity without the high pulse energy. Short pulses

generation requires broad bandwidth which means multiple longitudinal modes corresponding to the different wave length. However, random multimode is not enough for the pulse generation.

Mode-locking is required. In the following we will derive the time-domain characteristics of a multimode laser. The electric field of a multimode laser is

𝑒(𝑧, 𝑡) = 𝑅𝑒{∑ 𝐸_𝑚𝑒^𝑗(𝜔^𝑚^𝑡−𝑘^𝑚^𝑧+𝜑^𝑚⁾

𝑚

} (2.2.1)

where

𝜔_𝑚= 𝜔₀+ 𝑚∆𝜔 = 𝜔₀+2𝑚𝜋𝑐

𝐿 𝑎𝑛𝑑 𝑘_𝑚 = 𝜔_𝑚/𝑐 (2.2.2) With the Eq. (2.2.2), Eq. (2.2.1) can be rewritten as

𝑒(𝑧, 𝑡) = 𝑅𝑒 {𝑒^𝑗𝜔⁰^(𝑡−^𝑧^𝑐⁾∑ 𝐸_𝑚𝑒^{𝑗[𝑚∆𝜔(𝑡−}^𝑧^𝑐^)+𝜑^𝑚^]

𝑚

}

= 𝑅𝑒 {𝑎(𝑡 − 𝑧/𝑐)𝑒^𝑗𝜔⁰^(𝑡−^𝑧^𝑐⁾}

(2.2.3)

(18)

where 𝑎 (𝑡 −^𝑧

𝑐) = ∑ 𝐸_𝑚 _𝑚𝑒^{𝑗[𝑚∆𝜔(𝑡−}^𝑧^𝑐^)+𝜑^𝑚^].

Thus, the electric field can be regarded as a combination of a complex envelope function 𝑎 (𝑡 −^𝑧

𝑐) and the optical carrier 𝑒^𝑗𝜔⁰^(𝑡−^𝑧^𝑐⁾. In the laser cavity, both envelope and carrier functions travel at the speed of light and the time for the light taking one round trip around the resonator is

𝜏 =2𝑙 𝑐 = 2𝜋

∆𝜔 (2.2.4)

where 𝑙 is distance of the both sides of cavity and 𝜏 is also the time period of two pulses. To know more about the shape of the envelop we should deternine the mode amplitudes 𝐸_𝑚 and phases 𝜑_𝑚. For simplicity, we assume that there are N oscillating modes with equal amplitudes 𝐸₀ and with the phases identically zero. The process that making the modes held with fixed relative phases is known as mode-locking. The Eq. (2.2.3) becomes

𝑒(𝑧, 𝑡) = 𝑅𝑒 {𝐸₀𝑒^𝑗𝜔⁰^(𝑡−^𝑧^𝑐⁾ ∑ 𝑒^{𝑗[𝑚∆𝜔(𝑡−}^𝑧^𝑐^)]

(𝑁−1)/2

−(𝑁−1)/2

} (2.2.5)

Substitute m^′= m + (N − 1)/2

𝑒(𝑧, 𝑡) = 𝑅𝑒 {𝐸₀𝑒^𝑗𝜔⁰^(𝑡−^𝑧^𝑐^{)−𝑗∆𝜔(}^𝑁−1² ^)(𝑡−^𝑧^𝑐⁾∑ 𝑒^{𝑗[𝑚∆𝜔(𝑡−}^𝑧^𝑐^)]

𝑁−1

0

}

= 𝑅𝑒 {𝐸₀𝑒^𝑗𝜔⁰^(𝑡−^𝑧^𝑐^{)−𝑗∆𝜔(}^𝑁−1² ^)(𝑡−^𝑧^𝑐⁾1 − 𝑒^{𝑗[𝑁∆𝜔(𝑡−}^𝑧^𝑐^)]

1 − 𝑒^{𝑗∆𝜔(𝑡−}^𝑧^𝑐⁾ }

= 𝑅𝑒 {𝐸₀𝑒^𝑗𝜔⁰^(𝑡−^𝑧^𝑐⁾sin (𝑁∆𝜔/2) (𝑡 −𝑧 𝑐)) sin (∆𝜔

2 ) (𝑡 − 𝑧 𝑐)

}

(2.2.6)

The laser intensity, I, averaged over an optical cycle, is proportional to |𝑒(𝑧, 𝑡)|². At z=0, the intensity is

(19)

𝐼(𝑡)~|𝐸₀|²sin ²(𝑁∆𝜔𝑡/2) sin² (∆𝜔𝑡

2 )

(2.2.5)

The pulse duration is approximately determined by the number of modes, ∆𝑡 = ^2π

N∆𝜔= 1/𝜈.

Equivalently, the pulse width is equal to the inverse of the laser bandwidth. The peak intensity is proportional to N². Compare to the intensity of random mode phase laser which is proportional to N, the peak intensity has a factor 2 enhancement.

In the real situation, not only the mode phase is modulated, the mode amplitude would also be modulated during each round trip in the cavity. The pulse is shaped through the process. Modulation plays a key role in initiating and maintaining mode-locked laser operation.

2.2.2 Light intensity distribution near focus[11]

When the incoming ray focused by the lens, due to the diffraction of the light, the light cannot be totally converged to the focal point. The intensity distribution will appear in the vicinity of the focal point.

According to the Huygens-Fresnel principle, every point in the wavefront can be considered as an independent point source and propagate outward as a spherical wave. To get the intensity of a point,

t t

~N

~N²

Intensity Intensity Intensity

Figure 2.2.1 (a) Mode-locked laser with constant mode phase; (b) laser with random mode phase

(a) (b) a

(20)

one should sum the amplitude created by every source on the wavefront. In our situation which is an incident ray with a beam radius 𝒂 travelling parallel to the principal axis of the lens and then converging to the focal point, we assume that the converging process can be described as a

monochromatic spherical wave emerging from a circular aperture and converging toward the axial focal point showed as the following picture

The spherical wave propagating from a circular aperture with the diameter 2a (a>> 𝜆) and converge to the point O which is f distant from the center of aperture. The point P near the focus O is the position of observation specified by a vector 𝑅⃑ . We also assume that the radius of the aperture and

|𝑅⃑ | are both small compared to the radius of the wavefront W at the aperture and the amplitude of incident wavefront W is A/f. If a point Q is on the wavefront W, we determine that the distance

𝑅⃑

z 2a

W

C f

Q s

f𝑞

P z Figure 2.2.2 the incident ray converged by the lens

Figure 2.2.3 the spherical wave converging to O

O z

x z y

z

(21)

between Q and P is s and a vector f𝑞 is point from O to Q. By the Huygens-Fresnel principle, we can express the amplitude of the wave at point P as

𝑈(𝑃) = −𝑖 𝜆

𝐴𝑒^{−𝑖𝑘𝑓}

𝑓 ∬𝑒^𝑖𝑘𝑠

𝑠 𝑑𝑠. (2.2.6)

where 𝜆 is the wavelength of the wave. Consider the point P is near axis, hence approximately

s − f = −𝑞 ∙ 𝑅⃑ (2.2.7)

The Eq. (2.2.6) can be rewritten as

𝑈(𝑃) = −𝑖

𝜆𝐴 ∬ 𝑒^{−𝑖𝑘𝑞⃑ ∙𝑅}^⃑𝑑𝛺 (2.2.8)

where 𝑑𝛺 is the unit solid angle of the element Q. We take the origin of the Cartesian coordinate at O. The coordinates of points P and Q are (𝑥, 𝑦, 𝑧) and (𝜉, 𝜂, 𝜁). Using the polar coordinate on the x-y plane, rewrite the coordinates of P and Q as

{𝑥 = 𝑟 sin 𝜑 , 𝜉 = 𝑎𝜌sinθ

𝑦 = 𝑟 cos 𝜑 , 𝜂 = 𝑎𝜌𝑐𝑜𝑠𝜃 (2. 2.9)

𝜁 = −√𝑓²− 𝑎²𝜌² = −𝑓[1 −1 2

𝑎²𝜌²

𝑓² + ⋯ ] (2.2.10)

where a is the aperture radius and 𝜌 is a parameter which denote the distance of point Q to the center of the aperture in the unit of a. Furthermore, we get

𝑞 ∙ 𝑅⃑ =𝑥𝜉 + 𝑦𝜂 + 𝑧𝜁

𝑓 =𝑎𝜌 cos(𝜃 − 𝜑)

𝑓 − 𝑧[1 −1

2 𝑎²𝜌²

𝑓² + ⋯ ] (2.2.11) 𝑑𝛺 =𝑑𝑠

𝑓² =𝑎²𝜌𝑑𝜌𝑑𝜃

𝑓² ,0 < 𝜌 < 1, 0 < 𝜃 < 2𝜋 (2.2.12) To simplify the calculation, we introduce two dimensionless parameter u and v defined as

(22)

𝑢 =2𝜋 𝜆 (𝑎

𝑓)²𝑧 ; 𝑣 =2𝜋 𝜆

𝑎

𝑓𝑟 (2.2.13)

Substituting u and v to express k𝑞 ∙ 𝑅⃑ , we get

𝑘𝑞 ∙ 𝑅⃑ = 𝑣𝜌 cos(𝜃 − 𝜑) − (𝑎 𝑓)

2

𝑢 +1

2𝑢𝜌² (2.2.14)

The amplitude at the point P becomes

𝑈(𝑃) = − 𝑖𝑎² 𝜆𝑓²𝐴𝑒^𝑖(

𝑎 𝑓)²𝑢

∫ ∫ 𝑒−𝑖[𝑣𝜌 cos(𝜃−𝜑)−+1 2𝑢𝜌²] 2𝜋

0 1

0

𝜌𝑑𝜌𝑑𝜃 (2.2.15)

Use the zero order Bessel function

𝐽₀(𝑥) = 1

2𝜋∫ 𝑒^{𝑖𝑥𝑐𝑜𝑠(𝜗)}

2𝜋

0

𝑑𝜗 = 𝐽₀(−𝑥) (2.2.16)

Thus, we get

𝑈(𝑃) = −2𝜋𝑖𝑎² 𝜆𝑓² 𝐴𝑒^𝑖(

𝑎 𝑓)²𝑢

∫ 𝐽₀(𝑣𝜌)𝑒⁻²^𝑖^𝑢𝜌²

1

0

𝜌𝑑𝜌 (2.2.17)

Therefor we get the intensity distribution 𝐼(𝑢, 𝑣) near the focal point

𝐼(𝑢, 𝑣) = |𝑈(𝑃)|² = |2 ∫ 𝐽₀(𝑣𝜌)𝑒⁻²^𝑖^𝑢𝜌²

1

0

𝜌𝑑𝜌|

2

𝐼₀ (2.2.18)

where 𝐼₀ = (^𝜋𝑎²

𝜆𝑓²𝐴)²

2.2.3 Resolution

The numerical aperture (NA) is one of the important properties of lens in microscopy and is defined by NA=sinθ where n is the index of refraction of the medium and θ is the maximal half-angle

(23)

of the cone of light that can enter or exit the lens(figure 2.2.4). The optical resolution is related to the numerical aperture and the light intensity distribution discussed in the previous section.

On the focal plane, z=0(u=0), the intensity distribution is

𝐼(0, 𝑣) = |2 ∫ 𝐽₀(𝑣𝜌)

1

0

𝜌𝑑𝜌|

2

𝐼₀ (2.2.19)

Use the relation of Bessel function, 𝑑

𝑑𝑥[𝑥^𝑛+1𝐽_𝑛+1(𝑥)] = 𝑥^𝑛+1𝐽_𝑛(𝑥) (2.2.20) The Eq. (2.) becomes

𝐼(0, 𝑣) = (2𝐽₁(𝑣)

𝑣 )²𝐼₀ (2.2.21)

When 𝑣 increase from 0, we find that the situation, 𝐽₁(𝑣) = 0 , first appear at 𝑣 = 3.83, and

𝑣 =2𝜋 𝜆

𝑎

𝑓𝑟 = 3.83, → 2𝑟 ≈ 1.22𝜆𝑓

𝑎 (2.2.22)

Since ^𝑎

𝑓≈ NA, the diameter D of the first-order minimum of the Airy disk is

𝐷 =1.22𝜆

𝑁𝐴 (2.2.23)

According to the Rayleigh criterion, two points can just be distinguished when the maximum intensity of the Airy disk falls on the first-order minimum of the other one. Hence the optical resolution is just

𝑟 ≡0.61𝜆

𝑁𝐴 (2.2.24)

In the similar way, we can also discuss the resolution along the principal axis of the lens. Consider the situation 𝑣 = 0. The intensity distribution becomes

(24)

𝐼(𝑢, 0) = |2 ∫ 𝑒⁻²^𝑖^𝑢𝜌²

1

0

𝜌𝑑𝜌|

2

𝐼₀ = [sin (𝑢/4)

𝑢/4 ]𝐼₀ (2.2.25)

The first 𝐼(𝑢, 0) = 0 happens when 𝑢 = ±4𝜋. By the definition of 𝑢, we get 𝑢 =2𝜋

𝜆 (𝑎

𝑓)²𝑧, → 𝑧 = 2𝜆 (𝑎

𝑓)² ≈ 2𝜆

(𝑁𝐴)² (2.2.26)

From Eq. (2.2.24) and Eq. (2.2.26), we find that the larger the NA is or the smaller the wavelength is, the better resolution we get.

θ S Fluor 20X/0.75

∞/0.17 WD 1.0

Figure 2.2.4 the objective with NA=0.75

(25)

Chapter 3. material and methods

3.1 Laser and microscope

The excitation source used in our multiphoton microscope is generated by the tunable

titanium-sapphire laser (Tsunami, Spectra-Physics) pumped by a diode-pumped laser (Millennia X, Spectra-physics). The titanium-sapphire laser is tuned to satisfy the two-photon excitation condition.

The wavelength of the laser is 780 nm and the laser is pulsed with approximate 100 fs pulse width and at the repetition rate of 80 Mhz. The advantage of using the wavelength near-infrared is the reduction of the scattering and absorption when light passing through the biological sample.

To adjust the power of the laser, a half wave plate followed by a linear polarizer is set on the light path. By rotating the half wave with touching the linear polarizer, we can adjust the power and keep the polarization. Scanning of the laser beam on the focal plane is handled by a set of

galvanometer-driven mirrors (Model 6210, Cambridge Technology, Cambridge, MA). One mirror rotates along the horizontal axis and the other rotates along the vertical axis. By the rotation of the mirrors, plane scanning is achieved. We also called the scanning system x-y scanner. The x-y scanner is controlled by the computer and by each scan we get a frame of plane image at specific depth. There are several scanning modes which we could alter for different sizes of scanning plane, resolution of a frame image (pixels/frame), scanning speed (pixels/second). After the x-y scanner, the beam is directed into a microscope. The objective (20X, S Flour, NA 0.75, Nikon) is used to focus the laser beam on the sample. The sample is placed on a x-y-z movable stage which is controlled and located by the computer. By moving the stage in the z-direction with the sample fixed on, we can get 3-dimesional information of the sample. The emitted fluorescence signal excited by the laser will collected by the same objective. The collected signal then passes through the dichroic mirror and enters the detection box called PMT box by us. In the PMT box, the signal

(26)

will be separated into three detection channels by a set of dichroic mirrors (555DCLP, 495DCXR, Chroma Technology, Rockingham, VT) and additional band-pass filters (HQ460/50,

HQ525/50,HQ610/75, Chroma Technology). The signals are finally registered by the

photomultiplier tubes (R7400P, Hamamatsu, Japan) in the single photon counting mode. The output from the PMT are processed by a discriminator and transferred to the computer.

Two-photon microscope

Polarizer

Sampl XY scanner

460/50 610/75

Dichroic mirror 1/2 wave

Ti-sapphire laser

DM 555 525/50 DM 495

PMT box PMT PMT

Objective 20X

Figure 3.1.1 Laser and microscope system

(27)

3.2 Stain

DAPI, (4',6-diamidino-2-phenylindole) was first synthesized in Otto Dann’s laboratory at Erlangen and the fluorescence enhance when attached to DNA was found by Williamson and Fennell in 1975.

DAPI is a blue-fluorescence stain which preferentially binds to the double-stranded DNA which is especially rich in adenine-thymine sequences. The complex formed by the binding of DAPI and DNA produces approximately 20-fold fluorescent enhancement. The maximum excitation wave length is 358nm and the maximum emission wave length is 461nm for DAPI bound to DNA. DAPI also binds RNA. However the fluorescence is much weaker compared to the fluorescence of

DAPI-DNA complex. The wavelength of the DAPI-RNA complex also shifts from ~460nm yielded by DAPI-DNA complex to ~500nm.

When using DAPI for the quantitative DNA measurement, there are some points which is worth noting. DAPI is a base specific stain which has the high affinity A-T rich DNA sequence. Hence, the fluorescence intensity depends not only on the amount of DNA but also the content of the A-T base of the DNA. Moreover, the degree of chromatin condensation which depends on the method of cell preparation also affects the amount of the DAPI bound in the cell. For example, washing cells with 0.1N hydrochloric acid to remove histone proteins approximately doubles the amount of DAPI bound to DNA[12].

DAPI can be used on the live or fixed cells. However the efficiency of DAPI passing through the live cells is less than the fixed cells. The effectiveness of the stain is lower. DAPI is also popular in multicolor fluorescent techniques. Due to the blue fluorescence, DAPI has a vivid contrast to green, yellow, or red fluorescent probes of other structures.

(28)

Figure 3.2.1 chemical formula of DAPI

Figure 3.2.2 DAPI bind to DNA

Figure 3.2.3 Excitation and emission spectrum of DAPI

(29)

3.3 cell line CL1[14]

CL1 was the human lung cancer cell line established from a 64-yr-old man with a poorly

differentiated adenocarcinoma. By the xenograft experiment which demonstrates tumor formation, it shows that CL1 cell line still keep the characteristics of adenocarcinoma even if the cells have been cloned and passed on for more than 60 generations. To better understand the mechanisms underlying invasiveness and metastasis, five sublines (CL1-1, CL1-2, CL1-3, CL1-4 and CL1-5) with progressive invasiveness and metastatic capabilities were obtained through the in vitro selection process. The selection process is a loop that from CL1-0, CL1-1 is selected and CL1-2 is selected from CL1-1, and so on. The Transwell plates which have the polycarbonate membranes containing 8μm pores and coated with Matrigel are used for the selection of invasiveness. Cells were resuspended in RPMI, the cell culture medium, and seeded into the wells. After the incubation for 72 hours at 37^oC, remove the well and harvest the cells that migrated through the membranes and attached to the lower-chamber compartments. One round of the invasiveness selection is finished. We can culture the cells and expand them for next round selection.

By the similar way, the experiment of the invasion activity can be conducted. The result showed that the invasive potential of CL1-5 increased by 4- to 6-fold compared to CL1-0.

membrane

Cells CL1-0

Cells invade through the membrane 72 h

(30)

3.4 Preparation for the sample Cell culture

The cells we used in our experiment are CL1-0 and CL1-5 cell lines established by Chu et al. and provided by professor Huei-Wen Chen (National Taiwan University College of Medicine Graduate institute of Toxicology).

The cell lines are cultured in RPMI 1640 media (Invitrogen, Carlsbad, CA, USA) supplemented with 10% fetal bovine serum (GIBCO, Carlsbad, CA, USA) and 1% antibiotics (GIBCO), and maintained in an incubator with humidified atmosphere of 5% CO2 and 95% air at 37^oC. Every two days, to avoid the cell density becoming too high, we will split the cells in one cell culture flask into three new flasks and renew the cell culture medium.

Staining

The samples were stained by DAPI and CellTracker^TM Orange CMTMR Dye (Thermo Fisher Scientific). DAPI has blue fluorescence and CellTracker is orange. Two different fluorescent signals can be separated by the PMT box introduced in the section 4.1(DAPI excitation:358nm

emission:461nm, CellTracker excitation:541nm emission:565nm). In this thesis, only the DAPI signals will be considered for analysis. The CellTracker signal is adapted by another thesis.

Before staining, the cells need to be seeded on the cover slip which is coated by the 3mg/mL type I collagen diluted 20 times with Phosphate-buffered saline (PBS). The seeded cell concentration is 2X10⁵ cells/mL calculated with the counting chamber. The live cells are first stained with CellTracker of which the concentration is 0.5μM in serum-free medium in the incubator for 30 minutes. Then the cells are fixed with the 3.7% formalin solution before the staining of 0.5mg/L DAPI diluted by PBS from stock solution. To stain the DAPI, we drop 100μL DAPI on the

(31)

Parafilm and place the cover slip on the drop with the side seeded the cells. After staining, we add 10μL PBS solution between cover slip and slide, and mount them with nail polish coated around the cover slip.

3.5 Setting of microscope and operation procedure

The images of the samples are scanned by the system described in 3.1. The setting of the

microscope is in scan mode large FOV(340μm x 340μm scanning plane), scanning speed 20kHz, resolution of the image (512 x 512), and scanning along z-direction for 30 layers that the distance between each layer is 1μm. The highest intensity appears at the middle layer. After the z-scan for the area is finished, we can move the stage in x-y plane for the scan of other areas of the samples.

The laser intensity can be adjusted by rotate the half wave plate and its value is measured by a portable power meter placed after the polarizer. The excitation intensity for the z-scan is about 300mW.

Figure 3.4.1 the method to stain DAPI

Figure 3.4.2 complete sample

(32)

Chapter 4 Result and discussion

4.1 Image processing

The signals acquired by the microscope are separated to different channel by the wave bands, 610/75(nm), 525/50(nm), 460/50(nm). We take the images of DAPI in channel 525/50 for analysis since it shows the highest intensity in our system (signal of DAPI in 525/50 is a little stronger than in 460/50, and there is no DAPI signal in 610/75). The images are processed with a public domain program, ImageJ. To quantify DNA content of cells, we need to segment nucleus from the image and acquired the integrated intensity of each nucleus (the sum of the intensity value of each pixel in the nucleus). The following is the procedure to process the image and is all done with ImageJ.

With the 30 stacks of images, first sum their intensity into one image (image/stacks/z project/sum slices). To segment the nucleus, the image needs to be adjusted the threshold and transformed to

binary form. There are several methods which have different outcomes and may affect the result of the nucleus segmentation. The method used in this thesis is in image/adjust/auto threshold/method Huang. After autothreshold, use the function, watershed (process/binary/watershed), which can

separate two connected nuclei and analyze particles (analyze/analyze particles) for nucleus

segmentation. In the function, analyze particles, one can set the limit of minimum size (pixel^2) of the nucleus and exclude all small noise (set the minimum size limit, 150). The selection information is then stored in the ROI manger. To acquire the data, use measure in ROI manager on the original image.

(33)

Figure 4.1.1 Image processing and data acquisition

(34)

4.2 z-stack summation image

Considering the nucleus in the cell attached on the cover slip is not a completely flat object, the z-scan is used to quantify the total DNA content of the nucleus. The Figure (4.2.1) shows the integrated intensity of each cell from two image that one is single layer which has the highest intensity in the z-stacks and the other is the image of which the intensity is the summation of the z-stacks.

The spatial distributions along the z-axis of the nuclei on the cover slip are similar that most of them have the highest intensity in the same layer and the effect of z-summation simply is the linear intensity enhancement. The advantage of the enhancement is that the cell segmentation becomes more precise and it reflects the total DNA content. However it takes time which is proportional to the number of the scanning stacks. If the analysis only needs DNA content, the scan of z-stacks is unnecessary.

y = 1.8391x + 751.78 R² = 0.9799

0 20000 40000 60000 80000 100000 120000 140000

0 10000 20000 30000 40000 50000 60000 70000 80000

z-stacks summation

single layer

integrated intensity of each nucleus

Figure 4.2.1 integrated intensity of cells in z-stack summation image – in highest intensity single layer image

(35)

There is also a problem that should be noticed in this experiment. The enhancement ratio which defined as the slope in the Figure (4.2.1) is not a stable value. With the condition that the same sample and the successive scans, the enhancement ratios of the different z-stacks summation vary approximately from 2 to 9. The difference may be caused by the different overlapping volume of the two-photon excitation volume from different layers which is related to the laser intensity.

4.3 Comparison of CL1-0 and CL1-5

According to some measurements of DNA content by flow cytometry, the DNA content distribution of a group of the same species cells can be divide into three divisions which represent the different phases of the cycle, G1 phase, S phase and G2/M phase[13]. G1 phase is from the end of the mitosis until the DNA beginning to synthesis for cell duplication. When DNA is synthesizing until double content, the cell cycle is in the S phases. In G2/M phase, DNA in the cells remains double in G2 phase then entering the mitosis phase. Generally, most of the cells are in G1 phase that shows the largest peak in the distribution histogram. G2/M phase also has a smaller peak. However the nucleus in the late mitosis stage may be recognized as two objectives by the cell segmentation algorithm, some of the G2/M population will be classified into G1 phase.

As mention in the discussion 4.2 that laser intensity may not be stable, the fluorescence signal should be normalized before combining all the data together. To normalize, all integrated intensities of the cell in same image are divided by the median integrated intensity which has high probability in G1 phase. Then the peak of the normalized distribution will appear at the integrated intensity which has value equal to 1. There are 1324 CL1-0 nuclei and 984 CL1-5 nuclei for the analysis. The integrated intensity distribution histogram of CL1-0 and CL1-5 are shown below.

(36)

0 100 200 300

0.2 0.6 1 1.4 1.8 2.2 2.6 3 3.4 3.8

cell number

nucleus integrated density (normalized with the median intensity)

CL1-5

0 100 200 300

0.2 0.6 1 1.4 1.8 2.2 2.6 3 3.4 3.8

cell number

nucleus integrated density (normalized with median intensity)

CL1-0

The two distributions seem similar that both of them have the G1 peaks and without the G2/M peak.

Trying to differentiate two cell lines by the integrated intensity data, the average value and the standard deviation are calculated. The mean and standard deviation of CL1-0 and CL1-5 are (1.085, 0.514) and (1.163, 0.715). Though the difference looks small, that the both larger values of CL1-5 seems to suggest that CL1-5 has higher DNA content abnormality. Because of DNA

synthesis, the signals of normal and abnormal DNA content are merged in the distribution

histogram. To avoid the problem, another attempt to find the difference of cell lines is to count the nuclei of which the integrated intensity is larger than 2.6. There are 1.1% nuclei of CL1-0 and 4.5%

nuclei of CL1-5 satisfying the condition that also suggest the higher DNA abnormality of CL1-5.

Use the same method on the analysis of nuclei area. We can see that CL1-0 nuclei are larger than the CL1-5 nuclei and the dispersion of the CL1-0 area is also larger. From the Figure 4.3.3, it shows that the integrated intensity increase with the area. However the distribution histograms of

integrated intensity seem to be different from the distributions of the area. We also wonder how the distribution of average intensity is.

Figure 4.3.1 Integrated intensity distribution histogram of CL1-0 and CL1-5

(37)

0 1 2 3 4 5 6

0 200 400 600 800 1000

integrated intensity

area (μm^2)

CL1-5

0 1 2 3 4 5 6

0 200 400 600 800 1000

integrated intensity

area (μm^2)

CL1-0

0 50 100 150

cell number

area (μm^2)

CL1-0

0 50 100 150

cell number

area (μm^2)

CL1-5

To get the average intensity, integrated intensity of each nucleus is divided by its area. The origin average intensity is also normalized by multiplying the same value which makes the mean of average intensity equal to the mean of integrated intensity. The distribution of the average intensity shows that most of the nuclei of CL1-0 have the value 1 average intensity. But the nuclei of CL1-5 have the similar counts from average intensity 1 to 1.4 that may imply the disperse DNA density in the CL1-5 nuclei.

Figure 4.3.2 area distribution histogram of CL1-0 and CL1-5

Figure 4.3.3 integrated intensity – area diagram of CL1-0 and CL1-5

(38)

0 100 200 300 400 500

0.2 0.6 1 1.4 1.8 2.2 2.6 3 3.4 3.8

cell number

nucleus average density (integrated intensity/area)

CL1-0

0 100 200 300

0.2 0.6 1 1.4 1.8 2.2 2.6 3 3.4 3.8

cell number

nucleus average intensity (integrated intensity/area)

CL1-5

Table 4.4.1 Statistical value of some quantities of CL1-0 and CL1-5 Conclusion

Though the difference of the two cell lines can be recognize from the histograms, it is more convenient to have a value as the index of two cell lines furthermore the invasive ability. As the table showed below, the percentage difference between the standard deviation of the integrated density is the largest. And it may have the potential to measure the DNA content abnormality and the invasive ability of cells.

CL1-0 CL1-5

Average Standard deviation

Coefficient of variance

Average Standard deviation

Coefficient of variance Integrated

intensity 1.085 0.514 0.474 1.163 0.715 0.615

Area

(μm^2) 282.2 125.3 0.444 258.6 112.3 0.434

Average intensity

1.085 0.408 0.376 1.163 0.380 0.327

Figure 4.3.4 Nucleus average density distribution histogram of CL1-0 and CL1-5

(39)

References

[1] 國家衛生福利部, “民國 103 年主要死因分析,” (2016)

[2] Siegel, R., et al. (2013). "Cancer statistics, 2013." CA: A Cancer Journal for Clinicians 63(1):

11-30.

[3] American Cancer Society

[4] Team, T. N. L. S. T. R. (2011). "Reduced Lung-Cancer Mortality with Low-Dose Computed Tomographic Screening." New England Journal of Medicine 365(5): 395-409.

[5] Dyszkiewicz, W., et al. (2000). "Prognostic significance of DNA ploidy in squamous cell lung carcinoma: is it really worth it?" The Annals of Thoracic Surgery 70(5): 1629-1633.

[6] Volm, M., et al. (1985). "DNA distribution in non-small-cell lung carcinomas and its relationship to clinical behavior." Cytometry 6(4): 348-356.

[7] Bunn PA, Krasnow St, Makuch RW, Schlam ML, Schechter GP: Flow cytometry analysis of DNA content of bone marrow cells in patients with plasma cell myeloma: Clinical implications.

Blood 59528-535, 1982.

[8] Choma, D., et al. (2001). "Aneuploidy and prognosis of non-small-cell lung cancer: a meta-analysis of published data." Br J Cancer 85(1): 14-22.

[9] Boyd, R. W., Nonlinear optics, 3^rd ed. (Academic, London, 2008)

[10] DeVoe, R. J., et al. (2003). Voxel shapes in two-photon microfabrication.

[11] Born, M & Wolf, E., Principles of optics : electromagnetic theory of propagation, interference and diffraction of light, 7th(expanded) ed. (Cambridge University Press, Cambridge, UK; New York;, 1999)

[12] Kapuscinski, J. (1995). "DAPI: a DNA-specific fluorescent probe." Biotech Histochem 70(5):

220-233.

(40)

[13] Roukos, V., et al. "Cell cycle staging of individual cells by fluorescence microscopy."

(1750-2799 (Electronic)).

[14] Chu, Y.-W., et al. (1997). "Selection of Invasive and Metastatic Subpopulations from a Human Lung Adenocarcinoma Cell Line." American Journal of Respiratory Cell and Molecular Biology 17(3): 353-360.