The use of collagen or fibrin gels for the assay of human neutrophil chemotaxis

Journal oflmmunologiealMethods, 85 (1985) 137-151 Elsevier

137

JIM03712

The Use of Collagen or Fibrin Gels for the Assay of Human Neutrophil Chemotaxis L a i l a N . I s l a m , I.C. M c K a y a n d P.C. W i l k i n s o n Glasgow Universi(v, Department of Bacteriolog), and lmrnunolog~v, Western lnfirmarv, Glasgow GI I 6NT, U.K. (Received 24 July 1985, accepted 19 August 1985)

Neutrophil leucocytes are known to migrate actively into 3-dimensional gels of collagen or fibrin. In this paper, we have used such gels to study chemotaxis of h u m a n blood neutrophils towards gradient sources of formyl-methionyl-leucyl-phenylalanine (FMLP) using 2 assay systems. The first resembled the micropore filter assay in that neutrophils on the upper surface of collagen gels were allowed to invade in the presence of either an isotropic concentration or a gradient of FMLP. Neutrophils invaded the gel vigorously in both cases. The effect of the gradient was assessed by determining the population distribution at different levels in the gel. Cells moving randomly should be distributed normally, and directional locomotion should cause deviation from normal distribution. Such a deviation was seen, but was of marginal significance. A more direct demonstration of chemotaxis was achieved by the second assay in which an agarose slab containing FMLP was incorporated into a gel, and the paths of nearby neutrophils were filmed. These cells showed an unequivocal directional response to the F M L P gradient. Protein gels can thus be used in the same way as both the presently used filter assays and visual assays using plane substrata~ but with the advantage of providing a more physiological environment for the study of chemotaxis than either. Key words: chemotaxis - human neutrophils - protein gels

Introduction

There are 2 different sorts of assay in use for the study of leucocyte chemotaxis. The most widely used assays sample the distribution of cell populations following exposure to a gradient of attractant. Examples are the micropore filter and agarose assays. Less frequently used are assays in which the behaviour of individual cells is studied. For example, using time-lapse filming the paths of cells migrating towards a gradient source can be tracked. Both types of assay have generally been carried out using non-biological substrata such as flat glass or plastic surfaces and filters made of cellulose esters or other unphysiological materials. Three-dimensional matrices made from physiological materials are now becoming more popular. In some of the earliest studies of leucocyte locomotion, the migration of cells in serum (fibrin) clots was observed (Lewis, 1931). More recently, 3-dimensional collagen gels have been 0022-1759/85/$03.30 © 1985 Elsevier Science Publishers B.V. (Biomedical Division)

138 used in studies of a variety of cell types including fibroblasts (Bard and Hay, 1975), lymphocytes (Haston et al., 1982; Schor et al., 1983), neutrophils (Brown, 1982) and monocytes (Brown, 1984). Such gels may be prepared freshly from type I collagen from rat tail tendon (Elsdale and Bard, 1972; Schor, 1980). The collagen gel forms a fine fibrous 3-D meshwork through which cells can readily penetrate provided the collagen concentration is not too high. The gels are transparent and can be used for both types of assay mentioned above since they allow both direct observation of moving cells and the measurement of cell distribution after fixation. They provide a better approximation than most of the materials listed above to the environments through which cells have to move in vivo. Most of the studies published up till now using collagen gels have been studies of non-chemotactic locomotion. However, there is no reason why a collagen gel cannot be employed in the same way as the filter assay is used at present, namely to place an attractant source underneath the gel, to place cells on the upper surface of the gel and then to observe migration and the distribution of the cell population in response to the gradient. This in essence requires a measure of the vertical displacement of cells. A collagen gel can also be used in the same way that planar substrata are used at present, by placing a gradient source to one side of a microscopic field and then following the locomotion of cells in the horizontal plane through the gel towards the source. We have already used such techniques to study contact guidance of neutrophils in aligned gels (Wilkinson et al., 1982; Wilkinson and Lackie, 1983) and the chemotactic responses of lymphocytes (Wilkinson, 1985). Though we have used fibrin gels to study the effect of contact guidance on chemotaxis of human neutrophils (Wilkinson and Lackie, 1983), detailed descriptions of the use of collagen or fibrin gels for assaying neutrophil chemotaxis have not been published. Here we report that such gels can readily be used both for cell-distribution assays (in the same way as micropore filters) and for time-lapse filming of chemotactic responses. For the first type of assay, neutrophils were allowed to migrate into gels from their upper surfaces in response to gradients or uniform concentrations of formyl-Met-Leu-Phe. For the second, time-lapse cinematography was used to study chemotaxis of neutrophils across a microscopic field towards a source of the same peptide.

Materials and Methods Materials

Hanks' balanced salt solution (HBSS), lymphocyte separation medium and tissue culture multiwell plates with 24 flat-bottomed wells (1.7 × 1.6 cm) were obtained from Flow Labs., Irvine, Scotland. Dimethyl sulphoxide (DMSO, Grade I), 25% glutaraldehyde (Grade II), N-formyl-L-methionyl-L-leucyl-L-phenylalanine(FMLP), 3-(N-morpholino)propanesulphonic acid (MOPS), fibrinogen (human, fraction I, type I) and thrombin (human lyophilized) were obtained from Sigma, Poole, Dorset. Type I collagen was extracted from rat tail tendons with 3% acetic acid by slight modification of the methods of Elsdale and Bard (1972) and Schor (1980) using 2

139 centrifugation steps to remove any undissolved material before dialysis against distilled water. After dialysis, the resulting aqueous solution of collagen was adjusted to p H 4.0 and stored at - 2 0 ° C . F M L P was prepared as a stock solution (10 2 M in DMSO) and stored at - 2 0 ° C .

Cells Heparinized human blood was obtained from normal d o n o r s b y venepuncture. Neutrophils were purified by the standard procedure of dextran sedimentation followed by centrifugation through lymphocyte separation medium. Red cells were removed by hypotonic lysis. The neutrophils were resuspended in HBSS-MOPS (10 mM) at 10 6 cells per ml. For chemokinesis experiments, the cells were resuspended in a medium containing a chosen dilution of F M L P between 10 9 and 10 -6 M in HBSS-MOPS. The viability of cells in all experiments described here, judged by trypan blue exclusion exceeded 97%.

Preparation of collagen gels and assay of neutrophil locomotion Three-dimensional collagen matrices were prepared by adjustment of the molarity and the p H of the stock collagen solution to physiological levels by addition of appropriate volumes of HBSS, MOPS and distilled water as described by Shields et al. (1984). These were rapidly mixed to give a 10 ml solution of collagen (collagen A) at 1.5 m g / m l with F M L P added at any desired concentration. Another solution of collagen (collagen B) was prepared at 1 m g / m l without FMLP. Once these solutions were prepared the collagen formed a gel within a few minutes. Gels for chemokinesis assays were prepared by rapidly pouring a 0.5 ml portion of collagen A (containing F M L P at a uniform concentration of between 10 9 and 10 - 6 M ) into the wells of a multiwell tissue culture dish. Provided the collagen was used within 2 - 3 months of preparation, it gelled within 5 10 min at room temperature. The gels were allowed to set for 1 h before use. They were then overlaid with neutrophils in suspension in the same concentration of F M L P as was present in the gel, so that the F M L P concentration was constant throughout (Fig. la). For control experiments, F M L P was absent from both collagen and the cell suspension. Gels for chemotaxis assays were prepared as follows: a 0.4 ml portion of collagen A containing F M L P at a chosen uniform concentration of between 10 10 and 10 5 M was poured as above and allowed to set. This was to serve as a gradient source. One hour later a 0.3 ml portion of collagen B (without FMLP) was quickly and carefully layered over the first gel and allowed to set. This was to form a layer through which the F M L P could diffuse to form a gradient (Fig. lb). A suspension of neutrophils was added to the top of the collagen B layer 20-60 min later. In control experiments, the same procedure was followed except that collagen A lacked FMLP. The height of the combined layers A and B was approximately 3.5 m m and the gels were optically clear enough to allow good visualization of cells deep within the gel using the fine focusing adjustment of an inverted phase contrast microscope. The tissue culture dishes were immediately incubated at 37°C for cells to invade the gel. Neutrophils were allowed to invade the gels for various times. The gels were then fixed by addition of 2.5% glutaraldehyde in HBSS-MOPS at p H 7.4 for 30 min.

140

A

iiiiiiiiiiiiii Iii,

CelIs+FMLP

iiiii

i~

_ ~_:.~-i:;-:--:

~

-

~Collagen B NO FMLP

Collagen A+FMLP

~

Ct

~

j

\ /

P"

/

\

, \ Collagen or

Coverslip Fibrin Gel

l

Collagen A+FMLP

[ \

Agarose Metal FilmingChamber

Fig. 1. Schematic diagrams of the assays used. Above: population distribution assays. Below: visual assay. a: the assay for neutrophil invasion of collagen in an isolropic concentration of FMLP. b: the assay for neutrophil invasion of collagen in a gradient of FMLP. c: the visual assay using a metal filming chamber. The FMLP source is in a block of agarose which is overlaid with collagen or fibrin in which neutrophils were suspended before gelling. Hatched areas indicate solid media containing FMLP. The dotted areas indicate fluid media (Hanks') containing cells with or without FMLP. These cells settle on the surface of the collagen which they then invade. The area with broken hatching indicates collagen without FMEP.

Then the upper surface of the gel was washed thrice with HBSS-MOPS to remove non-invaded cells. The following measures of cell invasion were used. (i) A leading front measure, i.e., the distance that the leading 2 cells in the same focal plane in a microscope field ( x 200) had migrated (Zigmond and Hirsch, 1973). Four randomly chosen fields in each of 6 identical wells were counted. (ii) The vertical distribution of the neutrophil population through the gel was determined by counting the number of cells in each of a series of planes starting 40 ffm below the gel surface and moving down at 40/~m intervals until there were no more cells to be counted. The number of cells in 100 squares of a 10 x 10 graticulated eyepiece was measured in each plane at x 200 magnification. Ten to 16 readings were taken from 4 identical wells for the average cell number ( N ) at a particular distance (d, ffm) from the upper gel surface.

Preparation of fibrin gels Fibrin gels were prepared by addition of 1 part of thrombin at 1 N I H unit per ml to 9 parts of fibrinogen, final concentration, 1 mg per ml in HBSS-MOPS + FMLP. The mixture (0.4 ml) was rapidly added to dishes as described above. The fibrin gelled within 30 s at room temperature. It was then overlaid with HBSS-MOPS and left for 1 h. Neutrophils were then added and the gels incubated as described above.

141

Mathematical analysis of final cell distributions If cells start from 1 plane surface of a gel and move into it partly by random migration and partly by a superimposed directed movement of uniform velocity v normal to the surface, then their distribution at the end of the experiment should be given approximately by the equation: N = a(exp(-(d-

vt)2/2s2) + e x p ( - (d+ vt)2/2s2))

S

where N is the number of cells in focus at a depth d in the gel, a is a constant proportional to the total number of cells in the gel, s is the root-mean-square displacement attributable to random migration alone, t is the incubation time and v is the velocity of directional migration. This equation was made to fit the observed cell distributions for each microscopic field by adjusting the values of a, s and v until the value of chi-squared, which measures the discrepancy between observed and theoretical cell numbers, was as small as possible. The optimal values of s and v, as judged by this criterion, were taken as measures of random migration and directional migration respectively.

Visual assay of neutrophil chemotaxis in gels The method was adapted from that previously described for lymphocyte chemotaxis (Wilkinson, 1985). The assays were done in metal filming chambers which have a central circular hole to the under-surface of which a coverslip was attached (see Allan and Wilkinson (1978) for details of assays with these chambers). F M L P at 10 ~ M was incorporated into agarose (1%) containing HBSS-MOPS by melting the agarose, adding the F M L P and pouring the agarose onto a microscope slide to form a gel about 1 mm deep. A small slab of this agarose (1 × 2 ram) was cut out with a scalpel and transferred to the coverslip on the filming chamber. A solution of fibrinogen was prepared at 2 m g / m l and an equal volume of a neutrophil suspension at 2 × 10 6 c e l l s / m l was added and mixed. Thrombin was then adde& the mixture (0.4 ml) was added to the filming chamber and the gel allowed to set over the agarose slab (Fig. lc). Thus neutrophils were incorporated into the gel, not added to its surface as above. The chamber was then filled with HBSS-MOPS, sealed with an upper coverslip, and placed on the warmed stage (37°C) of an inverted microscope, and a field about 1 mm away from the edge of the agarose slab was chosen. The microscope was focused so that the behaviour of neutrophils within the fibrin gel could be observed using phase contrast optics. Locomotion of these neutrophils was then filmed by time-lapse cinematography using a lapse interval of 1 frame per 4 s. Once the films were developed, analysis was carried out by projecting the film onto paper and tracing out the cell path by dotting the centre of each cell at 40 s intervals. The dots were joined with straight lines. The response to the gradient was quantified by measuring the velocity [vector] of each moving cell averaged over the whole length of its track and the velocities were plotted on a vector scatter diagram. The y-axis was taken as the direction parallel to the gradient, migration towards the source being positive displacement, migration away from it being

142

negative displacement. The mean_+ SEM velocity components in the x and y directions were calculated for each cell population and Student's single-sample t-test was used to test whether these means differed significantly from zero and thus to provide evidence about chemotaxis towards the source. Though fibrin gels are 3-dimensional, the essential movement under consideration is in the horizontal plane towards the source and the assay is capable of measuring this movement.

Results

Time course of inuasion of gels by neutrophils In Fig. 2 is shown the time course for invasion of gels of collagen and fibrin, at 1 m g / m l , by neutrophils in the presence of uniform 10 s M FMLP. The course for both types of gel was similar. The leading front measure increased time but the relationship was not linear. For further experiments, an incubation

both time with time

LEADING FRONT

~um)

a)

400 ]

200

b)

-

100 -

0

0

I

I

i

1

2

3

0

i

i

i

l

2

3

HOURS

Fig. 2. Invasion by neutrophils of collagen (a) or fibrin (b) gels, both at 1 m g / m l in presence of FMLP (10- ~ M throughout). Time course for leading front assay for 2 separate experiments ( × ) and (O) in each gel _+SD.

143

of 2 h was chosen. If longer times were chosen, it was found that the leading front values for different gels began to diverge, possibly because of physical changes in the gel, such as contraction, that were difficult to control. If shorter times were chosen, cell penetration of the gel was insufficient to allow accurate plotting of cell numbers at different levels.

Chemokinesis of neutrophils in collagen gels: leading front value The dose-response curve for the leading front measure for neutrophils invading collagen gels in the presence of different uniform concentrations of FMLP is shown in Fig. 3. Similar dose-response curves were obtained in 4 similar experiments. The optimum concentration of FMLP in this assay was 10 ~ M. In the absence of FMLP, the leading front value was low.

Cell distribution within collagen gels The distribution of cells through collagen gels was plotted as mean cell number against distance from the top of the gel in the presence of a uniform concentration of 10 s M FMLP and is shown in Fig. 4. The distribution of cells moving randomly

LEADING FRONT

~um) 4 0 0 -i

300-

200 -

~oo ~

0 I

"St

1

I

I

I

-9

-8

-7

-6

Log [FMLP] (M)

Fig. 3. Dose-response curve. Neutrophils migrating into collagen gels (1.5 mg/ml) in different isotropic concentrations of FMLP. Leading front value at 2 h_+ SD.

144 into a gel would be identical to that expected for a simple diffusion ( Z i g m o n d and Hirsch, 1973); thus, if the logarithm of the cell n u m b e r were p l o t t e d against the square of the distance migrated, a linear plot would be expected. The d a t a of Fig. 4 were therefore t r a n s f o r m e d in this way and are shown in Fig. 5. This shows linear plots both for cells entering a gel in the absence of an a t t r a c t a n t a n d for cells r e s p o n d i n g to a uniform c o n c e n t r a t i o n of F M L P (it also shows a plot for a chemotaxis assay, see below). These plots are suggestive of r a n d o m l o c o m o t i o n in the absence of an a t t r a c t a n t a n d of chemokinesis (increased r a n d o m l o c o m o t i o n either due to an increase in cell speed or to recruitment of more cells or to both) in the presence of F M L P . In fact, the n u m b e r of cells entering the gel was greater in the presence of F M L P than in its absence (Fig. 4). The assay does not allow a direct measure of cell speed, though F M L P has been shown in other studies to increase the speed of neutrophils as well as recruiting extra l o c o m o t o r y cells (Shields and Haston, 1985).

CELL No. (N) 320 -

240 -

160 -

80-

\ ~x

o o

I

I

I

1

I

I

8O

160

240

320

400

480

DISTANCE (d, jJm)

Fig. 4. Distribution of neutrophils through collagen gels (1.5 mg/ml) after 2 h incubation in the presence of an isotropic concentration (10 s M) of FMLP (e) or in the absence of FMLP ( x ) after 2 h.

145

"-I

Log

10N

2.0

x

0

]

O

I

4

x

I

I

I

8

I

12

i

l

t6

]

20

d 2 x lO-4um

Fig. 5. Transformed plot of data from Fig. 4 and Fig. 7. Logarithm of cell numbers (logloN) against d 2. No FMLP (11); isotropic FMLP ( × ) from Fig. 4; FMLP gradient (O) from Fig. 7.

Migration of neutrophils into collagen gels in response to gradients of FMLP: leading front oalue Fig. 6 shows the dose-response curve for the leading front measure for cells migrating in gradients originating from sources of FMLP at different concentrations and shows that 10-8 M FMLP was the optimal concentration in this assay.

Cell distribution in gels If neutrophils exposed to gradients of FMLP respond by directional locomotion towards the peptide, it will be expected that the distribution of cells through the gel will no longer be that expected for a simple diffusion, but that there will be a flux of cells into the gel. If this is the case, the plot of log number of cells against the square of the distance migrated will not be linear. The effectiveness of the chemotactic response will depend on the amplitude of the gradient and, under the assay

146

LEADING FRONT (Urn) 500 -

400 -

300-

200

-

100-

..p,t [ -10

I

I

[

-9

-8

-7

[ -6

I -5

Log [FMLP](M) Fig. 6. Dose-response curve. Neutrophils migrating through collagen gels (l m g / m l ) towards gradient sources of F M L P at different concentrations leading front values at 2 h _+SD.

conditions described, this will depend on how long the FMLP, present at uniform concentration in collagen A, is allowed to diffuse through collagen B before cells are placed on the upper surface of the latter. Therefore, preliminary experiments were performed to determine the optimal time for gradient diffusion that was consistent with formation of a firm gel, and at the same time allowed good detection of the gradient by cells. A good chemotactic response was judged in this assay on the basis of the non-linearity of the plot of log cell no. against d=. It was found that a 20 min lapse for gradient diffusion before layering cells onto the upper gel was the minimum consistent with the practical limitation of having a firm gel to place the

147 CELL No.(N) 160-

120-

80-

40-

0 0

I

I

1

f

[

I

80

160

240

320

400

480

DISTANCE (djJm)

Fig. 7, Distribution of neutrophils through a collagen gel after 2 h incubation in the presence of a gradient of FMLP (10 -8 M at source) set up 30 rain before adding the cells.

cells on. In practice, 30 min was adopted as the optimum time for F M L P diffusion through the upper gel and the cells were allowed to migrate into the gel for 120 min. The distribution of cells through the gel in response to a gradient of F M L P (10 s M at source) is shown in Fig. 7 (cf Fig. 4). This is a non-linear transformed plot and the same data transformed to show log cell no. against d 2 is shown in Fig. 5. This plot is clearly non-linear. Analysis of these distributions by curve fitting as described in the Methods gave estimates of cell numbers entering the gel, displacement due to random migration and velocity of directional migration for each microscopic field studied. These estimates are summarized in Table I. The displacements attributable to random migration in Table I show no evidence of being influenced by the concentration gradient of FMLP, nor do the numbers of cells entering the gels. The second last column of the table is of interest as an indicator of whether the last 4 experiments (with uniform F M L P concentration) showed less evidence of directional drift than the first 7 experiments (with an F M L P concentration gradient). In the presence of a gradient the mean estimate of direc-

148 TABLE 1 S U M M A R Y OF M I G R A T I O N PARAMETERS ESTIMATED FROM F IN A L DISTRIBUTIONS OF CELLS THAT STARTED ON THE TOP S U R F A C E OF GELS A N D M I G R A T E D EITH ER IN A VERTICAL C O N C E N T R A T I O N G R A D I E N T OF FMLP OR IN A U N I F O R M FMLP CONCENTRATION

Experiment no.

no. of fields observed

Mean of all fields_+ SEM a

s/l~m

~,/l~m h

(proportional to number of cells entering the gel)

(root-mean-square displacement attributable to random migration)

(estimated

I

134 ± 5 118_+4 125-+5 122-+2 120 _+4 100 -+ 5 112_+3

59.4 _+3.9 52.4+5.6 36.1 -+6.5 65.9_+0.6 22.6 _+6.4 54.7 + 2.2 18.3_+6.4

102_+4 126±4 96-+2 109 ± 3

15.8_+6.5 12.9__+6.2 2.8±2.4 45.9 ± 2.1

velocity of directional migration)

With an FMLP concentration gradient

(0-10

s M)

l 2 3 4 5 6 7

10 10 10 10 10 16 12

7 498 _+269 9418_+897 17123_+410 16595_+643 13 223 ± 464 9 733 _+329 10143_+286

With a uniform concentration of FMLP (10 s M) a b c d

10 10 10 10

16440_+578 19668-+966 10361 _+396 19683 + 777

tional migration was 44.2_+ 7.0 ~ m / h , whereas in the absence of a gradient the mean was 19.49 ± 9.3 t~m/h. Comparison of the 2 sets of numbers in the velocity column gives P = 0.06 by Student's 2-tailed t-test or P = 0.04 by Wilcoxon's rank sum test. Arguably a 1-tailed test might be considered applicable on the grounds that negative chemotaxis is not observed in neutrophils, and if so then the P values would be half of those quoted. Even so, the statistical significance is rather marginal, possibly on account of the anomalous result of experiment d.

Visual assay of neutrophil chemotaxis through gels towards an F M L P source The evidence of chemotaxis from plots of the distribution of cell populations in gels as described above is marginal. A visual demonstration of directional locomotion of neutrophils through a gel towards a source of FMLP would provide more direct evidence. Neutrophils were incorporated into a fibrin gel which was cast over a slab of agarose containing FMLP (10 -~ M) (Fig. lc) and cells in a field at 1 mm distance from the agarose were filmed. The paths taken by the cells were tracked and the displacements determined and plotted on a vector scatter diagram as shown in Fig. 8. For each cell the starting point (at the beginning of the film sequence) was placed at the origin and the time-averaged velocity (displacement per rain) from that point is represented by a cross. The axis of the gradient is the y-axis and the source was at the north. It is evident from Fig. 8 that most of the cells have displaced

149 -20

X -15

X

x

X X X

X

-10

x&

X X

X

-5 X X

I

-10

I

1

I

!

I

-5

5

Io

15

20

--5

-10 Fig. 8. Vector scatter diagram showing time-averaged velocities of neutrophils in the visual assay in relation to the direction of an FMLP concentration gradient. The velocity (~m/min) of each cell is represented by a cross. The FMLP source was at the top (y +). The large cross represents the mean velocity + SEM in each of the x and y directions. Displacement in the y direction was highly significant (P < 0.001) but not in the x direction. This indicates directional locomotion of the cells towards the FMLP.

towards the source ( m e a n velocity in y-direction, + 8.7 _+ 1.1 ~ m / m i n ) . Th er e was no significant d i s p l a c e m e n t in the x-direction ( m e an velocity - 1.75 _+ 1.2 ~ t m / m i n ) . Similar e x p e r i m e n t s were carried out in collagen gels with similar results.

Discussion T h r e e - d i m e n s i o n a l fibrous matrices such as collagen and fibrin p r o v i d e a milieu for studying cell l o c o m o t i o n which is closer in character to those which cells e n c o u n t e r in vivo than the artificial substrata in m o r e c o m m o n use. Assay of c h e m o t a x i s of leucocytes in such gels has the a d v a n t a g e that this, unlike other assays, can be used b o t h to study the distribution of a cell p o p u l a t i o n following exposure to a stimulus and as a visual assay in which the detailed b e h a v i o u r of individual cells d u r i n g the response can be followed. P o p u l a t i o n assays have the

150

advantage that a very large sample is taken. However, due to the dimensions of the assay and probably to chance factors that affect both diffusion of the chemotactic factor and penetration of the gel by the cells, the distinction between population distribution in a gradient, in which a flux of cells responding chemotactically should be seen, and population distribution in isotropic attractant concentrations, which should be normal, is in practice a fine one, and is clearer in some experiments than in others (Table I). This is equally true of the filter assay and the agarose assay but the collagen gel assay has advantages over these assays in certain respects. It has been shown for example that leucocytes move through 3-dimensional matrices of collagen or fibrin by a mechanism which is largely independent of adhesive interactions with the substratum (Haston et al., 1982) and therefore, assays for chemotaxis using collagen gels are not confused by the cell-substratum interactions which are unavoidable using the glass or cellulose nitrate substrata of other assay systems. An unequivocal demonstration of chemotaxis through gels may be achieved by the visual assay described here in which cells were seen on time-lapse films to move directly towards the gradient source (an FMLP-containing agarose slab). In the visual assay reported in this paper, we used F M L P as a gradient source, but we have also used purified C5a, and sources of either immune complexes or LPS, both of the latter in the presence of fresh human serum. G o o d chemotaxis of human neutrophils to all these sources could be demonstrated. We have used 2 types of gel here, collagen and fibrin. Neutrophils migrated into both with a very similar time course when tested up to 2 h at least. We have not been able to demonstrate locomotion of monocytes into either gel, though Brown (1984) showed that the monocyte population which becomes detached from protein-coated plane substrata on overnight culture would also move into collagen gels. Lymphocytes migrate well into collagen gels (Haston et al., 1980; Schor et al., 1983; Shields et al., 1984), but have not been tested in fibrin. There are certain technical problems with gel assays. For example, protein gels are fragile, cannot be manipulated readily, and collapse if cut. They may become detached from the sides of tissue culture wells, and we scratched the sides of the wells to improve gel attachment. Counting cells in gels is not easy when there is any vibration, since they are prone to wobble, and we found that counting was best done when the laboratory was quiet. Using the metal filming chamber which is sealed and air-free, wobble is less of a problem. Another difficulty that may be encountered after a gel has been poured and has set is that cells placed on the surface of the gel sometimes do not move in, even though they are in locomotor morphology. This is due to some undefined change at the gel surface, possibly contraction or horizontal alignment of the gel, and we have found that covering the gel with fluid as soon as possible after it has set, rather than leaving it exposed to air reduces the incidence of this difficulty. The difference between the ease with which chemotaxis towards F M L P was demonstrated by filming the paths of neutrophils and the difficulty with which it was demonstrated in the population distribution assay highlights the problems of showing chemotaxis convincingly with the latter type of assay. The micropore filter assay is the classical population distribution assay and has been used in the great

151

majority of studies of neutrophil chemotaxis. It is a good assay for demonstrating that locomotion has been stimulated by a given factor, but is not an ideal assay for showing that the stimulated locomotion is chemotactic. The status of many of the putative chemotactic factors reported in the literature needs further evaluation usin 8 assays which give more direct evidence of chemotaxis.

Acknowledgements L.N.I. was supported by the Commonwealth Scholarship and Fellowship Plan (1983). PCW was supported by an MRC project grant. The authors are grateful to Dr. Wendy S. Haston and to Mr. James M. Shields for their suggestions and valuable discussions.

References Allan, R.B. and P.C. Wilkinson, 1978, Exp. Cell Res. 111, 191. Bard, J.B.L. and E.D. Hay, 1975, J. Cell Biol. 67, 400. Brown, A.F., 1982, J. Cel. Sci. 58, 455. Brown, A.F., 1984, Scanning Electron Microsc, 11,747. Elsdale, T. and J. Bard, 1972, J. Cell Biol. 54, 626. Haston, W.S., J.M. Shields and P.C. Wilkinson, 1982, J. Cell Biol. 92, 747. Lewis, W.H., 1931, Bull. Johns Hopkins Hosp. 49, 29. Schor, S.L., 1980, J. Cell Sci., 41, 159. Schor, S.L., T.D. Allen and B. Winn, 1983, J. Cell Biol. 96, 1089. Shields, J.M. and W.S. Haston, 1985, J. Cell Sci. 74, 75. Shields, J.M., W.S. Haston and P.C. Wilkinson, 1984, Immunology 51,259. Wilkinson, P.C., 1985, J. Immunol. Methods 76, 105. Wilkinson, P.C. and J.M. Lackie, 1983, Exp. Cell Res. 145,255. Wilkinson, P.C., J.M. Shields and W.S. Haston, 1982, Exp. Cell Res. 140, 55. Zigmond, S.H. and J.G. Hirsch, 1973, J. Exp. Med. 137, 387.