YJSBI 6610

No. of Pages 9, Model 5G

5 September 2014 Journal of Structural Biology xxx (2014) xxx–xxx 1

Contents lists available at ScienceDirect

Journal of Structural Biology journal homepage: www.elsevier.com/locate/yjsbi 5 6

Single virus detection by means of atomic force microscopy in combination with advanced image analysis q

3 4 7

Q1

8

Thomas Bocklitz a,b,c,⇑, Evelyn Kämmer a,b,c, Stephan Stöckel a,b,c, Dana Cialla-May a,b,c, Karina Weber a,b,c, Roland Zell d, Volker Deckert a,b,c, Jürgen Popp a,b,c a

9 10 11 12

Institute of Physical Chemistry and Abbe Center of Photonics, Friedrich Schiller University Jena, Helmholtzweg 4, 07743 Jena, Germany InfectoGnostics Forschungscampus Jena e.V., Zentrum für Angewandte Forschung, Philosophenweg 7, 07743 Jena, Germany Leibniz Institute of Photonic Technology, Albert-Einstein-Strasse 9, 07745 Jena, Germany d Department of Virology and Antiviral Therapy, Jena University Hospital, Friedrich Schiller University Jena, Hans-Knöll-Strasse 2, 07745 Jena, Germany b c

13

a r t i c l e

1 2 5 6 16 17 18 19

i n f o

Article history: Received 21 August 2014 Accepted 25 August 2014 Available online xxxx

20 21 22 23 24 25

Keywords: Image analysis Statistical modeling Atomic force microscopy Virus detection

a b s t r a c t In the present contribution virions of five different virus species, namely Varicella-zoster virus, Porcine teschovirus, Tobacco mosaic virus, Coliphage M13 and Enterobacteria phage PsP3, are investigated using atomic force microscopy (AFM). From the resulting height images quantitative features like maximal height, area and volume of the viruses could be extracted and compared to reference values. Subsequently, these features were accompanied by image moments, which quantify the morphology of the virions. Both types of features could be utilized for an automatic discrimination of the five virus species. The accuracy of this classification model was 96.8%. Thus, a virus detection on a single-particle level using AFM images is possible. Due to the application of advanced image analysis the morphology could be quantified and used for further analysis. Here, an automatic recognition by means of a classification model could be achieved in a reliable and objective manner. Ó 2014 Published by Elsevier Inc.

27 28 29 30 31 32 33 34 35 36 37 38 39

40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57

1. Introduction Viruses are very small obligate intracellular parasites, which by definition share a common architecture: A DNA or RNA genome is surrounded by a protective protein coat (capsid). Despite this rather simple basic structure, viruses come in an amazing variety of shapes and sizes. Viruses can range in sizes from 20 nm (e.g. Q3 viruses of the family Nanoviridae) (Bressan and Watanabe, 2011) to 1000 nm (Pandoravirus) (Philippe et al., 2013). Different arrangements of the protein layer and the genetic information allow a classification of viruses with respect to their morphology: Spherical viruses often consist of polyhedral or icosahedral nucleocapsids (for example rhinoviruses), whereas cylindrical nucleocapsids in form of helical arrays of capsid proteins are wrapped around a helical filament of nucleic acids (e.g. Tobacco mosaic virus). Many viruses are furthermore surrounded by an outer envelope of lipids, e.g. the influenza viruses. Finally, viruses of a complex structure bear a combination of icosahedral and helical

q

Q2 Q1

T.B. and E.K. share main authorship, due to equal contribution. ⇑ Corresponding author at: Institute of Physical Chemistry and Abbe Center of Photonics, Friedrich Schiller University Jena, Helmholtzweg 4, 07743 Jena, Germany. E-mail address: [email protected] (T. Bocklitz).

shape and may have a complex outer wall or head–tail morphology like many bacteriophages. Since a vast number of viruses can cause infectious disease, a rapid diagnosis is the goal of modern virus detection allowing for fast clinical actions in infection control, prophylaxis or specific antiviral treatment. Influenza viruses, responsible for the annually recurring flu season worldwide, are among the best examples of a global-scale demand for virus diagnosis and surveillance (Lofgren et al., 2007). Culture-based systems for virus isolation have been the ‘gold standard’ in clinical virology for decades, but are nowadays complemented or even replaced by a variety of serological tests and molecular technologies lead by real-time PCR (Okame et al., 2007; Leland and Ginocchio, 2007). The isolation of viruses in culture is considered time-consuming, labor-intensive and lacks the sensitivity needed to appreciably impact on medical decisions (Hodinka and Kaiser, 2013). Culture based isolation is not applicable for so-called ‘non-cultivable’ viruses like hepatitis C (Lecuit and Eloit, 2013). However, molecular testing is not without drawback: commonly used laboratory-developed assays and analyte-specific reagents require expertise, time, special equipment and have to undergo extensive verification and validation. The need to adapt primers and probes on a regular basis to highly variable viruses still requires culture for, e.g. influenza surveillance (Hodinka and Kaiser, 2013; Ogilvie, 2001).

http://dx.doi.org/10.1016/j.jsb.2014.08.008 1047-8477/Ó 2014 Published by Elsevier Inc.

Q1 Please cite this article in press as: Bocklitz, T., et al. Single virus detection by means of atomic force microscopy in combination with advanced image analysis. J. Struct. Biol. (2014), http://dx.doi.org/10.1016/j.jsb.2014.08.008

58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81

YJSBI 6610

No. of Pages 9, Model 5G

5 September 2014 Q1 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126

2

T. Bocklitz et al. / Journal of Structural Biology xxx (2014) xxx–xxx

Beside the described indirect methods, direct methods for the virus diagnostics exist. Electron microscopy (EM) is considered as a rapid method to identify viruses, as it allows a direct examination of specimen (Schramlová et al., 2010; Hazelton and Gelderblom, 2003; Zechmann and Zellnig, 2009). However, the attribute ‘direct’ conceals the facts that the sample preparation can be complex and that the virus particles are heavily altered due to the sample preparation. For example, transmission EM requires heavy metal staining or shadowing to increase contrast. Also, a complete dehydration is usually necessary. Pleiomorphic viruses and those lacking architectural uniformity are inefficiently analyzed via cryo-EM (Kuznetsov and McPherson, 2011). Further microscopic techniques for virus detection are scanning probe microscopy (SPM) methods, e.g. scanning tunneling microscopy (STM) (Ikai et al., 1994; Imai et al., 1993; Guckenberger et al., 1994) or atomic force microscopy (AFM) (Kuznetsov et al., 2012; Chen et al., 2013; Allison et al., 2010; Kuznetsov et al., 2001; Santos and Castanho, 2004). SPM is a non-invasive, inexpensive technique, which can work with a minute sample volume. This is a great advantage especially for viruses, which are difficult to cultivate. SPM measurements result in a three dimensional image, i.e. the height of a virion (a virus particle outside a host cell) is obtained next to the lateral dimension. For STM investigation, samples should be conductive, which is difficult to accomplish for biological samples. In contrast, AFM allows for an imaging of non-conductive materials. The images recorded in such a way (EM, STM, AFM) can be analyzed using image analysis approaches, as the measured quantity can be converted to a gray scale. The method best suited for such isolated virion images is the image moment method (Flusser et al., 2009). This method translates the morphology or shape of the image into quantitative features, which subsequently can be statistically analyzed. In this publication we use the word ‘morphology’ strictly in a biological sense, as it is common for virologist. In contrast, the shape of the virion is the type of appearance, for example a rod-shape. The latter excludes absolute height and volume. This separation yields a more comprehensive description as it circumvents the problem that there is a different meaning of the term ‘morphology’ in mathematical and biological sense. In this work, the image moment method in combination with atomic force microscopy (AFM) is used to discriminate viruses on a single-particle level. Five virus species from a wide range of hosts were chosen: As a plant virus, the Tobacco mosaic virus was imaged by AFM and subsequently analyzed. The Coliphage M13 and the Enterobacteria phage PsP3 were chosen as representatives of

bacteriophages. Finally, the Varicella-zoster virus (VZV) and the Porcine teschovirus (PTV) are examples of viruses, which rely on vertebrates as hosts. As a first approach, quantitative features of the viruses such as height, volume and area are evaluated and discussed. Subsequently, these features are accompanied by image moments and an automation of the analysis for the chosen virus species could be achieved.

127

2. Material and methods

134

2.1. Sample preparation

135

The PsP3 bacteriophage suspension was purchased from the Leibniz Institute DSMZ (German Collection of Microorganisms and Cell Cultures) and diluted with distilled, sterile filtrated water in a ratio of 1:10. The M13K07 bacteriophage stock solution (New England Biolabs) was diluted in a ratio of 1:100 with distilled and sterile filtrated water. The plant virus Tobacco mosaic virus (TMV) was prepared as described in Cialla et al. (2009) and diluted in a ratio of 1:105 with distilled and sterile filtrated water. The Porcine teschovirus (PTV) type 1 strain Talfan was amplified in porcine kidney cells PK-15 cultured in Eagle minimal essential medium (EMEM). In T75 Roux flasks PK-15 monolayers were infected and incubated for approximately 48 h until complete lysis. Thereafter, the supernatant was centrifuged at 4000 rpm (3774g; Variofuge 3.0R) to remove detritus. After a further ultracentrifugation for 90 min at 100000g (30000 rpm, rotor SW60Ti), a pellet of PTV was obtained. The virus pellet was resuspended in 100 ll PBS. For the Varicella-zoster virus (VZV) the strain vOka was used and amplified in human embryonic lung fibroblasts (HELF). For VZV preparation, HELFs were cultured in EMEM complemented with 25 mM HEPES, 1% non-essential amino acids, 2 mM L-glutamin, 1 mg/100 ml Ciprofloxacin and 10% fetal bovine serum at 37 °C and 1% CO2. The supernatant of infected cells showing complete lysis was centrifuged for 15 min at 4000 rpm (3774g; Variofuge 3.0R) to remove cell detritus. The inactivation of the viruses was carried out by irradiation with ultraviolet light. The suspensions of VZV and PTV were directly irradiated with UV light (20 min) and twofold serial dilutions with PBS were stored at 70 °C. The inactivation (30 min) of bacteriophages and the plant virus was performed after the first drying of the virus solution on the cover slip. Cover slips (18  18 mm) consisting of glass were applied as substrate. In order to remove organic residues, the glass slides were placed in

136

Fig.1. A scheme of the AFM setup is illustrated. For the AFM analysis, the sample was mounted in the sample holder of the inverse microscope and positioned guided by an oil immersion objective. The sample was scanned by the AFM probe. The vertical and lateral deflection of the cantilever (AFM probe) was recorded by an IR lasers and a four quadrant photodiode. The cantilever is moved with a nanometer precision by a piezoelectric scanner. The detector and controller utilized the signal from the stage and the diode for feedback regulation during AFM imaging.

Q1 Please cite this article in press as: Bocklitz, T., et al. Single virus detection by means of atomic force microscopy in combination with advanced image analysis. J. Struct. Biol. (2014), http://dx.doi.org/10.1016/j.jsb.2014.08.008

128 129 130 131 132 133

137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167

YJSBI 6610

No. of Pages 9, Model 5G

5 September 2014 Q1

176

a mixture of nitric acid and hydrogen peroxide in a ratio of 3:1 for 2 h followed by washing two times with distilled water. The clean glass slides were vacuum-dried and stored under argon. A volume of 10 ll of the virus dilutions were dropped on a cleaned cover slip and dried under ambient air conditions. In order to remove salts and medium residues, the samples were washed with 100 ll distilled and sterile filtered water. After a second drying, the virus particles adsorbed on the cover slip were ready to be measured.

177

2.2. Atomic force microscopy – AFM

178

204

An inverse microscope (Horiba Jobin Yvon) combined with an oil immersion objective (PlanApo, Olympus, 60, NA 1.45) was utilized to move the AFM tip on particles in the sample drop on the cover slip. To record the AFM images a setup from NanoWizard atomic force microscopy (JPK Instruments AG) was applied. A rotated, monolithic silicon AFM probe from NanoAndMore (Tap190Al-G) with a rectangle cantilever and a symmetric tip shape was utilized. The AFM probe had a cantilever force constant of around 48 N/m, a resonance frequency of 190 kHz and a tip radius of less than 10 nm. The AFM probe was suitable for non-contact and tapping mode. Different types of AFM tips, an untreated tip and a tip evaporated with a silver layer, were used for the AFM measurements. Due to the usage of different tips, a higher variance at the imaging of the virus particles is achieved and can be taken as proof that the subsequent analytical steps remain unbiased by the applied AFM tips. In order to preserve the virus particles from damages and to protect the tip from contaminations, the AFM images were recorded by using tapping mode in air. A scheme of the experimental setup is shown in Fig. 1. The AFM cantilever was coated with an aluminum reflection layer. An infrared laser was adjusted to a four quadrant photo diode, which detects the response of the AFM tip during the measurements. While measuring the tip is scanned over the sample controlled by a piezoelectric scanner with an accuracy of 0.5 nm. Sixty to one hundred single viral particles of each virus species were recorded. In the following we will term an AFM image of the height ‘AFM image’.

205

2.3. Calculations

206

All calculations were carried out in GnuR (R Development Core Team, 2008). The used packages are ‘akima’ (Akima, 2009), ‘IM’ (Rajwa et al., 2013) and ‘MASS’ (Venables and Ripley, 2002). Parts of the calculations were parallelized using the ‘foreach’ package (Analytics and Weston, 2013). Before the analysis, the images were pre-treated. All recorded AFM images were pre-processed with the JPKSPM data processing software (version 4.2.52). A background correction was done by means of a histogram line fit, which subtracts a polynomial fit from each scanned line of the AFM image. Furthermore, an interpolation step was done in order to account for different spatial resolutions. The common spatial resolution was chosen to be 1 nm. Thereafter, a 3  3 mean filter was applied and a cropping to the region of interest was carried out. From these sub-scans certain characteristics including volume and the deciles of the height were calculated. The deciles are the data points, which split the height distribution in ten equal parts, together with the maximal and the minimal height. Therefore, the elevenths decile is the maximum height, while the first decile is the minimal height. After a thresholding the area could be determined as well (see Fig. 2 C). The threshold was set to 50% of the maximal height (the sixth decile) of the virus. The 2D-Pseudo-Zernike-Moments (Xia et al., 2007) are defined via the Zernike-polynomials V pq ðr; #Þ in polar coordinates (r; #) with respect to two parameters (p; q)

168 169 170 171 172 173 174 175

179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203

207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228

229 231

3

T. Bocklitz et al. / Journal of Structural Biology xxx (2014) xxx–xxx

V pq ðr; #Þ ¼ Rpq ðrÞ exp ði  q#Þ:

ð1Þ

The parameters are commonly termed repetition q and order p and the radial function is given by

233

234

Rpq ¼

pjqj X s¼0

ð1Þk ð2p þ 1  sÞ!  r ps : s!ðp  jqj  sÞ!ðp þ jqj þ 1  sÞ!

ð2Þ

In this equation s is a summation index. The so-defined function system is composed of orthogonal functions on the unit circle. Therefore an image Iðx; yÞ can be described with the PseudoZernike-Moments (PZ-moments) after a transformation of the x–y-coordinates to polar coordinates is carried out. The PZmoments read

Z pq ¼

Z Z p þ 1 2p

p

0

0

236 237 238 239 240 241 242

243 1

V pq ðr; #ÞIðr; #ÞIðr; #Þ

p ¼ 0; 1; 2;    ; 0 6 jqj 6 p:

 rdrd#; ð3Þ

245

The symbol ⁄ denotes complex conjugation. If a translation into the center of mass and a normalization onto the first centered image moment is carried out these moments are translational and scale invariant (Chong et al., 2003). The absolute value of these moments is, due to the properties of the PZ-polynomials, rotational invariant. In this article we term these values Pseudo-ZernikeInvariants (PZ-invariants) as it is done by the ‘IM’ package (Rajwa et al., 2013). For the analysis presented here the repetition and order were set to 25 and only the first 200 values were utilized for the analysis. The PZ-invariants together with volume, area and height deciles were calculated for each virion. A visualization of the workflow to extract characteristics of every virus AFM scan is given in Fig. 2. As the values differ strongly in their order of magnitude a principal-component-analysis (PCA) with centering and scaling to unit variance of every feature was applied. Subsequently, a dimension reduction to 50 PCs was carried out and a linear-discriminantanalysis (LDA) was applied to build up a classification model for the virus species based on the first 50 PC scores. A ten-fold crossvalidation was used in order the quantify the performance of the model.

246

3. Results and discussion

267

Prior the analysis the images were pre-treated as described earlier. This step is important for the comparability of the images as they were recorded with slightly different parameters (scan dimension, resolution etc.). The most crucial of these parameters is the spatial resolution. Therefore every image was interpolated to a common spatial resolution of 1 nm both in x and y. Thereafter, the influence of small variations was corrected by a 3  3 mean filtering. At the end of the pre-treatment step a cropping to the approximated virus region was carried out. This was done using manually selected marker points and a polygon was cut from the entire image. In order to fully automate the analysis demonstrated here, a segmentation approach has to be applied. This can be achieved by the methodology previously described in Kylberg et al. (2012) and by Medyukhina et al. (2013). As a proof-ofprinciple of an automated recognition of viruses was intended, this step was not implemented. Manually choosing the region of interest (sub-scan) allowed to cross-check and the analysis presented here is not corrupted by possible artifacts produced by an automatic segmentation. After the pre-treatment was finished, features from the virion sub-scan were extracted, which included PZ-invariants, area, volume and the height deciles (see Fig. 2 B). The latter are the nine height values that divide the sorted data into ten equal parts together with the minimal and maximal height. Every part represents a tenths of the height distribution inside the regions of interest. The area is calculated from a binary

268

Q1 Please cite this article in press as: Bocklitz, T., et al. Single virus detection by means of atomic force microscopy in combination with advanced image analysis. J. Struct. Biol. (2014), http://dx.doi.org/10.1016/j.jsb.2014.08.008

232

247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266

269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292

YJSBI 6610

No. of Pages 9, Model 5G

5 September 2014 Q1

4

T. Bocklitz et al. / Journal of Structural Biology xxx (2014) xxx–xxx 300

A

100

Classification Model and Evaluation

Height [nm]

200

0

Height distribution Volume Image Moments Image Invariants

Interpolation Smoothing Cropping

Area

C

B Thresholding

300

100

Height [nm]

200

500nm 0

Fig.2. The pre-treatment and feature extraction of a single VZV image is sketched. Every image was background corrected (A). Thereafter, every background corrected image is interpolated to the same lateral resolution and a mean filter is applied to remove noise. These images are cropped to the region of interest and form a sub-scan (B). via a threshold a binary image (C) is generated. The red color illustrates the virus region, while the blue color represents the background contribution. From the binary image the area of the virus can be calculated, while with the pre-treated image (B) the height deciles, the volume and the PZ-invariants are calculated. All extracted features are used for a classification model. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

304

version of the image, which is constructed by a thresholding. The sum of all the pixel above the threshold times the squared spatial step size of 1 nm (reddish region in Fig. 2 C) is utilized as estimated area. The volume is calculated in a similar manner by the summation of the height values inside the region of interest times the squared spatial step size. These features (volume, area, height deciles and PZ-invariants) are discussed in the following and used for the classification of five viruses under investigation. In the discussion volume, area and maximal height are compared with literature values. The literature values are measured mostly with EM, thus the virions were measured in dried state. This may affect the direct comparison.

305

3.1. Virus AFM images and features interpretation

306

Five different virus species were investigated, which have diverse hosts and belong to five different virus families. The latter determines the morphology of the viruses. For each virus species, several single viral particles were measured by means of atomic force microscopy (AFM). The distinct morphology of the five virus species is presented in Fig. 3. For every virus species a representative AFM images is displayed. The maximum height, volume, and area were determined for all sub-scans and these quantities are illustrated in boxplots (Fig. 4). The Varicella-zoster virus (VZV) is a well-known representative of the Herpesviridae family. VZV causes varicella (chickenpox) as the primary infection and, when reactivated, shingles (herpes zoster) in humans (Smith and Arvin, 2009). VZV consists of a linear double-stranded DNA, which is surrounded by a capsid. This capsid is composed of 162 capsomeres and has an icosahedral structure. It

293 294 295 296 297 298 299 300 301 302 303

307 308 309 310 311 312 313 314 315 316 317 318 319 320

is surrounded by an amorphous tegument and a lipid envelope, in which glycoproteins are embedded. In Fig. 3 A an exemplary AFM image of VZV is shown. The measured heights are represented by the false color scale on the right hand side of each AFM image. The red colors illustrate large and the blue colors small height values. The VZV in Fig. 3 A has an approximate maximum height of 200 nm, which is conform with the literature (Almeida et al., 1962; Harson and Grose, 1995). The VZV normally appears in a fried egg structure, which is presumably caused by the amorphous structure of the tegument. The AFM image of VZV (Fig. 3 A) clearly displays this fried egg structure: The ‘egg white’ is represented by green to yellow color values and surrounds the reddish colored ‘egg yolk’. In this analogy, the ‘egg white’ represents the tegument and the lipid envelope, which encloses the ‘egg yolk’, the icosahedral capsid of VZV. The Porcine teschovirus (PTV) is the smallest examined virus species with an icosahedral capsid. The icosahedral capsid surrounds a linear single-stranded RNA. PTV lacks any further envelope, therefore it is a non-enveloped virus. It infects pigs and causes the so-called Teschen disease, a nonsuppurative polioencephalomyelitis (Chiu et al., 2012). The PTV belongs to the family Picornaviridae. An exemplary AFM image of a PTV virion can be seen in Fig. 3 B. The roundish structure of the PTV is clearly visible and the maximal height of this virion is around 31.8 nm (see color scale in Fig. 3 B). The height is slightly higher than the range of 25–30 nm in diameter known from the literature (Yamada et al., 2014; Chiu et al., 2014). This is true for most of the measured PTV particles as can be seen in Fig. 4 A. Here, the maximum heights of all measured viral particles of the five different virus species are presented in form of a boxplot: The height in nanometers is plotted logarithmically against the virus species. The median maximal

Q1 Please cite this article in press as: Bocklitz, T., et al. Single virus detection by means of atomic force microscopy in combination with advanced image analysis. J. Struct. Biol. (2014), http://dx.doi.org/10.1016/j.jsb.2014.08.008

321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350

YJSBI 6610

No. of Pages 9, Model 5G

5 September 2014 Q1

T. Bocklitz et al. / Journal of Structural Biology xxx (2014) xxx–xxx

5

Fig.3. AFM images of five different virus species are presented. (A) Varicella-zoster virus (VZV); (B) Porcine teschovirus (PTV); (C) PsP3 bacteriophage; (D) M13 bacteriophage and (E) Tobacco mosaic virus (TMV). The different morphologies and sizes of the viruses are clearly visible. The shown AFM images (A–E) were preprocessed using the JPKSPM data processing software (version 4.2.52) in combination with Gwyddion (version 2.36). The AFM measurements were carried out with tapping mode in air by using a silicon nitride AFM probe.

351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389

height of the PTV is 37.7 nm. The maximal heights of the VZV virions, with a median of 229.0 nm, are greater than values known from the literature (200 nm). Both viruses (VZV & PTV) show a broad size distribution, which is indicated by the large interquartile range, i.e. the box. The interquartile range of the VZV’s and PTV’s maximal height are 190.2–271.7 nm and 34.2–40.8 nm, respectively. One possible reason of the increased size may be an intake of water in presence of hypotonic distilled water or a hindered release of water during the drying of virus particles. Due to the pleomorphic tegument layer, especially VZV would be predestined for this kind of deformation. Another reason may be a systematic error in the background correction procedure, as the correct background is not known. In Fig. 4 B the volumes of the different virions are plotted as boxplot. The higher distance between the 1. quartile and the whisker below (compared to 3. quartile and the whisker above) of the VZV is visible. This can be understood considering the following: The lipid envelope can be affected by the measurement and tegument proteins might leak out. The result would be lower volumes and thus a higher variation in the lower segment of the volume distribution. Furthermore, a tip artifact can appear, i.e. the viral particle appears slightly larger through the radius of the tip (Hermann et al., 2011). Therefore, the variation of VZV particle with smaller volumes is higher. In comparison the variation of the volume is not so pronounced for the PTV virions. As a further non-enveloped virus, the Enterobacteria phage PsP3 was measured by using AFM. The PsP3 bacteriophage has a head– tail structure and belongs to the Myoviridae family. Pseudomonas putida is a potential host of the PsP3 bacteriophage. The head is about 130  5 nm in diameter and exhibits an elongated icosahedral symmetry (Campbell et al., 1995). The capsid protects the double-stranded DNA genome. The extended tail has a length of 197  9nm (Campbell et al., 1995). In Fig. 3 C an AFM image of a PsP3 bacteriophage is presented. It is evident that only the head of the bacteriophage could be measured. It is known from previous studies (Ivanovska et al., 2007; Droz et al., 1993; Dubrovin et al., 2008) that the tail can easily be dislodged from the bacteriophage’s head during AFM measurements. This is caused by an embedding of this structural part of the phage in salts or matrix residues,

which eventually leads to a displacement of the bacteriophage tail (Dubrovin et al., 2008; Droz et al., 1993). In addition, the head of the phage can be dented during the AFM measurements (Dubrovin et al., 2008; Droz et al., 1993). This is reflected in a large distribution of the measured maximum heights of the PsP3 phage’s head (see Fig. 4 A): The median height of the PsP3 phage is 101.7 nm, the first quartile is 88.0 nm and the third quartile 121.0 nm. Beside round to oval-shaped viruses, rod-like and filamentous viruses were analyzed. As a rather long representative, the M13K07 bacteriophage (M13 phage) was selected. This virus infects bacteria, namely Escherichia coli. Here, the single-stranded DNA genome is located in a helical capsid, capped by copies of two different coat proteins. The M13 phage is non-enveloped, belongs to the family of Inoviridae and has a length of 800 to 2000 nm (Li et al., 2013; Ploss and Kuhn, 2010; Mao et al., 2009; Chen et al., 2009; Makowski, 1994; Wolkers et al., 1995; Hemminga et al., 2010; Olofsson et al., 2001). An AFM image of two bacteriophages is shown in Fig. 3 D. With a diameter of 6–7 nm (Li et al., 2013; Ploss and Kuhn, 2010; Mao et al., 2009; Chen et al., 2009; Makowski, 1994; Wolkers et al., 1995; Hemminga et al., 2010; Olofsson et al., 2001), the M13 phage exhibits the smallest height profile among the selected virus species. This can be clearly seen from the boxplot (Fig. 4 A) with a median maximal height of 4.5 nm and a interquartile range of 3.6–5.4 nm. As no efforts were made to stretch the phages on the substrates, accumulation or overlaps of single phages can occur. In Fig. 4 C, a boxplot of the area of all measured virions is plotted. The calculated area of the M13 phages fluctuates considerably (Fig. 4 C) due to the small heights of the viruses. Another filament forming virus, the Tobacco mosaic virus (TMV), was measured. The TMV belongs to the Virgaviridae family and causes the mosaic disease in tobacco plants (Zechmann and Zellnig, 2009; Scholthof, 2004). The TMV is non-enveloped and the helical symmetric capsid surrounds the single-stranded RNA genome. An exemplary AFM image of a TMV virion is shown in Fig. 3 E. The rigid helical rod with a length of around 215 nm can be clearly seen. The TMV is shorter and higher than the M13 phage. In Fig. 4 A the height profile of TMV shows the lowest variance,

Q1 Please cite this article in press as: Bocklitz, T., et al. Single virus detection by means of atomic force microscopy in combination with advanced image analysis. J. Struct. Biol. (2014), http://dx.doi.org/10.1016/j.jsb.2014.08.008

390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428

YJSBI 6610

No. of Pages 9, Model 5G

5 September 2014 6

T. Bocklitz et al. / Journal of Structural Biology xxx (2014) xxx–xxx

A

B

5e+07

200

Volume [nm³]

100 50

5e+05

20 10

5e+04

5

Maximal Height [nm]

Volume

5e+08

500

Maximal Height

5e+06

Q1

M13 phage

PsP3 phage

TMV

PTV

M13 phage

VZV

C

PsP3 phage

TMV

PTV

VZV

5e+04 1e+03

5e+03

Area [nm²]

5e+05

Area

M13 phage

PsP3 phage

TMV

PTV

VZV

Fig.4. Boxplots representing the maximal height (A), volume (B) and area (C) of all measured viral particles of the five different virus species. The properties were plotted logarithmically against the viruses. The black lines illustrate the median, which separates the data into two parts, where 50% of the data are located. The median is the second quartile and surrounded by the box, which is formed by the first and third quartile, respectively. The position of the median and the box gives an impression of the underlying distribution of the data. The whisker are used to determine outlier, plotted as crosses.

433

indicated through the small difference of the median (17.2 nm) from the first quartile (16.5 nm) and third quartile (17.9 nm). The median height is about 17 nm and coincides with the values from previous studies (12–18 nm) (Guckenberger et al., 1994; Imai et al., 1993; Kuznetsov et al., 2001; Harder et al., 2013; Chen et al., 2013).

434

3.2. Image invariants for classification

435

In the previous section features of the AFM images were discussed. The discussion is based on descriptive statistics of parameters like maximal height, area and volume. With the help of boxplots these parameters were compared and discussed. For the automatic recognition of the virus species a predictive model like

429 430 431 432

436 437 438 439

a classification model has to be applied. As virus detection can be a critical task, for example in the case of human infecting viruses, a prediction based only on these simple quantitative features (e.g. maximal height, area and volume) is not yielding a reliable model. It is advisable to incorporate the height distribution and the morphology of the virus into such a model. In order to quantify the morphology of each virion, PZ-invariants were calculated of each single virion. The disadvantage of an incorporation of more information is that the data gets high-dimensional and an easy visual inspection is no longer possible. Such a problem can be solved by a principal-component-analysis (PCA) in combination with a linear-discriminant-analysis (LDA) as classification tool. A visualization of a PCA and LDA is given in Fig. 5. The first and the

Q1 Please cite this article in press as: Bocklitz, T., et al. Single virus detection by means of atomic force microscopy in combination with advanced image analysis. J. Struct. Biol. (2014), http://dx.doi.org/10.1016/j.jsb.2014.08.008

440 441 442 443 444 445 446 447 448 449 450 451 452

YJSBI 6610

No. of Pages 9, Model 5G

5 September 2014 Q1

7

T. Bocklitz et al. / Journal of Structural Biology xxx (2014) xxx–xxx

A

B

PCA loading plot

LDA score plot −2 0

2

4

6

−4 −2

0

2

4

6

5

10

−6

0

6 2

4

1e−08

2

4

6

1e−13

−6

−2

LD value 2

0

50

100

150

200

0

4 2 0

LD value 4

−4

M13 Phage PsP3 phage TMV PTV VZV

6

−4

1e−18

LD value 3

1e−23

Absolute of PC loading

−10

1e−03

LD value 1

−10 −5

Feature Index

0

5

10

−4 −2

0

2

4

6

Fig.5. A visualization of a principal component analysis (PCA) and linear discriminant analysis (LDA) is given. (A) Loading plot of the first (black) and third PC loading (red). The grayish line separates the basic features, such as volume, area and height deciles, from the PZ-invariants. (B) LDA score plot is visualized with the calculated four discriminants. It is obvious that the first and second LD can be used to distinguish between the virus species. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473

third loading of the PCA is given in Fig. 5 A. The features at the left of the gray line are volume, area and the height deciles. After the grayish line the PZ-invariants follow. As the scale of these values have a different range, the PCA was done in a way that every feature is centered and scaled to normality. So the mean of every feature was subtracted and the variance was scaled to one. Fig. 5 B is the composition of all LDA-score plots. For a five-class-problem (five virus species) the implementation in the ‘MASS’ package (Venables and Ripley, 2002) calculates four linear discriminants, which act as marker for certain groups. The investigation of this score plot shows that the combination of LD1 and LD2 can be utilized for the separation of all groups. This is an unusual result, but indicates that much more virus species can be separated using the presented approach. Nevertheless, the score plot is a visualization of the trained model and no judgment of the predictive ability of the model can be made from this plot. To check the predictive ability of the classification model a 10-fold cross validation was applied. The result of this procedure is given in Table 1. The confidence table indicates that the discrimination of the virus species is feasible using AFM images and image analysis in combination with supervised statistical methods as the accuracy of the recognition Table 1 A confidence table of the classifier evaluation (ten-fold cross-validation) is given. The accuracy and the mean sensitivity are 96.8% and 96.0%, respectively. While the M13 phage and PTV are perfectly recognized, the most errors occur in the discrimination of the VZV species. Predicted virus species True virus species

M13 phage

PsP3 phage

TMV

PTV

VZV

M13 phage PsP3 phage TMV PTV VZV Sensitivity Specificity

163 0 5 0 0 100% 98.4%

0 86 0 0 4 94.5% 98.9%

0 0 94 0 0 94.0% 100%

0 2 1 68 0 100% 99.3%

0 3 0 0 43 91.5% 99.2%

was 96.8%. The sensitivities for the virus species classification were 100% (M13 phage), 94.5% (Psp3 phage), 94.0% (TMV), 100% (PTV) and 91.5% (VZV), respectively. The recognition of the M13 phage and PTV was perfect, owing to their clearly distinguished structure. The high number of mis-classifications occurring for the VZV species are related to the fact that less images were available of that species. In cell culture, VZV virions tend to associate to cellular structures. Accordingly, VZV supernatants exhibit usually low virus titers. For this reason, the number of measured VZV virions is smaller compared to the other four virus species. Additionally, VZV has a broad morphological variance (Grose et al., 1995), i.e. dissimilar localized nucleocapsids, and thus different forms of the fried egg structure. The maximal heights, volumes and areas are varying as well. The envelope of the VZV can be affected during the AFM measurements. Therefore, the tegument proteins can get lost, so that the capsid remains partially unprotected. The unprotected icosahedral capsid of VZV has a diameter of around 100 nm (Harson and Grose, 1995; Nii, 1992; Grose et al., 1995), hence the capsid size is in order of the PsP3 phage’ head size. This is evident from Fig. 4 (A–C), as the whiskers of VZV and PsP3 phage overlap. Therefore, four VZV virions are false positives and classified as PsP3 species (see Table 1). Vice versa, three intact PsP3 phage’ heads, which are rather large, were predicted as VZV species. Damaged PsP3 virions form small viral particles, thus two PsP3 virions were assigned to the PTV species. One TMV particle is classified as PTV species (see Table 1). A possible reason is that TMV virions can loose their central RNA followed by breaking into shorter pieces (Guckenberger et al., 1994). Furthermore, TMV virions were classified as M13 phages. This may be a result of a high titer combined with large scan regions and a loss of fineness of the AFM image. Thus, many TMV virions were measured in one AFM scan. The virions are quite close to each other so that interruptions of single virions were not directly seen. Therefore, two or more TMV particle can be identified as one long virion. This leads to a mis-classification of five TMV virions as M13 phages species.

Q1 Please cite this article in press as: Bocklitz, T., et al. Single virus detection by means of atomic force microscopy in combination with advanced image analysis. J. Struct. Biol. (2014), http://dx.doi.org/10.1016/j.jsb.2014.08.008

474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509

YJSBI 6610

No. of Pages 9, Model 5G

5 September 2014 Q1

8

T. Bocklitz et al. / Journal of Structural Biology xxx (2014) xxx–xxx

517

Table 1 shows, that no virus is classified as false positive TMV particles, thus TMV has a 100% specificity. However, the M13 phage is least specific (with the lowest specificity of 98.4%) because five TMV virions are false positively identified as M13 phages. The specificities of the PsP3 phage, PTV and VZV were 98.9%, 99.3% and 99.2%, respectively. These results demonstrate that the truenegative-rate is quite low and a reliable virus species determination is possible.

518

4. Conclusion

519

562

In the present contribution virions of five different virus species were investigated by means of AFM imaging. In a first step easily accessible parameters of every virion were calculated. These quantitative features, including maximal height, volume and area, are in good agreement with the literature values. The boxplots of these features show, that a prediction can be done with these features, but only with high errors, as the whiskers show overlap (Fig. 4). An automatic prediction based on these features would not be robust, as the shape is not incorporated. In order to test, if an automation of the virus determination can be carried out, the features (height deciles, volume and area) were complemented by PZ-invariants. These invariants quantify the morphology and the shape of the virions. With both types of features a classifier was constructed and evaluated. In the presented study a PCA-LDA was utilized for this purpose. The tested virions of five different virus species could be classified with an accuracy of 96.8%. Two of the analyzed virus species (M13 phage & PTV) could be detected with 100% sensitivity. All sensitivities were better then 91.5%, indicating a high true-positive-rate for all virus species. The specificity and thus the true-negative-rate ranged from 98.4% to 100%. Overall, an accuracy of 96.8% was achieved, which indicates that a single virus detection can be achieved by using AFM images only. With the presented analysis an user independent classification of these five virus species could be demonstrated. The advantage is the fact that the user is not influencing the result of the prediction. Therefore, the present analysis regime can be used as automatic pre-sorting tool for subsequent analytical procedures or subsequent experiments. This is useful, if only a certain virus species is of interest. The presented method of single virus detection allows for a fast, reliable and user-independent determination of the virus species even without cultivation, thus can be used for the investigation of complex real-world samples for a point-of-care diagnostic. For this application the construction of databases for the identification of viruses has to be carried out. The presented analysis scheme is not restricted to AFM images. The method can be combined with all microscopic techniques or morphological measurements such as EM and SPM. The only requirement is that the morphology of interest can be measured with the technique. If that is the case, the workflow presented in that contribution can be applied. Thus, the morphology is used in a quantitative manner, allowing for automatic diagnosis and construction of databases. This yields to analytical methods based on morphological measurements.

563

Acknowledgment

564

Financial support of the German Research Foundation (DFG) for the research project PO563=13  1 and furthermore, the funding of the research project ‘FastVirus’ (2011FE9051) by the Thüringer Aufbaubank, the Free State of Thuringia as well as the European Union (EFRE) are gratefully acknowledged.

510 511 512 513 514 515 516

520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561

565 566 567 568

References

569

Bressan, A., Watanabe, S., 2011. Immunofluorescence localisation of Banana bunchy top virus (family Nanoviridae) within the aphid vector, Pentalonia nigronervosa, suggests a virus tropism distinct from aphid-transmitted luteoviruses. Virus Res. 155 (2), 520–525. http://dx.doi.org/10.1016/j.virusres.2010.12.005 . Philippe, N., Legendre, M., Doutre, G., Couté, Y., Poirot, O., Lescot, M., Arslan, D., Seltzer, V., Bertaux, L., Bruley, C., Garin, J., Claverie, J.-M., Abergel, C., 2013. Pandoraviruses: amoeba viruses with genomes up to 2.5 Mb reaching that of parasitic eukaryotes. Science 341 (6143), 281–286. http://dx.doi.org/10.1126/ science.1239181 . Lofgren, E., Fefferman, N.H., Naumov, Y.N., Gorski, J., Naumova, E.N., 2007. Influenza seasonality: underlying causes and modeling theories. J. Virol. 81 (11), 5429– 5436. http://dx.doi.org/10.1128/JVI.01680-06 , arXiv:http://jvi.asm.org/content/81/11/5429.full.pdf+html. Okame, M., Shiota, T., Hansman, G., Takagi, M., Yagyu, F., Takanashi, S., Phan, T.G., Shimizu, Y., Kohno, H., Okitsu, S., Ushijima, H., 2007. Anti-norovirus polyclonal antibody and its potential for development of an antigen-ELISA. J. Med. Virol. 79 (8), 1180–1186. http://dx.doi.org/10.1002/jmv.20906 . Leland, D.S., Ginocchio, C.C., 2007. Role of cell culture for virus detection in the age of technology. Clin. Microbiol. Rev. 20, 49–78. http://dx.doi.org/10.1128/ CMR.00002-06 . Hodinka, R.L., Kaiser, L., 2013. Is the era of viral culture over in the clinical microbiology laboratory? J. Clin. Microbiol. 51 (1), 2–8. http://dx.doi.org/ 10.1128/jcm.02593-12 . Lecuit, M., Eloit, M., 2013. The human virome: new tools and concepts. Trends Microbiol. 21 (10), 510–515. http://dx.doi.org/10.1016/j.tim.2013.07.001 . Ogilvie, M., 2001. Molecular techniques should not now replace cell culture in diagnostic virology laboratories. Rev. Med. Virol. 11 (6), 351–354. http:// dx.doi.org/10.1002/rmv.335 . Schramlová, J., Arientová, S., Hulínská, D., 2010. The role of electron microscopy in the rapid diagnosis of viral infections – review. Folia Microbiol. 55 (1), 88–101 . Hazelton, P.R., Gelderblom, H.R., 2003. Electron microscopy for rapid diagnosis of infectious agents in emergent situations. Emerg. Infect. Dis. 9 (3), 294–303. Zechmann, B., Zellnig, G., 2009. Rapid diagnosis of plant virus diseases by transmission electron microscopy. J. Virol. Methods 162 (1–2), 163–169. http://dx.doi.org/10.1016/j.jviromet.2009.07.032 . Kuznetsov, Y.G., McPherson, A., 2011. Atomic Force Microscopy in imaging of viruses and virus-infected cells. Microbiol. Mol. Biol. Rev. 75 (2), 268–285. http://dx.doi.org/10.1128/mmbr.00041-10 . Ikai, A., Imai, K., Yoshimura, K., Tomitori, M., Nishikawa, O., Kokawa, R., Kobayashi, M., Yamamoto, M., 1994. Scanning tunneling microscopy/atomic force microscopy studies of bacteriophage T4 and its tail fibers. J. Vacuum Sci. Technol. B 12 (3), 1478–1481. http://dx.doi.org/10.1116/1.587320 . Imai, K., Yoshimura, K., Tomitori, M., Nishikawa, O., Kokawa, R., Yamamoto, M., Kobayashi, M., Ikai, A., 1993. Scanning tunneling and atomic force microscopy of T4 bacteriophage and Tobacco mosaic virus. Jpn. J. Appl. Phys. 32 (6S), 2962– 2964 . Guckenberger, R., Arce, F.T., Hillebrand, A., Hartmann, T., 1994. Imaging of uncoated tobacco mosaic virus by scanning tunneling microscopy. J. Vacuum Sci. Technol. B 12 (3), 1508–1511. http://dx.doi.org/10.1116/1.587274 . Kuznetsov, Y.G., Chang, S.-C., Credaroli, A., Martiny, J., McPherson, A., 2012. An atomic force microscopy investigation of cyanophage structure. Micron 43 (12), 1336–1342. http://dx.doi.org/10.1016/j.micron.2012.02.013 (special issue on AFM in Biology & Bionanomedicine). Chen, S.-w.W., Odorico, M., Meillan, M., Vellutini, L., Teulon, J.-M., Parot, P., Bennetau, B., Pellequer, J.-L., 2013. Nanoscale structural features determined by AFM for single virus particles. Nanoscale 5, 10877–10886. http://dx.doi.org/ 10.1039/C3NR02706F . Allison, D.P., Mortensen, N.P., Sullivan, C.J., Doktycz, M.J., 2010. Atomic force microscopy of biological samples. Wiley Interdiscip. Rev. Nanomed. Nanobiotechnol. 2 (6), 618–634. http://dx.doi.org/10.1002/wnan.104 . Kuznetsov, Y.G., Malkin, A.J., Lucas, R.W., Plomp, M., McPherson, A., 2001. Imaging of viruses by atomic force microscopy. J. General Virol. 82 (9), 2025–2034 , arXiv: http:// vir.sgmjournals.org/content/82/9/2025.full.pdf+html. Santos, N.C., Castanho, M.A., 2004. An overview of the biophysical applications of atomic force microscopy. Biophys. Chem. 107 (2), 133–149. http://dx.doi.org/ 10.1016/j.bpc.2003.09.001 . Flusser, J., Zitova, B., Suk, T., 2009. Moments and Moment Invariants in Pattern Recognition. Wiley.

570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651

Q1 Please cite this article in press as: Bocklitz, T., et al. Single virus detection by means of atomic force microscopy in combination with advanced image analysis. J. Struct. Biol. (2014), http://dx.doi.org/10.1016/j.jsb.2014.08.008

YJSBI 6610

No. of Pages 9, Model 5G

5 September 2014 Q1 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716

T. Bocklitz et al. / Journal of Structural Biology xxx (2014) xxx–xxx Cialla, D., Deckert-Gaudig, T., Budich, C., Laue, M., Möller, R., Naumann, D., Deckert, V., Popp, J., 2009. Raman to the limit: tip-enhanced Raman spectroscopic investigations of a single tobacco mosaic virus. J. Raman Spectrosc. 40 (3), 240– 243. http://dx.doi.org/10.1002/jrs.2123 . R Development Core Team, 2008. R: A Language and Environment for Statistical Computing, R Foundation for Statistical Computing, Vienna, Austria, ISBN 3900051-07-0. URL:http://www.R-project.org. Fortran code by H. Akima R port by Albrecht Gebhardt aspline function by Thomas Petzoldt ([email protected]) enhancements and corrections by Martin Maechler, akima: Interpolation of irregularly spaced data, R package version 0.5-4 (2009). URL:http://CRAN.R-project.org/package=akima. Rajwa, B., Dundar, M., Irvine, A., Dang, T., 2013. IM: Orthogonal Moment Analysis, r package version 1.0 (2013). URL:http://CRAN.R-project.org/package=IM. Venables, W.N., Ripley, B.D., 2002. Modern Applied Statistics with S, 4th ed. Springer, New York . Analytics, R., Weston, S., 2013. foreach: Foreach looping construct for R. r package version 1.4.1. URL:http://CRAN.R-project.org/package=foreach. Xia, T., Zhu, H., Shu, H., Haigron, P., Luo, L., 2007. Image description with generalized pseudo-Zernike moments. JOSA A 24 (1), 50–59. Chong, C.-W., Raveendran, P., Mukundan, R., 2003. The scale invariants of pseudoZernike moments. Pattern Anal. Appl. 6 (3), 176–184. Kylberg, G., Uppström, M., Heldlund, K.-O., Borgefors, G., Sintorn, I.-M., 2012. Segmentation of virus particle candidates in transmission electron microscopy images. J. Microsc. 245 (2), 140–147. http://dx.doi.org/10.1111/j.13652818.2011.03556.x . Medyukhina, A., Meyer, T., Heuke, S., Vogler, N., Dietzek, B., Popp, J., 2013. Automated seeding-based nuclei segmentation in nonlinear optical microscopy. Appl. Optics 52 (28), 6979–6994 . Smith, C.K., Arvin, A.M., 2009. Varicella in the fetus and newborn. Semin. Fetal Neonatal Med. 14 (4), 209–217. http://dx.doi.org/10.1016/j.siny.2008.11.008 . Almeida, J.D., Howatson, A.F., Williams, M.G., 1962. Morphology of Varicella (Chicken Pox) Virus. Virology 16 (3), 353–355. http://dx.doi.org/10.1016/ 0042-6822(62)90261-1 . Harson, R., Grose, C., 1995. Egress of Varicella-zoster virus from the melanoma cell: a tropism for the melanocyte. J. Virol. 69 (8), 4994–5010 . Chiu, S.-C., Hu, S.-C., Chang, C.-C., Chang, C.-Y., Huang, C.-C., Pang, V.F., Wang, F.-I., 2012. The role of porcine teschovirus in causing diseases in endemically infected pigs. Vet. Microbiol. 161 (1–2), 88–95. http://dx.doi.org/10.1016/ j.vetmic.2012.07.031 . Yamada, M., Miyazaki, A., Yamamoto, Y., Nakamura, K., Ito, M., Tsunemitsu, H., Narita, M., 2014. Experimental teschovirus encephalomyelitis in gnotobiotic pigs. J. Comp. Pathol. 150 (2–3), 276–286. http://dx.doi.org/10.1016/ j.jcpa.2013.08.004 . Chiu, S.-C., Yang, C.-L., Chen, Y.-M., Hu, S.-C., Chiu, K.-C., Lin, Y.-C., Chang, C.-Y., Wang, F.-I., 2014. Multiple models of porcine teschovirus pathogenesis in endemically infected pigs. Vet. Microbiol. 168 (1), 69–77. http://dx.doi.org/ 10.1016/j.vetmic.2013.10.019 . Hermann, P., Hermelink, A., Lausch, V., Holland, G., Möller, L., Bannert, N., Naumann, D., 2011. Evaluation of tip-enhanced Raman spectroscopy for characterizing different virus strains. Analyst (Cambridge, U.K.) 136, 1148–1152. http:// dx.doi.org/10.1039/C0AN00531B . Campbell, J.I.A., Albrechtsen, M., Sørensen, J., 1995. Large Pseudomonas phages isolated from barley rhizosphere. FEMS Microbiol. Ecol. 18 (1), 63–74. http:// dx.doi.org/10.1111/j.1574-6941.1995.tb00164.x. Ivanovska, I., Wuite, G., Jönsson, B., Evilevitch, A., 2007. Internal DNA pressure modifies stability of WT phage. Proc. Natl. Acad. Sci. 104 (23), 9603–9608. http://dx.doi.org/10.1073/pnas.0703166104 , arXiv: http://www.pnas.org/content/104/23/ 9603.full.pdf+html. Droz, E., Taborelli, M., Wells, T., Descouts, P., 1993. Preparation of isolated biomolecules for SFM observations: T4 Bacteriophage as a test sample. Biophys. J. 65 (3), 1180–1187. http://dx.doi.org/10.1016/S00063495(93)81174-3 . Dubrovin, E.V., Voloshin, A.G., Kraevsky, S.V., Ignatyuk, T.E., Abramchuk, S.S., Yaminsky, I.V., Ignatov, S.G., 2008. Atomic Force Microscopy investigation of phage infection of bacteria. Langmuir 24 (22), 13068–13074. http://dx.doi.org/ 10.1021/la8022612, pMID: 18850726. arXiv: http://pubs.acs.org/doi/pdf/ 10.1021/la8022612. URL:. Li, T., Zan, X., Sun, Y., Zuo, X., Li, X., Senesi, A., Winans, R.E., Wang, Q., Lee, B., 2013. Self-assembly of Rodlike virus to superlattices. Langmuir 29 (41), 12777–12784. http://dx.doi.org/10.1021/la402933q, arXiv: http://pubs.acs.org/doi/pdf/ 10.1021/la402933q, URL:http://pubs.acs.org/doi/abs/10.1021/la402933q. Ploss, M., Kuhn, A., 2010. Kinetics of filamentous phage assembly. Phys. Biol. 7 (4), 045002 . Mao, C., Liu, A., Cao, B., 2009. Die Anwendung von Viren in Chemo- und Biosensoren. Angew. Chem. 121 (37), 6922–6943. http://dx.doi.org/10.1002/ange.200900231 . Chen, Y.-Y., Wu, C.-C., Hsu, J.-L., Peng, H.-L., Chang, H.-Y., Yew, T.-R., 2009. Surface rigidity change of Escherichia coli after filamentous bacteriophage infection. Langmuir 25 (8), 4607–4614. http://dx.doi.org/10.1021/la8036346, pMID: 19366225. arXiv: http://pubs.acs.org/doi/pdf/10.1021/la8036346, URL:http:// pubs.acs.org/doi/abs/10.1021/la8036346. Makowski, L., 1994. Phage display: structure, assembly and engineering of filamentous bacteriophage M13. Curr. Opin. Struct. Biol. 4 (2), 225–230. http://dx.doi.org/10.1016/S0959-440X(94)90312-3 . Wolkers, W.F., Haris, P.I., Pistorius, A.M.A., Chapman, D., Hemminga, M.A., 1995. FTIR spectroscopy of the major coat protein of M13 and Pf1 in the phage and reconstituted into phospholipid systems. Biochemistry 34 (24), 7825–7833. http://dx.doi.org/10.1021/bi00024a006, arXiv: http://pubs.acs.org/doi/pdf/ 10.1021/bi00024a006, URL:http://pubs.acs.org/doi/abs/10.1021/bi00024a006. Hemminga, M.A., Vos, W.L., Nazarov, P.V., Koehorst, R.B., Wolfs, C.J.A.M., Spruijt, R.B., Stopar, D., 2010. Viruses: incredible nanomachines. New advances with filamentous phages. Eur. Biophys. J. 39 (4), 541–550. http://dx.doi.org/10.1007/ s00249-009-0523-0, URL:http://dx.doi.org/10.1007/s00249-009-0523-0. Olofsson, L., Ankarloo, J., Andersson, P.O., Nicholls, I.A., 2001. Filamentous bacteriophage stability in non-aqueous media. Chem. Biol. 8 (7), 661–671. http://dx.doi.org/10.1016/S1074-5521(01)00041-2 . Scholthof, K.-B.G., 2004. Tobacco mosaic virus: a model system for plant biology. Annu. Rev. Phytopathol. 42, 13–34. http://dx.doi.org/10.1146/ annurev.phyto.42.040803.140322, URL:. Kuznetsov, Y.G., Larson, S.B., Day, J., Greenwood, A., McPherson, A., 2001. Structural transitions of satellite Tobacco mosaic virus particles. Virology 284 (2), 223– 234. http://dx.doi.org/10.1006/viro.2000.0914 . Harder, A., Dieding, M., Walhorn, V., Degenhard, S., Brodehl, A., Wege, C., Milting, H., Anselmetti, D., 2013. Apertureless scanning near-field optical microscopy of sparsely labeled tobacco mosaic viruses and the intermediate filament desmin. Beilstein J. Nanotechnol. 4, 510–516. http://dx.doi.org/10.3762/bjnano.4.60 . Grose, C., Harson, R., Beck, S., 1995. Computer modeling of prototypic and aberrant nucleocapsids of Varicella–zoster virus. Virology 214 (2), 321–329. http:// dx.doi.org/10.1006/viro.1995.0041 . Nii, S., 1992. Electron microscopic study on the development of herpesviruses. J. Electron Microsc. 41 (6), 414–423, arXiv: http://jmicro.oxfordjournals.org/ content/41/6/414.full.pdf+html. URL:.

Q1 Please cite this article in press as: Bocklitz, T., et al. Single virus detection by means of atomic force microscopy in combination with advanced image analysis. J. Struct. Biol. (2014), http://dx.doi.org/10.1016/j.jsb.2014.08.008

717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781

Single virus detection by means of atomic force microscopy in combination with advanced image analysis.

In the present contribution virions of five different virus species, namely Varicella-zoster virus, Porcine teschovirus, Tobacco mosaic virus, Colipha...
2MB Sizes 0 Downloads 6 Views