Viability testing and characterization of ... - Wiley Online Library

Viability Testing and Characterization of Germination of Fungal Spores by Automatic Image Analysis G. C. Paul, C. A. Kent, and C. R. Thomas' SERC Centre for Biochemical Engineering, School of Chemical Engineering, University of Birmingham, Edgbaston, Birmingham, B 15 2 T , United Kingdom Received August 4, 1992/Accepted December 21, 1992

Fungal spores are used in the laboratory for culture maintenance and at laboratory and other scales as inocula for fermentations. The spore swelling and germination processes constitute a major part of the lag phase, and the subsequent culture morphology and productivity can be greatly influenced by the initial concentration and condition of the spores. An image analysis method has been developed for assessing the viability and the germination characteristics of fungal spores in submerged cultures. Structural variations during germination, i.e., swelling, germ tube formation, and germ tube elongation, are measured in terms of distributions of spore volumes and of germ tube lengths and volumes. These measurements are fully automatic and give a very rapid assessment of spore viability. This image analysis method might be used as a tool in culture maintenance and for determining the quality of inocula for fungal fermentations. 0 1993 John Wiley & Sons, Inc. Key words: spore spore viability germination morphology image analysis Penicilliurn chrysogenurn

INTRODUCTION For consistency of fungal fermentations it is important to maintain a uniform inoculum, because once a mycelial structure has formed, it remains a discrete entity for long periods, extending and growing older. Early events can thus markedly influence the morphological development of these fermentations. This problem is compounded in commercial production where multiple steps of increasing size are used, leading to potential amplification of errors made in earlier stages of inoculum development. A low concentration of spores in an inoculum can limit germination.' This effect has been attributed to a deficiency in each spore of some metabolite vital for triggering the germination mechanism. The spore concentration must be increased until a certain level of this compound in the diffusate is reached, if good germination is to be achieved. Inoculum size can also influence growth, culture morphology, antibiotic production, and enzyme content of mycelia, as demonstrated by Smith and Calam" with Penicillium chrysogenurn. The differences induced by different inoculum spore concentrations were shown to be * To whom all correspondence should be addressed.

Biotechnology and Bioengineering, Vol. 42, Pp. 11 -23 (1993) 0 1993 John Wiley & Sons, Inc.

irreversible. It was shown that growth and penicillin production increased with the starting spore concentration in the inoculum, up to an optimal level for their shake flask experiments of 5 X lo3 spores mL-'. However, the usefulness of quoting an optimal spore concentration is limited unless the percentage viability of the spores is known. For culture maintenance and inocula preparation an assessment of the quality of spore preparations is, therefore, required. Inoculum quality might be greatly improved if not only the total concentration of spores in a preparation but also their percentage viability could be measured, ideally in the germination medium of the process. For determining the percentage viability of spore preparations, the plate count technique has been a useful tool. This is based on the assumption that when a dilute suspension is spread on a suitable medium, each individual spore will grow and produce an isolated colony. However, there are some basic problems associated with this method such as underestimation of percentage viabilities (because not all spores may form visible colonies on some sporulation media or because there might be more than one spore per colony), lack of reproducibility (due to uneven distribution of spores with varying numbers per colony), and long incubation times (2-4 days). The outcome depends largely on the composition of the medium as well as on the other growth conditions employed. Clearly a simpler and more rapid but still reliable alternative method would be valuable, especially if the spores were germinated in the fermentation medium rather on some other (solid) medium. The germination characteristics of fungal spores have been discussed by G ~ t t l i e b In . ~ the germination process, the dormant spore shifts from low to high metabolic activity. This process starts automatically if it is placed in suitable environmental conditions. Dormancy can also be broken by an activation process such as heat shock or by chemical treatment. Whatever the method of initiation, three basic structural changes during germination can be recognized by microscopic observation: spore swelling, germ tube emergence, and germ tube elongation. With the onset of germination, the spore begins to swell to several times its dormant diameter and a germ tube emerges. Usually a spore is considered germinated if the length of this germ tube reaches one-half of the largest dimension of

CCC 0006-3592/93/01011-013

the spore.5 The swelling and germ tube emergence in the germination process constitute a major part of the lag phase in fermentations inoculated with spores. Spore germination under different environmental conditions has been investigated by many authors using photomicroscopic method^.^,'^.'^ Trinci12 has shown that the germ tube growth occurs at an exponential rate after the onset of germination. The specific growth rate during the exponential phase may be up to 9 times that for normal mycelial growth in the equivalent liquid medium, because of endogenous spore reserves supplementing nutrients from the medium. From a practical point of view, measurements of such rapidly changing morphological events could be used to characterize germination and to determine the quality of the inoculum. A n automatic image analysis method is described here for determining the viability of fungal spores and assessing their germination characteristics. In such a method an image of spores on a microscope slide is digitized in both space and tone, so that it can be processed by a computer to extract interesting features, which can then be analyzed to obtain any measurements required. Automatic image analysis methods for morphological characterization of filamentous microorganisms have been r e p ~ r t e d . ~ .Image ~,’~ analysis has also been applied to measure the early growth and branching processes of Streptomyces tendae.’ However, the morphological changes occurring to spores during their germination have been overlooked despite their potential as an important tool for inoculum improvement. Structural variations during germination, i.e., swelling, germ tube formation, and germ tube elongation, are also characterized by the program.

SOFTWARE DESCRIPTION An overall flow diagram of the software is given in Figure 1.

The program consists of several phases, which are described below. Each consists of a combination of image processing operations, chosen to achieve a particular result. These operations are available on, or can easily be implemented on, many commercial image analyzers, although the speed of operation might vary with choice of hardware.

Setting Up Two types of parameters can be set by the user: (a) hardware parameters, and (b) image processing parameters. The former include the microscope lamp brightness, autofocus parameters, the calibration factor for the objective to be used, and control parameters for an automatic microscope stage (number and position of slides and total number of fields in each slide). Image processing parameters, such as detection levels and limits on some size and shape factors (the use of which is described below), are set, as well as other parameters concerning germ tube identification and elimination of debris. During setting up, correction for uneven slide illumination (“shading correction”) can also be

12

established, and manual editing selected if desired. Finally, an active measuring frame might be set up to prevent bias due to truncation of objects at the image edge. This phase is required only once for each set of measurements. Those described below are then executed in repeated cycles until all the selected slides and fields of views have been analyzed.

Image Processing An image of spores is captured from a monochrome video camera (e.g., Fig. 2a). In the present work images were of 512 X 512 pixels, each with a grayness (brightness) level from 0 (representing black) to 255 (white). Such images are of course two-dimensional projections of the objects in the microscope field of view. Dark objects such as spores are detected by their grayness level. The result of this detection is a binary image masking the parts of the gray image that are of interest, such as spores and germ tubes. As the grayness level of spores is not required for the analysis, all subsequent image processing is on this binary image. To avoid edge effects, only the portion of the binary image inside an active measuring frame is considered for subsequent image processing (image A, Figs. 1 and 2b). Image A may also contain many undesired objects such as debris (particularly dead mycelia), small dust particles (see Fig. 2b), or solids if a solid-containing medium (e.g., corn-steep liquor medium) is in use. The latter might of course have a wide range of sizes. Most of the unwanted objects are removed by two erosions. An erosion is an operation by which a layer of pixels is removed from the boundary of the binary image of the objects.” The first erosion removes only a few pixels (2 pixels for the test fermentation), and this eliminates smaller objects. The image is then rebuilt to its original form, excluding those small extraneous objects. This first erosion leaves spores, germ tubes, large unwanted debris, and a few small unwanted objects which cannot be removed because they are similar in size to spores (see Fig. 2c). The large debris can be identified by a larger erosion (typically 16 pixels for the test fermentations) which removes the desired objects and similar sized debris, leaving only the cores of the large debris. After rebuilding the latter to its original size, it can be subtracted from the earlier image, leaving a new image which contains spores and germ tubes and some spore-sized debris (image B). Objects in image B are then classified on the basis of a preset circularity parameter into definitely nongerminated spores (image C , Fig. 2d) and everything else, including germinating spores, i.e., spores with germ tubes (image D, Fig. 2e). The spores in image C are stored in image J (see Fig. 2f), which will eventually contain all the nongerminated spores. Image D may also contain nongerminated spores artifactually attached to germ tubes of other spores, and nongerminated spores touching one of those germinating, as well as debris. As described in detail below, touching objects and the germ tubes of the germinating spores in image D are separated by opening

BIOTECHNOLOGY AND BIOENGINEERING, VOL. 42, NO. 1, JUNE 5, 1993

I

I

I

lmane setuo

t Parameter setup

1

Image capture

I

DetecUon

I

I

I I

C5I3 Remove debris

Classification by

I

1.

Any germ tube without spores?

Ye.

no

t Any spore without

t

-

Perm tube?

I

I

1

kre lengths k circularityl

E5zI Calculations

Disk archive File storage

New sllde/new field

t Data analysis

Figure 1. Flow diagram of stages involved in characterization of fungal spore germination.

operations combined with selections using preset size and shape factors. An opening is an erosion followed by its reverse operation dilation (by which pixels are added to the periphery of an object). If the shape of an object is complex, e.g., it has a narrow neck or protuberances on its surface, such complex features will be irreversibly removed by the erosion. The

subsequent dilation will not be able to reconstruct the lost boundary details. Initially, germ tubes and spore bodies of germinating spores (“germ tube spores”) are separated by such an opening. The relatively narrow germ tubes are lost, leaving behind the wider germ tube spores (image E). Subtraction of the identified germ tube spores from image D gives the germ tubes of those spores (image F).

PAUL, KENT, AND THOMAS: IMAGE ANALYSIS OF SPORE GERMINATION

13

Figure 2. Spores of P. chrysogenum at different image processing steps (bar length 50 pm): (a) captured gray image (at 16 h, spores S1 and medium M2); (b) image A, binary image of original gray image in (a); (c) image B, binary image after removing small debris; (d) image C, binary image of nongerminated spores obtained by circularity measurement; (e) image D, binary image of germinating spores, connected nongerminated spores, and debris; (f) image J, binary image of nongerminated spores; (g) image K, binary image of germ tube spores; (h) image L, binary image of germ tubes; (i) Image L + K, binary image of germinated spores (germ tubes + germ tube spores).

A fixed size of opening would not be satisfactory because germinating spores of many sizes can be present. Too large an opening might remove relatively thin germinating spores as a whole, while too small an opening might not be adequate to remove thicker germ tubes. Therefore a multistage filtering approach has been incorporated in this phase where thinner germinating spores are bypassed by the initial preset maximum sized opening (preset at 6 and 8 pixels for the early and late stages of fermentation, respectively) and recycled in subsequent openings of de-

14

creasing size. This operation is continued until a lower preset limit (3 pixels for the test) is satisfied. The detailed algorithm of this process is given in Figure 3. Eventually images are obtained containing germ tube spores (image E), germ tubes (image F), and debris (image Xl). The germ tube spores (image E) obtained by the previous operations may sometimes remain attached to nongerminated or other germinated spores. The image is segmented to separate these connected objects, giving nongerminated and germ tube spores (image G). Segmentation is an op-


n (parameter)

I

1 I

i > i-limit?

I

1

Figure 3. Flow diagram of separation of germ tubes from germinating spores.

eration which disconnects touching objects. Image F, containing germ tubes, may contain many other objects, such as broken or dead hyphae, debris, etc. By superimposing this image and image G (of spores), objects originally attached to spores can be identified as possible germ tubes (image H), while the remainder are classified as false tubes, e.g., debris (image X2). Shape (circularity) and length criteria are applied to the objects in image H for the final identification of the germ tubes, which are relatively long, thin, and far from circular, whereas (for example) small spores attached to bigger ones are near circular. A lower length limit of germ tubes is also applied and can be justified as spores are only considered to have germinated when their germ tubes attain a length equal to or greater than half of the longest spore dimen~ion.~ This criterion also usefully eliminates small debris touching a spore. Through these operations, image L (Fig. 2h) is generated from image H and contains a binary image of germ tubes, Table I. Medium

M1 M2 M3 M4

while the rest of image H is stored in image I, which contains false tubes and spores. The spores are separated out from image I using a shape (circularity) filter, as spores are more circular than the false tubes. These small spores are nongerminated and are stored in image J (Fig. 2f) with previously identified nongerminated spores. Finally, the germ tubes in image L can be used to separate the spores in image G into nongerminated and germ tube spores, as each of the latter must be associated with a germ tube. The germ tube spores are held in image K (see Fig. 2g). The nongerminated spores obtained by this process are then also added to image J. Figures 2f, 2g, and 2h show typically binary images of the nongerminating spores, germ tube spores, and germ tubes of the germinating spores, respectively, identified from the original gray image of Figure 2a. During the process of image analysis an optional manual editor can be used if selected during setting up. This

List of media used for P. chrysogenurn germination. Reference and modification Deo and Gaucher4 Medium M1 with 0.2 g L-' Fe(NH4)2(S04)2 . 6H2O and supplemented with 0.6 g L-' of Na-EDTA Mou and Cooney6 Solids of M3 removed and supplemented with 1.0 g L-' Na-EDTA


15

editor might be used for removing unwanted objects in the image by using a mouse to select and exclude them from the subsequent image processing or for separating debris or a nongerminated spore artifactually attached to a germ tube of a germinating spore by drawing a line on the binary image between the connected objects. However, such artifacts are expected to be rare in defined media or any solid free medium, and manual editing slows down the image processing. The editor might be more useful when the medium contains solid particles.

Measurement and Calculations Primary measurements of individual objects in the images are perimeter and projected area. The raw data are then converted to diameter, length, and volume. Equivalent diameter D , and volume V, for nongerminated spores and germ tube spores can be calculated using the following expressions, which assume the spores are spherical and their two-dimensional projections circular:

Worthing, United Kingdom. One-millileter aliquots of frozen spore suspension containing ca. 3 x lo7 to 5 x lo7 spores mL-' were thawed at room temperature and then inoculated into 1-L Roux bottles containing 200 mL of sporulation medium (SmithKline Beecham Pharmaceuticals), Fifteen to 20 3-mm-diameter sterile glass beads were added to the bottles, which were tilted gently to and fro to disperse the spores uniformly on the agar surface. The cultures were incubated at 25°C for about 6-8 days to complete development of spores and then stored at 4°C. The spores were washed off from the sporulation mat using 60 mL 0.1% (v/v) Tween 80 with gentle agitation using glass beads. The conidial suspension was decanted into a sterile bottle and sonicated for about 15-20 min in an ultrasonic water bath (L&R Manufacturing Co., Kearny, NJ) to break up chains of conidia. Two spore stocks were used to test spore viabilities and germination characteristics. Stock S1 was fresh spores collected after 6-8 days of incubation while stock S2 was obtained from a 40-day-old culture.

Shake Flask Culture

where P is the perimeter. From Figure 2h it appears that germ tubes might be tapered, shaped like truncated cones. For the germ tubes found in this study it was shown that the assumption of cylindrical rather than truncated conical shape could not have caused errors in the estimated volumes of more than 5%. The parameters used to characterize the progress of germination are those describing the distributions of volumes of both nongerminated spores and germ tube spores and of term tube lengths and volumes.

The defined medium of Deo and Gaucher4 (medium M1) and the complex medium of Mou and Cooney6 (medium M3) were used to test the method. In order to investigate the effect of medium ingredients on germination, the Fez+ concentration of the defined medium M1 was increased twofold to give medium M2. Sodium-ethylenediaminetetraacetic acid (Na-EDTA) was also added to prevent precipitation. Medium M4 was prepared by removing the corn-steep solids from medium M3 by passing it through a glass microfiber filter, Whatman grade GF/A (Whatman Int., Maidstone, UK). Again Na-EDTA was incorporated to prevent precipitation during sterilization. Table I summarizes the media. For each germination medium 10 mL of the spore suspension was inoculated into a 500-mL conical flask containing 50 mL of that medium to give 5 X lo6 to 6 X lo6 spores mL-' (the standard concentration range for inoculation of this fungus). Five milliliters of 20% (w/v) glucose solution (sterilized separately) was added to the shake flask, which was then incubated in a shaker at 200 rpm, 2.5 cm stroke at 25°C. Approximately 1 mL sample was collected from the culture flask every 4 h from 8 h after inoculation and fixed with 2 drops of lactophenol cotton blue strain (BDH Chemicals Ltd, Poole, England). The samples were stored at room temperature for up to 7 days. No change in the germination characteristics was observed for this period of storage.

MATERIALS AND METHODS

Counting Chamber/Slide Preparation

$70% v, = 6

where A is the projected area, D , is the equivalent diameter, and V , is the volume of the spores. If the shape of a germ tube is considered to be cylindrical, its mean diameter D , and length L , can be derived from the projected area and perimeter measurements using the following expressions:

D,

=

Lt

=

P - J m(3) 4

P

+ J P 2 - 16A

(4) 4 and the volume of the germ tube can be obtained from

rD,A v, = 4

Sporulation Culture An industrial production strain of Penicillium chrysogenum

was obtained from SmithKline Beecham Pharmaceuticals,

16

A Helber counting chamber (Weber Scientific Int., Middlesex, UK.) with a depth of 20 p m was used to estimate the spore concentrations. The chambers were supplied without grid lines to ease the image processing. When germination


Table 11. Parameter values used for the image analysis of P. chrysozenurn germination. Parameters Spore diameter limits, p m Minimum Maximum Germ tube length limits, p m Minimum Maximum Circularity limits Spores Germ tubes

Typical values

Values preferable at early stage

2-3 5-16

5-8

2-3 5-16

2-3 10-50

2-3 10-20

2-3 10-50

51.2 >1.2

51.2 >1.2

51.2 >1.2

was significant, the culture was diluted with an equal volume of distilled water before analysis. Approximately 0.04 mL of the (diluted) culture was placed on the counting surface. The chamber was closed with a thick optically flat coverslip.

Image Analysis of Spores For image analysis, a Quantimet 570 image analyzer (Leica Cambridge, Cambridge) connected by a Sanyo CCTV camera (Sanyo Electric Co., Basel, Switzerland) to a Polyvar optical microscope (Reichert Jung, Optische Werke AG, Wien, Austria) was used. Slides of samples to be analyzed were placed on an eight-slide stage mounted on the microscope which was drivable in three dimensions with 2.5 p m step size in the x - y directions and 0.25 step size in the vertical direction. The magnification used was

Values preferable at late stage

2-3

200 times, which corresponded to an interpixel distance of 0.538 p m . For automatic image processing and analysis a computer program was developed using Turbo C 2.0 (Borland, Scotts Valley, CA). The C-library ADSOFT 1.01 (Leica Cambridge, Cambridge, UK) provided Quantimet image analysis functions. Typical values of the parameters used in the characterization are listed in Table 11. These were set using samples from a preliminary germination experiment and could then be used for other experiments with the same organism and medium. Settings for the lower limit of diameter and for the maximum circularity depend on the type of spores in use and can be obtained from a preliminary experiment on unswollen spores. To discriminate between nongerminated spores attached to some other spore and a genuine germ tube, the minimum circularity for germ tubes was set equal to the maximum for unswollen spores. The parameter for

Figure 4. Gray images of P. chrysogenum taken at 20 h germinated in medium M2 with spore stock (a) S1 and (b) S2. Photographs were directly taken from image analyzer display. Bar length 50 p m .


17

Table 111. Comparison between automatic and manual editing methods with solid containing corn-steep liquor medium (M3). Incubation Time (h) 8 12 16 20 24 a

Spore concentration (mL-'

x lob)

Automatic

Manual

3.5 4.1 4.7 3.9 0.38d

4.9 5.4 6.2 5.7 0.44a

Automatic

Manual

Automatic

Manual

13.2 53.0 81.4 95.0 93.0b

9.7 39.5 70.3 91.5

15 18 22 21 12

38 47 42 36 28

w.nb

Nongerminated spores only. Percentage of germinated spores calculated using an initial spore concentration of 5.5 X 10' spores mL-'.

the lower limit of germ tube length was set to eliminate artifacts caused by small debris touching the spores. The four parameter values described above could be kept constant throughout a germination test on P. chrysogenum spores. The other two parameters were settings for the maximum spore diameter and the maximum length of a germ tube and were found by a preliminary trial. These parameters are used to distinguish between spores, germ tubes, and debris. It was found that the accuracy of the measurements could be improved if different values of these parameters were used in the early and late stages of germination (Table 11). This was particularly important when large amounts of debris and media particles were present. Starting from a default value, 3-5 iterations were usually needed for the adjustment of one of these parameters; for two parameters and a maximum total of 10 iterations the tuning time was less than 10 min. A total of 400 spores (nongerminated + germinated) were analyzed for each sample. The total number of fields for analysis of this number of spores was in the range of 25 to 32, 17 to 13 spores per field respectively. The samples taken toward the end of the germination often contained long branched and entangled hyphae. The measurements on individual germ tubes could not be performed on these samples, and only nongerminating spores could be counted. In order to estimate the proportion of these spores (and by difference the proportion of germinated spores), a fixed number of fields was analyzed. Using the average of the total number of spores per field measured on earlier samples, an estimate could be made of the total number of spores in that number of fields, and the proportion of the different spore types could be estimated. For both the defined medium M2 and the solid-containing corn-steep liquor medium M3 the automatic method and a manual editing method using the optional editor were compared. In the manual editing method, the spores were separatedldisconnected from debris andlor corn-steep liquor particles by marking the latter on the image using a mouse, so that they could be eliminated before subsequent image analysis. Following the measurement of all the fields of view for a particular sample a statistical analysis of the data was performed to give means, standard deviations, and histograms of the parameters. The volume of the sample analyzed was obtained by converting the total number of

18

Analysis time (min)

Percent germiantion

fields analyzed (giving the area of sample measured) and the hemocytometer chamber depth into volume. Concentrations of the germinating and nongerminated spores were calculated. Volumes of the individual spore types and germ tubes per unit volume of the medium were also obtained.

RESULTS The freshly prepared spore stock (S1) and the 40-day-old spores (S2) did not germinate identically. Figure 4 shows samples taken at 20 h from shake flasks containing defined medium M2. The spores shown in Figure 4a were grown from stock S1, whereas those of Figure 4b were grown from stock S2. It is clear that the latter did not germinate as well as the former, indicating that the germinability of the spores deteriorated with the duration of storage. These very different spore preparations combined with the range of media used gave an excellent test of the image analysis method. Errors found in the estimates of spore concentration (number of spores per milliliter of inoculum) depended on whether the medium contained undissolved solids or was solid-free. With defined media M1 and M2 and solid free corn-steep liquor medium M4, the percentage error in a single concentration measurement was found by multiple slide analysis to be less than 8.0% (in all cases regardless of spore type), when at least 400 spores (nongerminated + 100

90 0 80

Spores S I Medium M41

0 Spores S2 Medium M2 0 Spores S2 Medium M4

E:

2 70 4

.,C 60

,EiL a 20 10 0

'

-//& 1 40

0

5

10

15

20

25

30

35

Time (h)

Figure 5. Time course of germination of P. chyrsogenum spores in different inoculum media and using two ages of spores.


I 50

I

n

/. A

E 2135 u

0

0

:I20 m

0

?pores S I T c G i m

MI

Spores SI Spore, SI Spores S2 Spores SZ

MZ

Medium Medium Medium Medium

-~

M4 MZ Y4

3 u

$105

.-c E

: 90 w , C

g

75

L

0

Q

60

P

49

-c m C

2

30

5

10

15

20

25

30

35

40

Time ( h )

Figure 6 . Time course of mean volumes of nongerminated spores of P. chrysogenum in different inoculum media and using two ages of spores.

germinated) were analyzed per sample. For the percentage of spores germinated, the maximum errors were 4.0 and 6.5% with defined media (M1 and M2) and solid-free complex medium (M4), respectively. The larger error with M4 medium was due to the presence of precipitated materials that appeared during sterilization. Through the comparisons that were made of the automatic and manual editing methods, systematic errors in the former could be estimated. When the concentration of the original spore suspension was counted by both methods, the difference was found to be 7.2% (manual method greater). This error was due to the presence in the original suspension of relatively many spores in conidial chains, the separation of which by automatic image analysis was poor. Comparisons were also made on three samples from an S1 inoculated medium M2 shake flask culture, taken at 16, 20, and 24 h. The percentage errors were all less than 4.4% in these cases, with corresponding errors for the percentage germinated spores less than 5.0%. The comparison has also

Time ( h )

Figure 7. Time course of mean volumes of germinated spores of P. chrysogenum in different germination media and using two ages of spores.

been made with solids present in the medium (corn-steep solids in medium M3). Table 111 summarizes the results for five samples from a shake flask culture. The automatic analysis resulted in systematic errors in the range of 5% to 36% (compared to the results of the manual editing method). In this case the errors were due to spores being attached to solid particles, from which separation by automatic image analysis was sometimes not possible. However, the percentage errors were less for the late samples (after 20 h) because the spores in those samples had swelled and germinated and were more distinct. Whatever medium was being used, the time of analysis for one sample using automatic analysis varied between 10 and 22 min depending on the complexity of the samples, i.e., on the presence of debris, or spores connected in conidial chains. Using the method with manual editing the analysis time was twice as long. Figure 5 shows the percentage of germinated spores during incubation in the three different media and with two ages of spores. The media had a marked influence on germination. The proportion of the spores which formed germ tubes and the time of germ tube formation were also affected by the medium. Table IV summarizes the germination characteristics obtained by image analysis. It shows that at 24 h, 89% of the spores had germinated in the complex medium M4, compared to only 16% in the defined medium M1. When the concentration of Fe2+ salt in M1 was increased twofold (to give medium M2), the germination level increased from 16% to 33%. The defined media M1 and M2 required longer incubation times for germination than the complex medium. For example, 50% spore germination took 36 h in medium M2 but only 13 h in complex medium M4. With medium M1 the germination was only 21% in 36 h. Figure 5 also shows that germinability of spores deteriorated significantly with storage. After 24 h of incubation the 40-day-old stock (S2) had only germinated 21% and 45% in media M2 and M4, respectively, compared to 33% and 89% germination with fresh spores (stock S l ) . Achieving given germination levels for the old spores required increased incubation times compared to fresh spores (Table IV). The mean volumes of nongerminated and germ tube spores during germination are presented in Figures 6 and 7, respectively. Spherical growth of the spores progressed with continuing incubation. The media had a marked effect on spore swelling (growth) and their eventual size, with higher growth in the defined media. In the defined media (M1 and M2) spores increased continuously in volume. The mean volumes of the nongerminated spores changed from 30 to 150 ,urn3 over 36 h, while the germ tube spores went from 120 to 350 ,urn3. The spores swelled approximately 4 to 12 times before forming germ tubes. In the complex medium the spherical growth is significantly less than in the defined media (Figs. 6 and 7). The maximum mean volume of the nongerminated spores was 65 pm3 and that of the germinated ones 110 ,urn3. The mean volume of the


19

Table IV. Comparison of germination characteristics of P. chrysogenum spores grown in different media and using two ages of spores.

Medium

Spore type

M1 M2

s1

Incubation time for 50% germination, h

Germ tube specific growth rate p, h-’

a

0.155 0.193 0.187 0.226 0.194

16 33 21 89 45

s1 s2 s1 s2

M4

a

Percentage germination at 24 h

36 d

13 26

Total germination was less than 50%

germ tube spores was approximately constant after 8-12 h of incubation. The mean volumes of spores alone might not be adequate to represent the swelling process, as each sample contains a wide range of spores sizes, whether germinated or not.

Image analysis can also provide the distributions of spore sizes as shown in Figures 8 and 9 for defined medium M2 and complex medium M4, respectively, with spore stock S1. In M2 the sizes of the nongerminated spores were widely distributed, while in M4 there was a narrower size

250

200

( a ) 12 h

(b) 16 h

h

0

150

iff? Q,

100

k

r+

50

0

,60

180

300

420

540

660

60

180

300

420

540

660

n

J.

Volume ( p m ) 250

200

(d) 24 h

( c ) 20 h

h 0

150

iff? k

100

k 50

0

f 2 m n

60

180

300

420

540

660

60

180 3,

300

420

540

e

660

Volume ( p m )

Figure 8. Comparisons of distributions of nongerminated and germinated spore volumes of P. chrysogenum grown in medium M2 using spores germinated spores. S1: . I )nongerminated spores;

20


250

200

(b) 12 h

h 0

150

Q)

3 ZlOO I 4

L 50

0

60

180

300

420

540

660

60

180 3

300 420

540

660

Volume ( p m ) 250

200

( d ) 20 h

(c) 16 h

h 0

150

$CY Q)

L k

100

50

0

60

180

300

420

540

660

60

-180

300

420

540

660

Figure 9. Comparisons of distributions of nongerminated and germinated spore volumes of P. chrysogenum grown in medium M4 using spores S1: . I ) nongerminated spores; germinated spores.

range. The latter indicates that possibly corn-steep liquor contains some agent which stimulates spore germination as soon as a spore has swollen to a critical size. Besides spore characteristics, germ tube lengths of the germinating spores could also be found by image analysis. Figure 10 shows the mean length of the germ tubes during germination in different media with both spore stocks (S1 and S2). Mean germ tube length increased with germination time. Differences of germ tube growth with the different medium and spore ages are evident in this figure. The complex medium resulted in a more rapidly growing culture than was found with the defined media. Volumes of the germ tube material per unit volume of the inoculum culture were obtained by image analysis (Fig. 11). Table IV compares the corresponding specific growth rates of the germ tubes for the different media and

spores. With the complex medium M4 the specific germ tube growth rate was over 45% greater than with M1. With the older spores the specific growth rates decreased significantly.

DISCUSSION A n image analysis method has been described for deter-

mining the viability and the germination characteristics of fungal spores. It offers a number of advantages over photomicroscopy or colony counting (on solid medium): it is rapid, is more accurate and consistent, can discriminate between nongerminated and just germinated spores, and in particular can be used on spores germinating in the actual submerged fermentation medium. When the method is applied to samples from submerged cultures, it pro-


21

'

A

Spores S 1 Medium M I Spores S l Medium MZ Spores S l Medium M 4

0 Spores SZ Medium MZ

0

5

10

15

20

25

30

35

I 40

Time (h)

Figure 10. Time course of mean germ tube lengths of germinated spores of P. chrysogenum in different germination media and using two ages of spores.

vides measurements on spore preparations that are very appropriate for assessment of the quality of an inoculum. The manual editing method provides accurate and precise measurements, while the automatic method is usually of acceptable accuracy and is very fast. Although the method has been tested using spore preparations of P. chrysogenum, it should be applicable to spores of other fungi and of similar speed and accuracy provided the parameter values used to characterize spore germination (as listed in Table 11) are properly set. The method gave a satisfactory performance in eliminating debris and artifacts from the images when spores were germinated on defined or solid-free complex media. With a solids-containing medium the performance of the fully automatic method was seriously degraded. It is for this reason that manual intervention through screen editing, as used in the manual editing method, has been made available in the software. Such manual intervention might be chosen at any stage during analysis, particularly for fields containing large quantities of debris and media particles. However, use of editing would significantly reduce the speed of analysis. Each potential user of the method could check the accuracy of the automatic method by using the editing facility and choose a balance between speed and accuracy that is suitable for his or her application. Even with manual editing, the method described here is a significant advance on alternatives. Figure 5 shows how the method might be used to investigate the relationship between incubation time and percentage of germination with different media and different spore preparations. From Figure 5 it can be seen that there is a significant effect of culture medium on spore germination and on the germination lag. The complex media (M3 and M4) yielded high percentage of germinations at a faster rate than did the defined media (M1 and M2). The method can therefore be used to select or formulate media for inoculum preparation that will give highest and most rapid germination rates possible.

22

Figure 5 also shows that spore viability decreased significantly due to storage for a period of 40 days at 4°C on agar. Such losses of viability, and changes in germination characteristics with the medium used, are important where exceeding a critical concentration of spores in the inoculum is important to the development of a fermentation. In order to achieve consistent inocula, the proportion of nonviable spores in each inoculum should be taken into account to ensure there is a necessary concentration of viable spores. The program makes such determinations possible. Systematic investigations are also possible to determine loss of spore viabilities with storage time, temperature, and the preservation method used. Image analysis thus provides a versatile tool for determining the quality of spore preparations and for testing the effectiveness of preservation techniques for the culture maintenance. Figures 6 to 9 show how the program can also be used to investigate the pregermination and germination processes. Both spherical growth (swelling) and germination are influenced by the medium used as well as the age of the spores. With the defined media (M1 and M2) spore germination is associated with larger spherical growth of spores than that observed in the complex media (M3 and M4). It appears that in the nutrient-rich corn-steep liquor media (M3 and M4) the viable spores germinated as soon as they reached a certain size, but this did not occur with the defined media, which allowed swelling to continue to larger sizes. The program will permit detailed investigation of the relationship between swelling and germination of spores. Besides characterizing swelling and germination, the method provided the mean length of germ tubes of the germinated spores (Fig. 10) and might thus be used to study early hyphal development from spores. With Aspergillus niger spores Anderson and Smith2 observed a correlation between the degree of spherical growth and branching of the germ tube. Larger spores possessed thicker germ tubes and a greater degree of branching, whereas lower

0.8

1

Spores S1 Medium

0.7

MZ

Spores S l Medium M4 0 Spores S2 Medium MZ

-1

0

0.6 0 v

D

0.5

.

0.4

.

0.3

.

I

3

c

/

i 01 L.

0

g 0.2

-

3 0

> 0.1 n n . -.~,

0

5

10

15

20

25

30

35

40

Time (h)

Figure 11. Time course of volume of germ tube materials per unit volume of culture of P. chrysogenum grown in different media and using two ages of spores.


spherical growth resulted in thinner, elongated, and relatively unbranched germ tubes. An earlier report' showed that when the nitrogen source of the defined medium was substituted by corn-steep liquor, thin and long hyphae with less branching resulted. In the present study comparatively less spherical growth was observed with corn-steep liquor medium, which is consistent with these two previous observations.',' For complete studies of early hyphal extension and branching phenomena the present program might be used in conjunction with that of Tucker et al.13 This would have the advantage over the alternative method of Reichl et al.9 of being usable on samples from submerged fermentations rather than on a single spore in a growth chamber on a microscope.

CONCLUSIONS It has been demonstrated that the new image analysis method proposed here can be used for determining the viability of and for characterizing the germination of fungal spores in submerged culture. The method might be used in the laboratory for routine maintenance of inoculum quality of spore preparations. It also offers fungal physiologists a new tool for the investigation of spore germination and its subsequent influence on productivity. Effects of medium formulation on germination could be studied using this new method for development of better inoculation methods as part of the optimization of fungal fermentations. This work was supported by the Science and Engineering Research Council, United Kingdom. The Association of Commonwealth Universities, United Kingdom, is thanked for providing a scholarship to G. C. Paul.

References 1. Allen, P. J. 1957. Properties of a volatile fraction from underspores of Puccinia graminis var. Tritici affecting their germination and development. I. Biological activity. Plant Physiol. 32: 385-389. 2. Anderson, J.G., Smith, J.E. 1972. The effects of elevated temperatures on spore swelling and germination in Aspergillus niger. Can. J. Microbiol. 18: 289-297. 3. Cox, P. W., Thomas, C. R. 1992. Classification and measurement of fungal pellets by automated image analysis. Biotechnol. Bioeng. 39: 945 -952. 4. Deo, Y. M., Gaucher, G. M. 1984. Semicontinuous and continuous production of penicillin-G by Penicillium chrysogenum cells immobilized in K-carrageenan beads. Biotechnol. Bioeng. 26: 285 -295. 5. Gottleib, D. 1978. The germination of fungal sprees. Meadowfield Press, Shildon, England. 6. Mou, D-G., Cooney, C.L. 1983. Growth monitoring and control through computer-aided on-line mass balancing in a fed-batch penicillin fermentation. Biotechnol. Bioeng. 25: 225-255. 7. Packer, H.L., Thomas, C. R. 1990. Morphological measurements on filamentous microorganisms by fully automatic image analysis. Biotechnol. Bioeng. 35: 870-881. 8. Pirt, S.J., Callow, D. S. 1959. Continuous flow culture of the filamentous mould Penicillium chrysogenum and the control of its morphology. Nature 184: 307-310. 9. Reichl, U., Buschulte, T.K., Gilles, E.D. 1990. Study of early growth and branching of Sfrepfomyces fendae by means of an image processing system. J. Microsc. 138: 55-62. 10. Russ, J. C. 1990. Computer assisted microscopy. The measurement and analysis. Plenum Press, New York. 11. Smith, G.M.,Calam, C.T. 1980. Variations in inocula and their influence on the productivity of antibiotic fermentations. Biotechnol. Lett. 2: 261-266. 12. Trinci, A. P. J. 1971. Exponential growth of the germ tubes of fungal spores. J. Gen. Microbiol. 67: 345-348. 13. Tucker, K.G.,Kelly, T., Delgrazia, P., Thomas, C. R. 1992. Fully automatic measurement of mycelial morphology by image analysis. Biotechnol. Prog. 8: 353-359. 14. Ynagita, T. 1957. Biochemical aspects of the germination of conidiospores of Aspergillus niger. Arch. Mikrobiol. 26: 329-344.


23