Early infection and spread of a conditionally replicating adenovirus under conditions of plaque formation

Virology 423 (2012) 89–96

Contents lists available at SciVerse ScienceDirect

Virology journal homepage: www.elsevier.com/locate/yviro

Early infection and spread of a conditionally replicating adenovirus under conditions of plaque formation Andrew Hofacre a, Dominik Wodarz b, Natalia L. Komarova c, Hung Fan a,⁎ a b c

Department of Molecular Biology and Biochemistry, Cancer Research Institute, University of California, Irvine, CA 92697, USA Department of Ecology and Evolutionary Biology, 321 Steinhaus Hall, University of California, Irvine, CA 92697, USA Department of Mathematics, Rowland Hall, University of California, Irvine, CA 92697, USA

a r t i c l e

i n f o

Article history: Received 23 September 2011 Returned to author for revision 18 October 2011 Accepted 21 November 2011 Available online 20 December 2011 Keywords: Adenovirus Conditionally replicating adenovirus Green fluorescent protein Adenovirus infection

a b s t r a c t Conditionally-replicating adenoviruses (CRAds) and other oncolytic viruses replicate selectively in tumor cells, presenting a potential cancer treatment approach. To optimize application of these viruses, understanding of early spread of these viruses in target cells is important. Here we used a recombinant adenovirus expressing enhanced jellyfish green fluorescent protein (EGFP) in place of the EIA and EIB genes (AdEGFPuci). Infection of susceptible cells (AD-293) under plaque formation conditions (MOIb b 1) on gridded culture dishes and daily monitoring allowed visualization of initially infected cells, as well as spread to neighboring cells. We determined key parameters of early infection, including the rate and efficiency of spread from the initially infected cell to other cells. It was noteworthy that a minority of initially infected cells ultimately resulted in plaques. The approaches elucidated here will be useful for determining early infection parameters for CRAds of therapeutic interest. © 2011 Elsevier Inc. All rights reserved.

Introduction Oncolytic viruses that replicate selectively in and lyse tumor cells have been of considerable interest as novel approaches to cancer therapy (Chiocca, 2002; Wong et al., 2010). Their side effects may be less severe than traditional cancer treatments such as chemotherapy, surgery, and radiation. Extensive laboratory research and clinical trials have been conducted with several engineered oncolytic viruses based on adenovirus, reovirus, herpes simplex virus type I, poxvirus, poliovirus, Newcastle disease virus, measles virus, and vesicular stomatitis virus (Alemany, 2007; Garber, 2006; Kirn et al., 1998; Lin and Nemunaitis, 2004; Parato et al., 2005; Russell and Peng, 2007; Toth et al., 2011). Conditionally-replicating adenoviruses (CRAds) have been extensively studied as oncolytic viruses (Toth et al., 2010). CRAds typically lack the EIB and/or EIA genes; these viral genes inactivate the cellular p53 and RB tumor suppressor genes that are frequently mutated or inactivated in human cancers. Thus CRAds will replicate in tumor cells lacking p53 and/or Rb, although other mechanisms of tumor selectivity may be operative (Hall et al., 1998; Lane, 1998; Turnell et al., 1999). In particular Adenovirus dl1520 (ONYX-015), defective in the EIB 55 Kd protein, has been studied in vitro and in vivo (Bischoff et al., 1996; Kirn et al., 1998). While Addl1520 and similar

⁎ Corresponding author. Fax: + 1 949 824 4023. E-mail address: [email protected] (H. Fan). 0042-6822/$ – see front matter © 2011 Elsevier Inc. All rights reserved. doi:10.1016/j.virol.2011.11.014

viruses have shown some efficacy in Phase I clinical trials on cancer patients, so far larger trials have not demonstrated sustained and reliable treatment success (Liu and Kirn, 2008; Ries and Brandts, 2004). One contributory factor could be that the kinetics of oncolytic virus spread in tumors is not well understood, so optimal methods of delivery may not have been employed. One approach to understanding kinetics of viral spread has been through mathematical modeling. Previously mathematical analyses of experimentally measured spreading for several viruses, including human cytomegalovirus (Lam et al., 2006), vesicular stomatitis virus (Haseltine et al., 2008) and T7 bacteriophage (Fort, 2002; Fort and Mendez, 2002; Yin and McCaskill, 1992; You and Yin, 1999), have generated models for spread and plaque formation. We have modeled oncolytic virus spread in tumor cells and obtained interesting predictions (Komarova and Wodarz, 2010; Wodarz, 2003; Wodarz and Komarova, 2009; Zurakowski and Wodarz, 2007). However application of the modeling to particular oncolytic viruses requires empirical determination of key parameters in the models. Another approach recently employed by several research groups (including the experiments described here) involves generation of recombinant CRAds expressing fluorescent proteins (Borovjagin et al., 2010; Hofacre and Fan, Unpublished results; Le et al., 2006a; Ono et al., 2005; Ugai et al., 2010). Infected cells can be detected by fluorescence under UV illumination, and importantly cells or cultures can be sequentially imaged at different times, allowing temporal tracking of viral infection (Le et al., 2006b; Ono et al., 2005). We have used this approach to characterize very early events in CRAd virus infection and spread.

90

A. Hofacre et al. / Virology 423 (2012) 89–96

In the experiments described in this study, we have used a CRAd expressing enhanced jellyfish green fluorescent protein (EGFP) in the context of an adenovirus type 5 vector defective in EIA and EIB (AdEGFPuci). Infection of AdEGFPuci onto complementing cells (AD-293) at low multiplicity in gridded dishes under conditions of plaque formation (MOI b b1) followed by sequential daily monitoring for infected cells allowed determination of several key parameters and properties of AdEGFPuci in these cells. The approaches elucidated here will be useful for determining early infection parameters for CRAds such as Addl1520. Results Determination of early kinetic parameters of adenovirus infection in AD-293 cells In these experiments we established a system to study the kinetics of establishment and spread of CRAd infection. Such information would be important for understanding spread of oncolytic adenoviruses such as dl1520 in vivo. We employed a recombinant adenovirus type-5 expressing EGFP gene in place of EIA and EIB (AdEGFPuci, Fig. 1). While removal of the viral sequences was necessary to accommodate the EGFP gene within the packaging limits for adenovirus, AdEGFPuci will replicate in human 293 embryonic kidney cells that constitutively express adenovirus EIA and EIB viral proteins (Graham et al., 1977). AdEGFPuci replicated efficiently in 293 cells, giving viral stocks with titers of 10 9–10 11 pfu/ml. In this study establishment and spread of AdEGFPuci was carried out under conditions of very low multiplicity infection of HEK293 cells in monolayer, under plaque formation conditions (including overlay with 0.65% agar) (Graham and van der Eb, 1973). The 293 cells were AD-293 cells obtained from a commercial supplier (Agilent Technologies, La Jolla, CA). A titered AdEGFPuci stock was diluted and infected onto AD-293 cells in 60 mm tissue culture dishes with 2 mm grids under standard conditions. The AdEGFPuci dilutions were chosen so that less than 30% of the grid squares ever showed an infected cell; under these conditions it was very likely that any infected cells appearing in a square at any time resulted from one initial infection. The dishes were scanned under UV illumination daily, and each square was scored for the presence or absence of green (virus-infected) cells. Results from two infections are shown in Fig. 2A. Grid squares with one or a small number of cells are shown as green dots, while those with larger numbers of green cells (developing plaques) are indicated by stars. In one experiment (Experiment 4) the numbers of fluorescent cells in each square were scored until they exceeded 10 cells. The cumulative number of grid squares showing evidence for infection with respect to days post-infection is shown in Fig. 2B. In these

MLP E1A

E1B

LITR Ψ

L1

L2

L3

L5

L4

E3 RITR

Adenoviral genome IX

CMV GFP

E2B

E2A

E4

IVa2

AdEGFPuci

Fig. 1. Organization of AdEGFPuci. Schematic diagram of the EGFP recombinant Ad viruses, AdEGFPuci (ΔE1A/E1B), is shown. For reference the transcriptional map of adenovirus is shown at the top of the figure. The adenovirus genome contains early (E1–4), intermediate (IX and IVa2) and late (L1-5) genes flanked by left and right inverted terminal repeats (LITR and RITR, respectively). MLP: major late promoter, Ψ packaging signal.

low multiplicity infections, the time between initial infection and appearance of green fluorescence was 2 days ± 1 day (Fig. 2B). This lag likely was influenced by the average time for entry and uncoating of virus into the cells, as well as the time required for expression of the viral genome and build-up of EGFP to levels where fluorescence could be detected. (Since EGFP was driven by the CMV promoter, transcription of EGFP mRNA should have occurred with “immediate early” kinetics.) Surprisingly under these conditions a minority (28.1 ± 4.0%) of initial infections progressed to spreading infection and a viral plaque (Fig. 2B). Several patterns of infection within individual grid squares were observed over time, as shown in Figs. 3A and B. In virtually all cases, infection within a square began with one fluorescent cell, as would be expected for the low multiplicity of infection used. In some cases the individual infected cell persisted for a few days (or longer) and then it disappeared — mostly like due to lysis of the infected cell (pattern A). This could represent infection of a cell with a defective AdEGFPuci genome, or productive infection that did not successfully spread. A second pattern (B) started with one infected cell, followed by subsequent appearance of additional fluorescent cells, and ultimately a large region of local infection — i.e. a viral plaque. At the viral plaque stage, cytopathic effect (CPE) in the region of infection was visible by light microscopy, and ultimately a cleared region was evident — surrounded by green fluorescent cells. In another pattern (C) a grid square showed an infected cell that persisted for a few days, after which it disappeared (presumably due to lysis); however subsequently another infected cell appeared which either disappeared after a few days (pattern C), or it progressed to a spreading infection and formation of a plaque (pattern D). We interpret patterns C and D as showing an initially infected cell that spread infection to another cell, which either did (pattern D) or did not (pattern C) result in widespread infection and development of a plaque. The data for all of the infections are presented in Supplemental Data — Table S1, and the relative frequencies of the different patterns will be presented below. Analysis of the kinetics of infection in the infected grid squares with particular patterns allowed determination of additional parameters of initial infection. The distribution of the number of days that the initially infected cell persisted in grids with patterns C and D is shown Fig. 4A. In these squares the initially infected cell was clearly productively infected since it spread infection to an additional cell(s). At the same time the fact that the initially infected cell disappeared before appearance of secondarily infected cells allowed estimation of the lifetime of the initially infected cells. The results indicated that an initially infected fluorescent cell persisted for 2.5 ± 1 days on average before it dies. Since the average time between infection and appearance of detectable fluorescence was 2 ± 1 days (Fig. 2B), the average life of an initially infected AdEGFPuci infected cell was 4.5 ± 2 days under these conditions, assuming that binding, penetration and uncoating were relatively rapid — 2 days before it showed green fluorescence, and 2.5 days after it turned green. For infected grid squares with patterns C and D, the number of days that a grid showed no fluorescence after disappearance of the initially infected cell before appearance of a secondarily infected cell(s) is plotted in Fig. 4B. This was typically 1 day (76.9 ± 0.9%). Thus, on average the secondarily infected cells in pattern C and D grids appeared at 5.5 days post-infection. Since it took infected cells 2 days to show EGFP fluorescence (Fig. 2B), the initially infected cells released infectious virus that spread to another cell at approximately 3.5 days in order for secondarily infected cells to appear at 5.5 days. There were rare cases where multiple infected cells appeared in a grid without a single infected cell appearing first (pattern E, not shown, 2 cases out 265, b1%). In both of these cases the multiply infected cells first became evident at 7 and 9 days post-infection respectively. We attribute them to instances where the initially

A. Hofacre et al. / Virology 423 (2012) 89–96

A

91

1 2 3 4 5 6 7 8 9 10 11 12 13 A B C D E F G H I J K L MNO P Q R S T U VWX Y Z 15 16 17 18 19 20 21 22 23 24 25 26

1 2 3 4 5 6 7 8 9 10 11 12 13 A B C D E F G H I J K L MNO P Q R S T U VWX Y Z 15 16 17 18 19 20 21 22 23 24 25 26

1 2 3 4 5 6 7 8 9 10 11 12 13 A B C D E F G H I J K L MNO P Q R S T U VWX Y Z 15 16 17 18 19 20 21 22 23 24 25 26

1 2 3 4 5 6 7 8 9 10 11 12 13 A B C D E F G H I J K L MNO P Q R S T U VWX Y Z 15 16 17 18 19 20 21 22 23 24 25 26

1 2 3 4 5 6 7 8 9 10 11 12 13 A B C D E F G H I J K L MNO P Q R S T U VWX Y Z 15 16 17 18 19 20 21 22 23 24 25 26

1 2 3 4 5 6 7 8 9 10 11 12 13 A B C D E F G H I J K L MNO P Q R S T U VWX Y Z 15 16 17 18 19 20 21 22 23 24 25 26

1 2 3 4 5 6 7 8 9 10 11 12 13 A B C D E F G H I J K L MNO P Q R S T U VWX Y Z 15 16 17 18 19 20 21 22 23 24 25 26

1 2 3 4 5 6 7 8 9 10 11 12 13 A B C D E F G H I J K L MNO P Q R S T U VWX Y Z 15 16 17 18 19 20 21 22 23 24 25 26

1

5

B

2

3

6

7

Experiment 1 540

135

No. of Grid Squares

8

Experiment 4

540

135

> 1 infected cell

120

120 plaques

105

105

90

90

75

75

60

60

45

45

30

30

15

15

0

4

> 1 infected cell plaques

0 0

1

2

3

4

5

6

7

8

9

Days

0

1

2

3

4

5

6

7

8

9

10 11 12 13 14 15 16 17

Days

Fig. 2. Kinetics of AdEGFPuci infection under conditions of plaque formation. (A) Representative infection of AD-293 cells seeded onto gridded tissue culture dishes was infected by a diluted stock of AdEGFPuci under plaque formation conditions (MOI b b 0.01) and the dish was observed under UV illumination for infected cells every 24 h. Results for each grid square for the first 8 days are shown. The dots indicate the presence of a fluorescent cell. The stars represent the presence of multiple fluorescent cells and spreading infection — i.e. a plaque. (B) Data from one of three similar experiments (Experiment 1) were collected in the panel on the left, and the cumulative number of squares (out of 540) showing one or more infected cells is shown (diamonds). The cumulative number of grid squares showing plaques is also shown (squares). For these experiments when the number of infected cells in a grid was >2, the grid was scored as one with spreading infection but the number of infected cells was not recorded. For the first three experiments, the percentage of grids showing initial infection that progressed to plaques was 27.8 ± 4.8%. Data from a fourth experiment is shown on the right – in this experiment the number of infected cells were also counted and recorded for those grids with spreading infection (>2 infected cells) for the first 16 days (see Fig. 6).

infected cell was not observed, either due to observational error, or because it died soon after it began to show green fluorescence and between observations. Patterns of spread from initially infected cells The patterns of spread from initially infected cells for AdEGFPuci in AD-293 cells, using results from four independent experiments are shown in Fig. 5A. The fates of the initially infected cells were categorized as spread of infection to 0, 1, 2 or >2 secondarily infected cells (1 → 0, 1 → 1, 1 → 2 and 1 → >2, respectively) and the number of events in each category is shown. Secondary infections were characterized as the state of infection (number of fluorescent cells) 3–5 days after appearance of an initially infected cell. One area of uncertainty was for cells in pattern A, if the initially infected cell persisted for more than 3 days. These grid squares could represent cells infected with a defective virus (green fluorescence but no

lysis), or they could reflect spread of infection from the originally infected cell to a single secondarily infected cell where the originally infected cell lysed on the same day that the secondarily infected cell began to show fluorescence. As a simplifying assumption, grid squares in pattern A where the single infected cell persisted for more than 3 days were categorized as spread of infection to one additional cell. Thus grid pattern A was subdivided into pattern A1 (single fluorescent cell persisting for ≤3 days) and pattern A2 (single fluorescent cell persisting for ≥ 4 days); pattern A1 was assigned to the 1 → 0 category, and A2 was assigned to the 1 → 1 category. Fig. 5A also shows the ultimate results of the infections for each grid at 14 days post-infection, where spreading infection (S) resulted in a plaque while abortive infection (A) resulted in no plaque. As mentioned above, the overall percentage of initially infected cells that gave rise to spreading infection and a plaque was 28.1 ± 4%. For grids where the initially infected cell spread infection to >2 cells, spreading infection and a plaque resulted 100% of the time (37

92

A. Hofacre et al. / Virology 423 (2012) 89–96

A

A1

B

A2

A

A

B

B

C

C

D

D

1

2

3

4

= single cell

5

6

7

8 (days) 1

2

3

4

5

6

7

= plaque

Fig. 3. Patterns of AdEGFPuci infection. (A) Diagrams of patterns of infection for AD-293 infected with AdEGFPuci (Fig. 2) are shown. In virtually all cases, grid squares initially showed a single infected cell. Pattern A, persistence of one infected cell for several days, represented by a triangle, followed by disappearance of the fluorescent cell. Pattern B, one infected cell followed by additional fluorescent cells, which ultimately led to a viral plaque, represented by a star. Pattern C, one infected cell that persisted for several days, then disappeared for at least one day, before fluorescence reappeared on a subsequent day in the same square, but then ultimately fluorescence ceased. Pattern D, similar to pattern C, except that additional fluorescent cells reappeared in the square, ultimately resulting in a plaque. Pattern A and B may also be represented by two subcategories, A1 and A2, that could result from overlap of EGFP fluorescence from the primary infected cell simultaneously with a neighboring secondarily infected cell rather than a singly infected cell that persisted for an extensive number of days. (B) Photo montages (100 ×) of grids showing the patterns diagramed in (A).

grids). If the initially infected cell spread infection to just 2 cells, then there was an intermediate probability of spreading infection (53%, 17/32 grids). It was noteworthy that a significant fraction of these grids did not progress to spreading infection (15/32 grids). If the initially infected cell spread infection to just one secondary cell, then the overall rate of the secondarily infected cell giving rise to spreading infection and a plaque was 24.7% (24/97 grids) — slightly lower but not statistically different from the overall rate for the single infected cells in the initial infection. In the analysis in Fig. 5A, we were concerned that the simplifying assumption (assigning grids with pattern A2 to the 1 → 1 category) could have included abortively infected cells that never spread infection to another cell. This could have at least partially caused the apparently lower efficiency of spreading infection in the 1 → 1 category compared to the initially infected cells. To circumvent this concern, we analyzed data from the subset of grid patterns C and D, where a single secondarily infected cell appeared after disappearance of the initially infected cell (Fig. 3). For these squares there was no uncertainty about the replication competency of the virus in the initially infected cell. Patterns of spread of infection for the next (tertiary) round of infection from these cells are shown in Fig. 5B. The analysis from Fig. 5B indicated that the secondarily infected single cells behaved essentially the same as the initially infected cells in terms of the efficiency with which they spread infection to tertiarily infected cells, and the likelihood of giving a spreading infection. (In fact most of the events in the 1 → 1 category of Fig. 5A came from grid patterns C and D (97) and fewer came from pattern A2 (8)). Spread of infection — comparison with a stochastic null model The data gathered from four independent experiments (Fig. 2A and Table S1) showed various pathways that AdEGFPuci infection in AD-293 cells can follow, and provided a statistical description of the different outcomes. In particular, we found that infection always started with one infected cell, and in 28 ± 4% of the cases spreading infection occurred and a plaque was formed within 14 days of observation. A similar percentage (21 ± 10%) of initially infected cells resulted in plaques in a fourth experiment, where the numbers of infected cells were recorded for grid squares with >2 secondarily

infected cells. When initial infection spread to 2 infected cells, in more than a half of the cases spreading infection and a plaque resulted. It was also noteworthy that an initially infected cell immediately spread infection to >2 cells in a minority (13.4%) of times. A striking observation was that when infection spread to three or more infected cells, this resulted without exception in plaque formation. Here we show that these observations are incompatible with a simple null-model whereby infection from the initially infected cell spreads stochastically according to a so-called “birth-death” process. In the null-model, the probability of infection extinction can be predicted starting from the number of cells secondarily infected from an initially infected cell. As described above the probability of an initially infected cell forming a plaque was found to be 28%, which corresponded to an extinction probability p1 of ~0.72. As shown in Fig. 5B, when initial infection spread to one secondarily infected cell, the extinction rate was similar to p1. If the same laws of spread apply to grid squares where initial infection spread to other numbers of cells, the probability of infection extinction starting with i secondarily infected cells would be pi ≥ p1i. In particular in a non-spatial model, where infected cells spread infection according to laws of mass-action, we have pi = p1i, and in spatial models the probability of extinction would be higher. As a consequence, the probability of plaque formation when there are i secondarily infected cells would be lower than 1–(0.72) i (Fig. 6A). The number distribution of secondarily infected cells taken from Experiment No. 4 is shown in Fig. 6B, where 16 grid squares showed spread of infection to > 2 cells. Calculation of the probabilities for each value of i (3 through 10) and weighting for the number of grids in each i category resulted in an overall predicted extinction probability for squares with > 2 secondarily infected cells to be 0.19 (3 out of 16 squares), compared to the observed 0%. If the number distribution of secondarily infected cells in experiments 1–3 was similar to experiment 4, then the null-model would predict 7 out of 37 grid squares with > 2 secondarily infected showing extinguishments of infection, while none were observed. Thus the results did not fit the stochastic null-model when the number of secondarily infected cells was > 2, although they were consistent with it for 1 or 2 secondarily infected cells.

A. Hofacre et al. / Virology 423 (2012) 89–96

A

Lifespan of Initial Fluorescent Cells 50

No. of Grid Squares

45 40 35 30 25 20 15

10 5

0

1

2

3

4

5

6

7

8

Days

B

Time Between Fluorescence (1 o – 2o Infected cells)

Percent of Grid Squares

100.0

80.0

60.0

40.0

20.0

0.0

1

2

3

4

5

6

7

8

Days Fig. 4. Lifespan of initially infected cells. (A) Data from Experiments 1–4 (Fig. 2) were analyzed for the length of persistence of initially infected cells. The number of days that the initially infected cells persisted in grid squares with patterns C and D of Fig. 3 are shown. (B) For grid squares with patterns C and D, the number of days between disappearance of the initially infected fluorescent cell and reappearance of one or more secondarily infected cells is shown. Results are plotted as percent of the total grid squares analyzed (129).

Discussion In these experiments, we characterized the initial dynamics of infection for the CRAd AdEGFPuci under conditions of plaque formation — i.e. low MOI under an agar overlay. The cells used were AD-293, a derivative of HEK293 (293) cells that constitutively express Ad EIA and EIB proteins (Graham et al., 1977). These cells could support replication of AdEGFPuci, which contained EGFP inserted in place of the EIA and EIB coding regions. Performing the infections on gridded culture dishes allowed repeated monitoring at daily intervals for signs of infection (green fluorescence). This approach allowed determination of several parameters and properties of initial AdEGFPuci infection, including patterns of infection spread, and some of the results were surprising. First, a minority (~28%) of initially infected cells resulted in spreading infection and a plaque. Second, spread of infection from initially infected cells to secondarily infected cells was generally inefficient — in ~74% of cases the initially infected cell spread infection to no or only one secondary cell. Third, the probability of infection leading to spreading infection and a plaque did not fit a stochastic null-model when initial infection spread to more than two secondarily infected cells — in 100% of these cases plaques resulted. In this system, the average time for appearance of green fluorescence was 2 ± 1 days, and the average lifetimes of

93

initially infected cells were 4.5 ± 2 days. The average time for spread of infection to secondarily infected cells was ~ 3.5 days. The low efficiency of spread of infection from an initially infected cell to secondary cells observed here was surprising. The traditional view is that an infected cell produces many infectious progeny viruses (i.e. a large burst size), so that multiple secondarily infected cells are expected. However for eukaryotic viruses such as adenovirus, this has not been carefully evaluated for conditions of plaque formation per se (Mittereder et al., 1996). The use of EGFP-expressing virus allowed systematic identification of all initially infected cells and tracking of their fates, and this approach uncovered the discrepancy between number of initially infected cells and ultimate plaque number. The use of AD-293 cells may have facilitated this analysis. In these cells it was possible to distinguish primary from secondarily infected cells due to the relatively long time between initial and secondary infection (3.5 days). In preliminary experiments we have tested infection of AdEGFPuci in the original 293 cells. In 293 cells the infection appeared to spread more rapidly, and more secondarily infected cells resulted from each initially infected cell. The reason for the difference between AD-293 and 293 cells is under investigation. One possibility could be that the levels of EIA and EIB proteins expressed in these cells differ but RT-PCR did not detect differences in levels of EIA transcripts (Hofacre and Fan, Unpublished results). One potential factor in these experiments could have been that repeated scanning of the infected cultures under UV light might have mutagenized the virus or infected cell, resulting in the discrepancy between number of initially infected cells and resulting plaques. However, the scanning was conducted rapidly at low magnification, which minimized the UV exposure to any given cell or virus. To directly address this possibility, duplicate infections of AdEGFPuci on AD-293 cells under plaque formation conditions were established; one set of cultures was monitored daily under UV for green fluorescence while the others were not. At 11 days post-infection both sets were assayed for plaques by GFP fluorescence. The titer of the AdEGFPuci stock measured on cells without UV monitoring was 5.3 × 10 8 pfu/ml, while the same stock measured on the cells with daily UV monitoring gave a similar titer of 4.9 × 10 8 pfu/ml, within the margin of experimental error. Therefore daily UV monitoring did not appreciably affect the efficiency of plaque formation. It was necessary to study infection of AdEGFPuci in EIA and EIB complementing cells such AD-293 because AdEGFPuci does not encode EIA and EIB proteins. The fact that AD-293 or 293 cells constitutively express EIA and EIB proteins could influence the rates of replication and spread compared to wild-type adenovirus infected into cells that do not express these proteins. In wild-type adenovirus infection EIA is the first viral transcript expressed, and the EIA proteins are required for transcriptional induction of other viral early proteins including those encoded by EIB. Thus AdEGFPuci might replicate more rapidly in AD-293 or 293 cells than wild-type virus infected into cells that do not constitutively express EIA and EIB. We have generated a recombinant adenovirus expressing EGFP as well as EIA and EIB (AdEGFPwt, (Hofacre and Fan, Unpublished results)). This virus replicates efficiently in AD-293 cells, and it also forms plaques in some but not all non-complementing cells (e.g. HeLa and MCF7, but not A549). Infection of MCF7 cells with AdEGFPwt also showed a discrepancy between number of initially infected cells and resulting plaques (ca. 26%), indicating that the results reported here reflect a more general phenomenon. In future experiments it will be possible to use the approaches described here to determine the parameters of early infection for a virus that resembles wild-type adenovirus in terms of EIA and EIB expression. The motivation for these studies was to understand the parameters of viral infection and spread for oncolytic adenoviruses such as dl1520. To this end we have also generated a recombinant

94

A. Hofacre et al. / Virology 423 (2012) 89–96

A

(1

o

3 1 day

o

2

3 1 day

o

3

Fate of Infection)

3 1 day

14d

0

Plaques

A

91 A 1 >2

97

S

24

1 265 A 2 >2

32 >2

S

17

S

37

37 78 (28.1

B

(1

o

3 1 day

o

2

3 1 day

o

3

4.0%)

Fate of Infection)

3 1 day

14d

0

Plaques

A

47 A 1 1

28

>2

S

4

A

97 2 7

>2

S

5

S

15

1 265

>2 15

24 (21.0

9.8%)

Fig. 5. Spread of AdEGFPuci early rounds of infections. (A) The fates of initial infection for experiments 1–4 (Fig. 2) are diagrammed, starting from the initially infected cells. The values in the box represent the number of infected cells, values below each box represent the number of infectious events observed in the boxed category, after each round of infection by AdEGFPuci in AD-293 cells. Percentages of the infected cells showing each fate are indicated over the arrows to each secondary infection category; the errors among the different experiments are indicated (standard error of the mean). (B) The spread of infection from a secondarily infected cell when there is only one cell secondarily infected cell. The flow chart is equivalent to that represented above in (A), except the fates of the secondarily infected single cells are expanded upon. The percentages of each fate for secondary singly infected cells were not statistically different from those for the primary singly infected cells in (A).

adenovirus expressing EGFP in place of EIB, but wild-type for EIA (Addl1520EGFPΔE3 (Hofacre and Fan, Unpublished results)). This virus replicates in 293 and AD-293 cells, as evidenced by CPE at high multiplicity infections and plaques at lower MOIs. As described in Fig. 6, we also analyzed the stochastics of early viral infection. The overall probability of extinction of infection from an initially infected cell infected at low MOI (one infectious virus per cell for any cell infected) was determined to be 0.72 ± 0.04 (Fig. 2). When the initially infected cell spread infection to one or two cells, the fates of those infections were consistent

with a stochastic null-model, in which the secondarily infected cells behaved like the initially infected cell (Fig. 5). However, when infection spread to >2 cells, the results were inconsistent with the stochastic null-model; in all of these cases spreading infection and a plaque resulted. A likely explanation was that when infection spread to >2 cells, some of those cells were infected by more than one virus (i.e. an MOI > 1). The amount of infectious virus produced per infected cell typically increases as MOIs increase (e.g. 1 vs. 10) (Mittereder et al., 1996), which would reduce the extinction probability p.

A. Hofacre et al. / Virology 423 (2012) 89–96

Probability of Plaque Formation

A

Materials and methods

1.0

Cell lines 0.8

0.6

0.4

2

4

6

8

10

No. of Secondarily Infected Cells

B

AD-293 cells used in these studies and were purchased commercially (Agilent Technologies — La Jolla, CA). AD-293s are derived from the commonly used HEK293 cell line, but have improved cell adherence and plaque formation properties. HEK293 cells are human embryonic kidney cells transformed by sheared adenovirus type 5 DNA (Graham et al., 1977). AD-293 cells, like HEK293 cells, express the adenovirus E region (EIA and EIB), allowing the production of infectious virus particles when they are infected or transfected with E1-deleted adenovirus vectors. AD-293 cells were grown in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS). AD-293 cells were cultured at 37 °C, 95% humidity, and 5% CO2.

Experiment 4 Adenovirus

4

Occurrence

95

3

2

1

0 1

2

3

4

5

6

7

8

9

10

No. of Secondarily Infected Cells Fig. 6. Probability of a spreading AdEGFPuci infection. (A) The probability of plaque formation when an initially infected cell spreads infection to different numbers of secondarily infected cells according to a stochastic null-model is shown (circles). The observed values are shown (squares). (B) The distribution of AdEGFPuci virus spread from primary infected cells to secondarily infected cells in grid squares where the number of secondarily infected cells was > 2 (Experiment 4, 16 squares).

Recently other researchers have also employed CRAds expressing expressing fluorescent proteins (e.g. EGFP or RFP), either as fusions to viral structural proteins (Le et al., 2006a, 2006b) or, as in this study, as independent proteins driven by the CMV promoter (Ono et al., 2005) to characterize virus infection. Efficiency of infection and spread in A549 lung cancer cells (Ugai et al., 2010), glioblastoma cell lines or primary tumor cells (Idema et al., 2010) , or in an orthotopic breast cancer tumor model (Borovjagin et al., 2010) were carried out. In particular Idema et al. (Idema et al., 2010) measured parameters for CRAd spread after initial infection, and they used the measured values to test mathematical models for oncolytic virus spread. The studies described here differ from the previous reports in that infections were conducted at extremely low MOIs, which insured that any infected cell was infected with only one infectious virus, and that surrounding cells were not infected by input virus. This allowed characterization of the spread of infection from an infected cell to neighboring uninfected cells. In contrast the other studies were either conducted at high MOIs (Ugai et al., 2010), or at MOIs of 1 but where neighboring cells were also likely infected (Idema et al., 2010). In summary, these studies have allowed elucidation of several key parameters and principles for establishment of spreading infections in the AdEGFPuci — AD-293 system. While the values of these parameters are likely to differ for other virus — cell systems, the same underlying principles are likely to apply. Analogous measurement of these parameters will provide insight into the spread of wild-type adenovirus, and also oncolytic adenoviruses such as dl1520 in more relevant target cells.

A recombinant adenovirus type-5 (Ad5) virus that expresses enhanced jellyfish green fluorescent protein (EGFP), in which the EGFP gene driven by the immediate early promoter of human cytomegalovirus (CMV) was substituted into the EIA and EIB regions (AdEGFPuci), was used in these studies (a gift from Drs. Toai Nguyen and Luis Villarreal, Center for Virology Research, UCI). Though removal of the viral sequences was necessary to produce a recombinant viral DNA within the packaging limits for adenovirus, AdEGFPuci will replicate in HEK293 cells (Graham et al., 1977). AdEGFPuci viral stocks, with titers of 10 9–10 11 pfu/ml, were generated from infected AD-293 cells and recovered from cultures (when extensive CPE was evident) by freeze–thaw three times followed by CsCl2 banding via previously established techniques (Tollefson et al., 2007). Adenovirus infections and plaque assays Adenovirus infections were performed as follows. 24 h prior to infections, 5 × 10 5 AD-293 cells were seeded onto a 60 mm cell culture dish with a marked grid of (2 mm squares)(Corning — Corning, NY). 24 h post-seeding (cell density of ~ 70–80% confluence) 1 ml of dilutions of AdEGFPuci stocks (~ 10 2–10 1 pfu/ml dilutions) were applied directly to the cell monolayers and incubated while rocking at 37 °C, 95% humidity, and 5% CO2 for 1 h. Subsequently, non-absorbed virus was aspirated, and the cell monolayer was overlaid with 3 ml of 0.65% agarose in 1 × DMEM/10% FBS. AdEGFPuci infected cells were cultured under the above conditions (without rocking) for at 8 days or longer, and the cells were fed by overlay with 2 ml–0.65% agarose/1 × DMEM/10% FBS every 3–4 days. Quantification of AdEGFPuci infection and plaque development AdEGFPuci infected cells were scanned under UV illumination daily (100 × magnification) using an Axiovert 200M phasecontrast fluorescent microscope (Zeiss, Germany). Each grid square was scored for the presence of green (virus-infected) cells and monitored daily for at least 8 days or longer. Squares showing AdEGFPuci infection (fluorescent cells) were recorded at 24 h intervals and compiled using Microsoft PowerPoint (see Fig. 2A). The data sets were for four experiments and the subsequent fates of each positive grid square are presented in Supplementary Data — Table S1. Supplementary materials related to this article can be found online at doi: 10.1016/j.virol.2011.11.014.

96

A. Hofacre et al. / Virology 423 (2012) 89–96

Acknowledgments We thank David Ornelless, Luis Villarreal, and Toai Nguyen for gifts of recombinant adenonvirus and advice. We would also like to thank Mina Kalantari-Dehaghi and Mary Frances Ypma-Wong for advice on the plaque assays. References Alemany, R., 2007. Cancer selective adenoviruses. Mol. Aspects Med. 28 (1), 42–58. Bischoff, J.R., Kirn, D.H., Williams, A., Heise, C., Horn, S., Muna, M., Ng, L., Nye, J.A., SampsonJohannes, A., Fattaey, A., McCormick, F., 1996. An adenovirus mutant that replicates selectively in p53-deficient human tumor cells. Science 274 (5286), 373–376. Borovjagin, A.V., McNally, L.R., Wang, M., Curiel, D.T., MacDougall, M.J., Zinn, K.R., 2010. Noninvasive monitoring of mRFP1- and mCherry-labeled oncolytic adenoviruses in an orthotopic breast cancer model by spectral imaging. Mol. Imaging 9 (2), 59–75. Chiocca, E.A., 2002. Oncolytic viruses. Nat. Rev. Cancer 2 (12), 938–950. Fort, J., 2002. A comment on amplification and spread of viruses in a growing plaque. J. Theor. Biol. 214 (3), 515–518. Fort, J., Mendez, V., 2002. Time-delayed spread of viruses in growing plaques. Phys. Rev. Lett. 89 (17), 178101. Garber, K., 2006. China approves world's first oncolytic virus therapy for cancer treatment. J. Natl. Cancer Inst. 98 (5), 298–300. Graham, F.L., Smiley, J., Russell, W.C., Nairn, R., 1977. Characteristics of a human cell line transformed by DNA from human adenovirus type 5. J. Gen. Virol. 36 (1), 59–74. Graham, F.L., van der Eb, A.J., 1973. A new technique for the assay of infectivity of human adenovirus 5 DNA. Virology 52 (2), 456–467. Hall, A.R., Dix, B.R., O'Carroll, S.J., Braithwaite, A.W., 1998. p53-dependent cell death/ apoptosis is required for a productive adenovirus infection. Nat. Med. 4 (9), 1068–1072. Haseltine, E.L., Lam, V., Yin, J., Rawlings, J.B., 2008. Image-guided modeling of virus growth and spread. Bull. Math. Biol. 70 (6), 1730–1748. Hofacre, A., and Fan, H. (Unpublished results). Fluorescent Imaging of a dl1520-like Conditionally-replicating Adenovirus. Idema, S., Dirven, C.M., van Beusechem, V.W., Carette, J.E., Planque, R., Noske, D.P., Lamfers, M.L., Vandertop, W.P., 2010. Objective determination of the oncolytic potency of conditionally-replicating adenoviruses using mathematical modeling. J. Gene Med. 12 (7), 564–571. Kirn, D., Hermiston, T., McCormick, F., 1998. ONYX-015: clinical data are encouraging. Nat. Med. 4 (12), 1341–1342. Komarova, N.L., Wodarz, D., 2010. ODE models for oncolytic virus dynamics. J. Theor. Biol. 263 (4), 530–543. Lam, V., Boehme, K.W., Compton, T., Yin, J., 2006. Spatial patterns of protein expression in focal infections of human cytomegalovirus. Biotechnol. Bioeng. 93 (6), 1029–1039.

Lane, D.P., 1998. Killing tumor cells with viruses—a question of specificity. Nat. Med. 4 (9), 1012–1013. Le, L.P., Le, H.N., Dmitriev, I.P., Davydova, J.G., Gavrikova, T., Yamamoto, S., Curiel, D.T., Yamamoto, M., 2006a. Dynamic monitoring of oncolytic adenovirus in vivo by genetic capsid labeling. J. Natl. Cancer Inst. 98 (3), 203–214. Le, L.P., Le, H.N., Nelson, A.R., Matthews, D.A., Yamamoto, M., Curiel, D.T., 2006b. Core labeling of adenovirus with EGFP. Virology 351 (2), 291–302. Lin, E., Nemunaitis, J., 2004. Oncolytic viral therapies. Cancer Gene Ther. 11 (10), 643–664. Liu, T.C., Kirn, D., 2008. Gene therapy progress and prospects cancer: oncolytic viruses. Gene Ther. 15 (12), 877–884. Mittereder, N., March, K.L., Trapnell, B.C., 1996. Evaluation of the concentration and bioactivity of adenovirus vectors for gene therapy. J. Virol. 70 (11), 7498–7509. Ono, H.A., Le, L.P., Davydova, J.G., Gavrikova, T., Yamamoto, M., 2005. Noninvasive visualization of adenovirus replication with a fluorescent reporter in the E3 region. Cancer Res. 65 (22), 10154–10158. Parato, K.A., Senger, D., Forsyth, P.A., Bell, J.C., 2005. Recent progress in the battle between oncolytic viruses and tumours. Nat. Rev. Cancer 5 (12), 965–976. Ries, S.J., Brandts, C.H., 2004. Oncolytic viruses for the treatment of cancer: current strategies and clinical trials. Drug Discov. Today 9 (17), 759–768. Russell, S.J., Peng, K.W., 2007. Viruses as anticancer drugs. Trends Pharmacol. Sci. 28 (7), 326–333. Tollefson, A.E., Wold, W.S.M., 1–9, 2007. Adenovirus Methods and Protocols, 2nd ed. Methods in Molecular Medicine, Vol. 1, pp. 1–9. 130,. Toth, K., Dhar, D., Wold, W.S., 2010. Oncolytic (replication-competent) adenoviruses as anticancer agents. Expert Opin. Biol. Ther. 10 (3), 353–368. Toth, K., Kuppuswamy, M., Shashkova, E.V., Spencer, J.F., Wold, W.S., 2011. A fully replication-competent adenovirus vector with enhanced oncolytic properties. Cancer Gene Ther. 17 (11), 761–770. Turnell, A.S., Grand, R.J., Gallimore, P.H., 1999. The replicative capacities of large E1Bnull group A and group C adenoviruses are independent of host cell p53 status. J. Virol. 73 (3), 2074–2083. Ugai, H., Wang, M., Le, L.P., Matthews, D.A., Yamamoto, M., Curiel, D.T., 2010. In vitro dynamic visualization analysis of fluorescently labeled minor capsid protein IX and core protein V by simultaneous detection. J. Mol. Biol. 395 (1), 55–78. Wodarz, D., 2003. Gene therapy for killing p53-negative cancer cells: use of replicating versus nonreplicating agents. Hum. Gene Ther. 14 (2), 153–159. Wodarz, D., Komarova, N., 2009. Towards predictive computational models of oncolytic virus therapy: basis for experimental validation and model selection. PLoS One 4 (1), e4271. Wong, H.H., Lemoine, N., Wang, Y., 2010. Oncolytic viruses for cancer therapy: overcoming the obstacles. Viruses 2 (1), 78–106. Yin, J., McCaskill, J.S., 1992. Replication of viruses in a growing plaque: a reaction– diffusion model. Biophys. J. 61 (6), 1540–1549. You, L., Yin, J., 1999. Amplification and spread of viruses in a growing plaque. J. Theor. Biol. 200 (4), 365–373. Zurakowski, R., Wodarz, D., 2007. Model-driven approaches for in vitro combination therapy using ONYX-015 replicating oncolytic adenovirus. J. Theor. Biol. 245 (1), 1–8.