Macrophage dependence of peripheral sensory nerve regeneration: Possible involvement of nerve growth factor

\leuron, Vol. 6, 359-370, March, 1991, Copyright 0 1991 by Cell Press

Macrophage Dependence of Peripheral Sensory Nerve Regeneration: Possible Involvement of Nerve Growth Factor Michael C. Brown,* V. Hugh Perry,+ E. Ruth Lunn,* “iiamon Gordon,* and Rolf Heumanns $University Laboratory of Physiology ‘arks Road 53xford, OX1 3PT -Department of Experimental Psychology 5. Parks Road 3xford, OX1 3UD 5ir William Dunn School of Pathology i. Parks Road 3xford, OX1 3RE !ngland IAbteilung Neurochemie Max-Planck-lnstitut ftir Psychiatric \m Klopferspitz 18a -18033 Planegg-Martinsried Jermany

Summary The levels of NGF and NCF receptor mRNA, the degree of macrophage recruitment, and the ability of sensory and motor axons to regenerate were measured in C57BLI Ola mice, in which Wallerian degeneration following a nerve lesion is very slow. Results were compared with those from C57BU6) and BALB/c mice, in which degeneration is normal. We found that in C57BUOla mice, apart from the actual lesion site, recruitment of macrophages was much lower, levels of mRNA for both NGF and its receptor were raised only slightly above normal, and sensory axon regeneration was much impaired. Mokor axons regenerated quite well. These results provide in vivo evidence that macrophage recruitment is an important component of NGF synthesis and of sensory (but riot motor) axon maintenance and regrowth.

introduction The degenerated distal stump of sectioned peripheral lerves provides an excellent environment for regengerating axons (Ram6n y Cajal, 1928). It possesses a scaffold matrix, the Schwann cell basal lamina, which contains laminin (Cornbrooks et al., 1983), a well tnown substratum for growth cones. It also contains 1 population of cells, of which the Schwann cells are arobably the most important, that secrete a variety +f neuroactive agents which attract and encourage growth (Politis et al., 1982; Lundborg et al., 1982) even iIf axons of central neurons (Benfey and Aguayo, 1982). The best known of these trophic factors is nerve );rowth factor (NGF; Heumann et al., 1987a). Macro,3hages recruited into peripheral nerves after axotomy ‘Perry et al., 1987) are probably important for initiating jchwann cell mitosis (Beuche and Friede, 1984) and nducing them to manufacture NGF (Heumann et al.,

1987b) by secreting interleukin-I (Lindholm et al., 1987). The importance of actively dividing Schwann cells for axonal regrowth is shown by the poor growth through frozen segments of peripheral nerve (Hall, 1986; but see Sketelj et al., 1989) or when mitosis is inhibited (Hall and Cregson, 1977). We recently described a strain of mouse in which nerve degeneration following axotomy is extremely slow (Lunn et al., 1989). These mice are the C57BUOla strain (referred to in a previous publication [Lunn et al., 19891 as C57BL/6/0la, the name given by their supplier). In these animals the majority of axons in the isolated distal nerve segment continue to conduct action potentials for at least a week, ultrastructural morphology is maintained, and there is an almost complete absence of recruitment of myelomonocytic cells and of Schwann cell mitosis (except in the first few millimeters surrounding the lesion). Knowing the experiments of Langley and Anderson (1904), who were unable to detect any growth of axons from one proximal nerve stump into another intact proximal nerve stump,weassumed that intact peripheral nerveswere impenetrable to axons and that, therefore, regrowth of crushed axons down distal nerve stumps would be extremely poor in C57BUOla mice. To our surprise, motor axons succeeded in returning to the soleus muscle at nearly the same speed and in approximately the same numbers as in other mouse strains following sciatic nerve crush in the mid-thigh (Lunn et al., 1989). They reached their targets while most of the original axons in the sciatic nerve were still intact by growing along intact unmyelinating Schwann cells (Remak cells) that were still associated with undegenerated unmyelinated axons (Brown et al., 1989, j. Physiol., abstract). The aim of the present experiments was to determine, whether, as might be predicted in the absence of macrophage invasion and Schwann cell mitosis, NGF production was reduced in the distal nerve stump in C57BLIOla mice. In addition, we hoped to determine whether sensory axons, some of which are known to be NGF dependent, would regrow as well as motor axons in C57BUOla mice. It has been reported that antibodies to NGF do not interfere with regeneration of sensory fibers (Rich et al., 1984; Diamond et al., 1987).

Results Slow Degeneration and Failure of Macrophage Recruitment in C57BUOla Nerves Electron phenous (Lunn et in nerves axotomy. to exclude

microscopy of the distal segment of cut sanerves shows that, as in the sciatic nerve al., 1989), there is very little degeneration taken from C57BUOla mice a week after Seven days after a nerve section designed reinnervation, >70% of myelinated and un-

hlacrophage 3 il

Dependence

of Sensory Axon Regrowth

Table 1. Axonal Counts from the Saphenous Nerve Distal to t?e Injury, Expressed As a Percentage of Control Values (’ Days After Injury) -__ Myelinated Unmyelinated

k57BUOla _~ Near lesion “0 m m from hALB/c _Idear lesion “0 m m from

cut

Crush

cut

Crush

lesion

78.5 71.5

90.4 76.7

76.4 77.5

306.0 61.6

lesion

1.3 0.0

12.5 0.0

2.7 0.3

200.0 63.0

Data are based on counts on two normal nerves, two cut nerves, and one crushed nerve of each strain. (For typical values of absolute numbers of axons see Figure 5.) Note that, as would be expected, nearly all the original axons in the distal nerve stump of BAlB/c mice have degenerated by 7 days (Cut data), but that .kter crush, allowing rapid regeneration, many unmyelinated axems are present both near and 10 m m from the lesion. In the case of C57BUOla mice, over 70% of myelinated and unmyelinated axons are still intact in the distal nerve stumps (Cut data). In .rddition, many regenerated unmyelinated axons are present near the crush site (306% compared with 76.4%), but these do not extend as far as IO m m from the lesion (61.6% compared with “6.7%).

myelinated fibers remain in nerves from the mutant C57BUOla mice, whereas virtually no myelinated or unmyelinated axons remain in the distal stump of BALBlc mice (Table 1). The lack of degeneration 7days ,dter axotomy is associated with a lack of the normal !.ecruitment of F4/80-positive macrophages into the distal nerve stump of both the sciatic and saphenous herves of the C57BUOla mice (Figure 1; Figure 2) at slistances greater than 3 m m from the site of nerve ,;ection. At the lesion site, substantial numbers do ,lccumulate and cause a swelling in the nerves, and : he nerve both proximal and distal to the crush site is ihiny as a result of the presence of intact myelin. In Z57BL16J and BALBlc mice, on the other hand, macro;ohages are recruited into the whole of the distal ,;tump, and there is a greater than 7-fold increase in -heir number over the first 5 days (Figure 1; Figure 2). in addition large numbers of macrophages are found n the connective tissues and epineurium of the lerves at 3 days. Comparison of NCF mRNA and NCF Receptor mRNA Levels in Segments of Sciatic Nerve A rapid, transient rise in NGF mRNAafter explantation and culture of innervated target tissues and sciatic nerve segments has been described (see Heumann et al., 1990). Such a macrophage-independent, rapid response after dissection and culture was also found

‘igure

1. Tranverse

Sections

of Saphenous

and Sciatic Nerves Stained

in C57BU6) mice and was virtually identical in C57BL1 Ola mice. The peak of NCF mRNA levels was at 6 hr; thereafter it decreased to constant but still elevated levels in both strains during the 72 hr of measurement (Figure 3A). We nexttested whether C57BUOla nerves were still responsive to macrophage-conditioned medium added at 3 days for a period of 3 hr (Heumann et al., 1987b). Again, the nerves of both strains responded well by reincreasing their levels of NGF mRNA. In peripheral nerves NGF receptor expression is restricted exclusively to Schwann cells (Bandtlow et al., 1987). The levels of mRNA for NGF receptor in cultured nerves rise, albeit slowly, in both strains from an average of 64 fg/mg wet weight to 505 fg/mg wet weight at 72 hr. Notably, the NGF receptor mRNA in nerves is not influenced appreciably by macrophageconditioned medium (Figure 3B), which is in agreement with results obtained previously in rats (Heumann et al., 1987b). We now challenged in vivo our previous suggestion that diffusion into the nerve of secretory products released from macrophages could be an important element in stimulating NGF production. Table 2 gives the levels of mRNA for NGF before and 2, 3 and 5 days after sciatic nerve section in vivo. The results are different for all three strains of mouse studied. The most striking finding is that at all times after section the level of mRNA is lowest for the C57BL/Ola mice. At 5 days after section in these whole distal nerve segments, the levels of NGF mRNA were highest in BALBlc followed by C57BU6J mice. As we have found large numbers of macrophages localized near the lesion site in C57BLlOla mice and as macrophages are involved in the up regulation of mRNA for NGF (Heumann et al., 1987b), it seemed possible that this region of the severed sciatic in these mice might be responsible for much of the small rise seen in assays of the whole distal segment at 5 days. Separate assays of the top 3 m m of the severed stumps and of the rest of the sciatic distal to this point were therefore made. The results confirm that at the lesion site, there are high levels of NGF mRNA in the nerve in both C57BL16) and C57BUOla mice (104 f 42 and 81 f 37 fglmg wet weight, respectively). However, in only C57BUOla mice, the portion of nerve we// away from the lesion site had negligible amounts of mRNA for NGF (22 + 5 fg/mg wet weight; Figure 3C). This can be compared with the value from the equivalent region in C57BU6J mice (92 f 13 fg/mg wet weight; Figure 3C), which is not much less than the value for the lesion site in this strain. Control (nonsectioned) levels were at 7.9 + 2 fglmg wet weight (Figure 3C). These values are very similar to those measured pre-

for Macrophages

with a Polyclonal

Antibody

F4/80

japhenous nerves are on the left; sciatic nerves are on the right. (A and D) Normal nerves from C57BU6J mice. (B and E) Nerves from Z57BU6J mice 6 days after section. (C and F) Nerves from C57BUOla mice also 6 days after section. Sections are all >3 m m distal to !he point of section. Bars, 25 urn (A, B, and C); 50 urn (D, E, and F).

Neuron 362

Sciatic

Saphenous 2015-

,I

T 1

Figure 2. Counts of F4/80-Positive Ceils in the Distal Stumps of the Sciatic and Saphenous Nerves of C57BL/Ola, C57BU6J, and BALBlc Mice at Various Times after Axotomy Open squares and continuous lines represent C57BUOla mice; crosses and dotted lines, C57BU6J mice; open diamonds and dashed lines, BALBk mice. Bars give standard deviations.

-0

12 3 4 5 Days post axotomy

o&--,-7 7 012345 Days post axotomy

6

viously in the Wistar rat, in which nonlesioned basal levels were at 3.9 f 1 fg/mg wet weight and 64 + 4 fg/mg wet weight in the distal segment 3 days after section (Heumann et al., 1987a). In spite of the lack of regulation of NGF receptor by macrophage-conditioned medium, the results in Figure 3C demonstrate that not only NGF, but also NGF receptor mRNA levels are much lower (4.9-fold) in the C57BUOla mice compared with the C57BU6J mice in the bottom portion of the sciatic nerve 5 days after section. The same is true for all the earlier time points in which the nerve was assayed as a whole (Table 2). Comparison of Retrograde Degeneration and Regeneration of Severed Cutaneous Sensory Axons in C57BLIOla and Other Mice Electrophysiological Results The distance reached at different times by fibers with a low stimulus threshold in the proximal stump is shown in Figure 4A. By 3 days, potentials were recordable between 2 and 6 mm distal to the crush point in both C57BU6J and C57BUOla mice. For nerves from C57BU6J mice, the distance reached by axons grew progressivelywith time, and by 7days, axons had usually reached the end of the portion of nerve dissected for study, approximately 16-18 mm. The growth rate (theslopeoftheregression lineforthedistanceversus time graph) was m2.9 mm/day. The low-threshold afferents in BALB/c mice regrew only slightly more slowly (-2.5 mm/day) than those in C57BL/6J mice. As will be seen later, the axons in BALB/c mice do better in terms of the number of fibers returning. In nerves from C57BUOla mice, by contrast, no obvious further gains in distance were made up to 10 days after sec-

Table 2. Levels of mRNAs for NGF and NCF Receptor

in Whole

2 Days

3 Days

C57BUOla C57B U6J BALBlc

47 f 3 100 262 f 19 166 60 f 25 105 Average unsectioned

Distal Sciatic Nerve Segments NGF Receptor

NCF mRNA (fg/mg wet weight) Strain

tion, although studies at longer times after section suggest a continued, slow elongation by some axons (see below). Thus in this strain, cutaneous myelinated axons appear to start growing well, but then slow down to a considerable extent. The same is true for fibers that could be activated by stimulating the proximal stump with longer pulses (1 ms) and greater voltage. The comparative regrowth of these fibers is shown in Figure 48. The difference is not as marked as that for low-threshold afferents because regrowth in the high-threshold fibers (probably C fibers) is slower in the normal C57BL/6) mice (~2 mm per day) and progresses further in the C57BL/Ola mice (some 7 mm) before tailing off. The progress of regeneration of myelinated afferent fibers in C57BUOla mice was followed electrophysiologically for up to 28 days. This reveals that there is a very slow growth of myelinated axons into the distal stump; thevalue of the slopeof the regression line for distance reached by the fastest axons versus time is 0.47 mm/day. However, when the peak-to-peak size of the compound action potential recorded ~3 mm below the crush point was compared with the values from C57BU6j mice at different times after nerve crush, it seems that the number of fibers succeeding in making this slow growth is very small. This is borne out by the results described below. Electron Microscopic Results The figures for axonal counts at 7 days given in Table 1 show that after nerve crush many regrowing unmyelinated axons are present near the lesion in nerves from C57BUOla mice; there are more than 3 times as many unmyelinated axons as found in a normal control nerve. In contrast, there are no more unmyelinated axons 10 mm from the lesion in the crushed

5 Days

f 6 71 * 5 * 15 108 f 4 * IO 225 f 27 level: 7.9 f 2

2 Days

Various

Days after Section

mRNA (fg/mg wet weight) 3 Days

84 f 15 293 + 22 2131 f 281 2549 f 264 288 + 4 2538 f 561 Average unsectioned level: 64 + 30

5 Days 1362 f 376 3526 f 246 3024 f 177

lnacrophage

Dependence

of Sensory Axon Regrowth

:63

A a-

6J Ola

A

,g 6J a 15 .!F

o/-0 *

-l

,

0

24

time in culture

48

72 mocrophage cond. medium

(hours)

5 Days post saphenous nerve crush

1 10

B 20

1

B

I-

J,' 3-

&

4

T 8.1 6J <’ /’

-----4

Ola

6J 010

OL----5 0 Days post saphenous nerve crush

0 time

24

in culture

72

40

, . (hours)

Figure 4. Regeneration

mge

macro

cond.

medium

C NGF-receptor

NGF

6J 125

6 3e

100

1

6J

Ola 010

5000 y .P 4000

1

e

Assessed

10

Electrophysiologically

(A) The distance reached by regenerating low-threshold afferents in the saphenous nerves of C57BUOla mice (open squares and continuous lines), C57BU6J (crossesand dashed lines), and BALBl c (open diamonds and dotted lines) mice at different times after nerve crush. Bars give standard errors. (B) The distance reached by regenerating high-threshold afferents in the saphenous nerves of C57BUOla (open squares and continuous line) and C57BU6J (crosses and dashed line) mice at different times after nerve crush. Bars give standard errors.

2 3000

2000

T

1000

0

r \ 2 ‘;; B E F

- basal levels Figure 3. Levels of mRNA for NGF and Its Receptor In Vivo

In Vitro and

Levels of mRNAs for NCF (A) and NCF receptor (B) in cultured nerve segments. Data are from C57BUOla mice (open squares and continuous lines) and C57BU6J (crosses and dashed lines). The bars at the right show the effect of adding macrophageconditioned medium to nerves at 72 hr and incubating for an additional 3 hr. Data are from C57BU6J mice (open bars) and from C57BUOla (closed bars). In (C) mRNA levels for NGF and NCF receptor are shown from distal sciatic nerve segments 5 days after section in vivo. Data are obtained from the bottom portion of the nerve. The top 3 m m segments adjacent to the lesion site were removed and measured separately (see Results). Figures at each time point are the mean of assays on at least two, but usually three, separate samples of tissue.

nerve than in the nerve that had been cut to prevent reinnervation. Thus the microscopic data support the conclusions drawn from the electrophysiological experiments at 7 days. Total counts of myelinated and unmyelinated axons were made in montages of electron micrographs of whole saphenous nerves between 28 and 30 days after nerve crush or section (nerves were cut in two C57BU Ola and two C57BU6) mice). The counts were made both proximal and distal to the lesion site (-6 mm proximalandl0mmdistal tothecrushsite).Anyaxons found in the distal nerves are derived from regenerating fibers, since at 28 days all the original centrally disconnected axons have degenerated, even those from C57BUOla mice. In this experiment, the results from the cut nerves did not differ from those from the crushed nerves. This is because of the length of time allowed for reinnervation and because the nerve section was deliberately made at a point on the saphenous nerve where the proximal and distal nerve ends

N.WVJn 364

O la

6J

Balb/c

Figure 5. Long-Term Regeneration sessed by Electron Microscopy

As-

Numbers of myelinated (top three graphs) and unmyelinated (bottom three graphs) axons in proximal and distal saphenous nerve stumps in control nerves (open circles and continuous lines joining data from one mouse) and experimental nerves crushed 28-30 days previously (crosses and dashed lines). Results are from C57BUOla mice in the graphs on the left, C57BU6J mice in the middle graphs, and BALB/c mice in the graphs on the right.

800 B g 600 B 400 2 E5 200 0

o+-Prox.

Did.

did not spring apart after sectioning. Hence the data from nerves that were crushed and those that were cut are not distinguished in the figures. Counts were also made from the contralateral (control) nerves. The data are presented separately for myelinated and unmyelinated fibers in Figure 5. Taking the myelinated axons first, it can be seen that in the C57BUOla mice there are fewer axons in the proximal stump of the experimental nerves than in the control nerves and many fewer in the distal stump. Thus there is some degree of retrograde axonal loss of myelinated fibers in these mice following axotomy. Many of those that remain do not succeed in sending axons into the distal stump by28 days. On average, the reinnervation success of the myelinated fibers (percentage of axons in the proximal stump sendingaxons intothedistal stump)isonlyl9% (Table 3). This figure is corrected for the slight decrease in axonal numbers in the distal stump compared with

Table 3. Reinnervation Success of Myelinated and Unmyelinated Axons in Saphenous Nerves of Three Strains of Mice 28 Days after Axotomy

Myelinated Unmyelinated

C57BUOla

C57BU6)

BALBlc

19% 61%

74% 64%

95% 147%

The mean percentageof axons in the proximal nerve stumpgaining access to the distal stump is corrected for the normal reduction in axonal numbersas the nerve travelsdistally by expressing it as a percentage of the mean percentage of proximal axons normally found in the control distal stump.

the proximal that is evident even in the intact nerve and is due to axons branching off from the saphenous nerve to innervate the skin as the nerve progresses distally. In C57BU6) mice there does not appear to be any retrograde loss of myelinated axons in the proximal stump (Figure 6, top), but the reinnervation success is by no means perfect at 74% (Table 3). In contrast, in the BALB/c mice reinnervation is nearly complete (95%). The bottom of Figure 5 shows that for all three mouse strains there is a substantial loss of unmyelinated axons from the proximal stumps. Reinnervation success is also poor in the two C57 strains, only ~60% (Table 3). However, the axonal counts suggest that in the BALB/c mice there is a considerable excess of reinnervating unmyelinated fibers, nearly 58% more than necessary to compensate for the lesion-induced losses. This must be due to multiple sprouts arising from some of the regrowing axons. Failure of Sensory Fiber Regeneration and Success of Motor Fiber Regeneration in the Sciatic Nerve in C57BUOla Mice The sciatic nerve, like the saphenous, develops a marked swelling at the site of a crush. In view of the failure of the axons in the purely sensory nerve to regenerate, it seemed interesting to determinewhether sensory fibers in such a mixed nerve, in which motor regeneration was successful, were impeded in their attempts to regrow. To answer this question, we recorded from dorsal and ventral roots separately on stimulating the nerve below the lesion site to deter-

tvtacrophage 345

Dependence

of Sensory Axon Regrowth

BablC

Figure 6. Poor Sensory but Good Motor Regeneration in the Sciatic Nerves of CS7BUOla Mice

c

Compound action potentials recorded in vitro in dorsal roots (top) and ventral roots (bottom) on stimulating the posterior tibia1 nerve near the ankle (42 m m below the crush site) in the sciatic nerves 29 days after the operation. Scale bar on the left is 0.25 mV for C57BUOla records (on the left) and 0.5 mV for BALB/c records (on the right). Note the very small potential recorded in the dorsal root of the C57BUOla mouse.

If++-

whether regenerating fibers were motor or sensory. The typical results of such an experiment carried out 29 days after crush of the sciatic nerve are illustrated in Figure 6. In the records from the BALB/c mouse, the sizes of the compound action potentials evoked on stimulating the posterior tibia1 nerve at the ankle (~12 mm below the crush) are about equal in 1he dorsal and ventral roots, but the ventral root po-ential is some 30 times larger than the dorsal root one n the records from the C57BUOla mouse. Only at ;timulation distances close to the crush site could arge potentials be evoked in the dorsal roots of 257BUOla mice. As a simple measure of the relative iuccess of afferent and efferent axons in regrowing, rhe ratio of the size (peak height) of the dorsal root ootential evoked by stimulating at least 10 mm distal io the crush site was expressed as a fraction of the peak height of the ventral root potential elicited from the same spot. At 2 weeks, this ratio was on average 3.8 for data from three C57BL/6) mice and 0.015 for data from four C57BUOla mice. Data from experiments covering the time interval from 23 to 34 days after sciatic nerve crush gave an average dorsal root potential of 1 .I mV and an average ventral root potential of 1.04 mV, which yields a ratio of 1.03 for four BALB/c mice. The corresponding figures for four C57BUOla mice were as follows: dorsal root potential, 0.04 mV; ventral root potential 0.68 mV, giving a ratio of 0.06. Putting it another way, although the motor fibers in C57BUOla mice seem to be not quite as good as those in BALBlc mice (O&3/1.04 = 65%) when tested over a long distance, the sensory fibers are much worse (0.0411.1 = 4%). Reinnervation of muscle spindle afferents was investigated following sciatic nerve crush in the thigh by recording the sensory discharge in the nerve to the soleus muscle in response to muscle stretch. Between 14 and 21 days following nerve crush, when motor reinnervation is complete (Lunn et al., 1989), all the soleus muscles from eight C57BL16) mice contained tnine

stretch-responsive afferents. In contrast, only two out of six muscles from C57BLlOla mice yielded responses,and thesewerefrom singleunitsthat needed large stretches to activate only a transient firing. Effect of NCF Administration on Regeneration of the Saphenous Nerve Daily injections of NGF were made into thesaphenous nerves of C57BUOla mice to determine whether regeneration could be improved. Injections were made with calibrated glass capillaries pulled to a fine tip on a microelectrode puller. These were inserted under the perineurium, and a volume of ~0.5 PI (containing 2.5 bg of NGF) was injected manually with a 1 ml syringe connected to the capillary by a fine piece of tubing. In six of the eight mice the injections began on day 3 following saphenous nerve crush; in the other two, injections began on day 1. Injections were made progressively more distally each day, at points below the crush that would have been reached by axons in saphenous nerves from C57BU6J mice on the different injection days. The mean furthest distance reached by 6 days was 4.9 f 0.57 mm (f SEM) in NCFtreated nerves and 2.8 f 0.49 mm in nerves injected with 0.1% serum albumin carrier solution (in which the NGF was dissolved). Thus NGF had a weak (but significant; p < 0.02, by a two-tailed t test) effect, but did not enhance the regeneration rate to a degree approaching that found naturally in other mouse strains. In threeof the mice the regeneration of unmyelinated axons was tested by using stronger shocks to the nerve. The mean distance reached was also increased from 3 mm in controls to 5.5 mm. Discussion Slow Degeneration and Low NGF Production in C57BUOla Mice We previously showed that degeneration in the sciatic nerves of C57BUOla mice was very slow (Lunn et al.,

Neuron 366

1989). In the saphenous nerve there is also very little degeneration even 7 days after axotomy, and macrophage numbers do not increase during this time. An additional new finding is that the distal sciatic nerve stump in C57BL/Ola mice is a poor source of mRNAs for either NGF or its receptor when the nerve is left in vivo. The low NGF mRNA levels in the transected sciatic nerves of C57BUOla mice are not due to a general lack of response to macrophages. In cultured nerve segments, macrophage-conditioned medium stimulates large rises in NGF mRNA levels in C57BLIOla as well C57BU6) mice (Figure 3A). Moreover, C57BL/ Ola-derived macrophages are clearly able to accumulate at the site of a lesion, promoting local swelling. They can also bring about axonal degeneration when transferred to other mouse strains (Perry et al., 1990b) and cause the local production of NGF (see Results). This suggests that the low levels of mRNAs for NGF and its receptor in the bottom portion of C57BL/Ola nerves may result from secondary events due to the absence of macrophages, rather than from intrinsic defects of signaling by macrophages or response to macrophages by nonneuronal cells. It has been shown previously that the long-term NGF mRNA levels (3 days after section or later) reflect the local production of NGF in the distal rat sciatic nerve (Heumann et al., 1987a). In C57BUOla mice the low NGF mRNA level at 5 days, i.e., the failure of NGF production to rise normally, fits well with the recent observation that macrophages not only are involved in the process of degeneration of axons and myelin, but also stimulate the synthesis of NGF in cultured nerve segments (Heumann et al., 1987b). The major responsible agent is the macrophage secretory product interleukin-1 (Lindholm et al., 1987). The macrophage dependence of the long-term regulation of NGF synthesis, however, does not rule out the possibility that NGF production is modulated in addition by macrophage-independent mechanisms, especially during the initial 3 days after section. First, the rapid, transient increase in NGF mRNA occurs within hours in the absence of macrophages and is mediated byatranscriptional activation of an AP-1 site located in the first intron of the NGF gene (Hengerer et al., 1990). Second, glucocorticoids have been shown recently to interfere negatively with the NGF gene transcription in the sciatic nerve(Lindholm etal., 1990) involving a region of the NGF promoter that does not contain the AP-1 site. This negative regulation by glucocorticoids may well be one of the reasons for the much higher levels of NGF mRNA in cultured nerves, e.g., at 3 days (290 k 31 fg/mg wet weight) compared with those found in vivo (165 k 37fg/mg wet weight). In fact, the peak levels of mRNA for NGF in vivo are an order of magnitude lower than the peak levels attained in vitro(see Figure3), which indicates that there is a potential for considerable improvement in NGF supply in the whole animal. Up regulation of the Schwann cell NGF low-affinity

receptor and its mRNA requires the lack of Schwann cell contact with intact axons (Taniuchi et al., 1988; Heumann et al., 1987b). NGF receptor mRNA is not regulated by macrophage secretory products. However, as axonal destruction after axotomy is speeded by macrophage recruitment (Lunn et al., 1989), macrophages indirectly accelerate production of mRNA for NGF receptor, thus leading to 4.9-fold differences between the nerves of C57BL/Ola and C57BL/6) mice at 5 days (Figure 3C). This is compatible with the idea that a major mechanism of Schwann cell NGF receptor regulation resides in the direct interaction between the Schwann cells and their axons. It remains, however, to be investigated whether low-affinity NGF receptors expressed on the Schwann cell surface are functionally involved in the formation of a favorable substrate for sensory nerve cell regeneration (Sandrock and Matthew, 1987). Failure of Afferent Nerve Fiber Regeneration in C57BUOIa Mice In view of the low levels of NGF in the distal stump of C57BL/Ola mice, it was interesting to compare the regeneration capacity of sensory neurons with that of motoneurons, which are apparently not responsive to NGF (although they transiently express low-affinity NGF receptors during regeneration for as yet unknown reasons [Enfors et al., 19891). The results suggest that afferent fiber regeneration is dramatically slowed a few days after beginning fairly normally into the slowly degenerating distal nerve stumps of C57BL/ Ola mice. Results similar to our own have been found (Bisby and Chen, 1990) using the pinch test and transport of radiolabeled proteins to follow the rateof afferent fiber regeneration in the sciatic nerves of C57BLi Ola mice. The remote possibility that the afferent fibers themselves might be at fault in the C57BL/Ola mice can be ruled out by our finding that these afferents can regrow when the distal stump is damaged at closely spaced intervals such that degeneration is directly and mechanically produced (unpublished data). The initial growth of afferents in C57BUOla mice may be possible because at the actual lesion site macrophage recruitment does occur, and this may provide enough trophic support to begin growth. Farther down the nerve, where it is still very largely intact and undegenerated, the few regrowing axons seem to be associated with Remakcells. Interestingly, thesecells, unlike intact myelinating Schwann cells, normallyexpress N-CAM and Ll glycoproteins (Mirsky et al., 1986). An obvious and appealing explanation for the poor regeneration of afferent fibers is the relative lack of NGF and/or its receptor in the distal nerve stumps of C57BUOla mice. In support of this view, there was an improvement of regeneration after repeated injections of NGF into the saphenous nerve. Control injections indicated that the effect was not simply a consequence of injury-induced degeneration. The undramatic natureof the improvement is perhaps not

Macrophage 367

Dependence

of Sensory Axon Regrowth

surprising in view of the difficulty of delivering the material to the right place and the fact that the supply was intermittent rather than continuous. In any case, NGF in the absence of NGF receptors on Schwann cells, as would be the situation in our C57BUOla mice, may not be of much service to axons (Johnson et al., 1988). Our finding that regeneration of proprioceptive afferent fibers as well as cutaneous fibers was slow is interesting, as there is some evidence that muscle spindles may be NGF sensitive (Sekiya et al., 1986) and some very large dorsal root ganglion cells possess high-affinity receptors for NGF (Verge et al., 1989). During early development, however, proprioceptive afferent fibers are not responsive to NGF (Davies, 1986). It is therefore possible that a lack of other NGFlike substances may be involved in the poor regeneration of afferent fibers in C57BUOla mice. A possible role for NGF in promoting afferent fiber regeneration has been investigated and discussed previously by other workers. NGF significantly increased the number of myelinated axons growing across a gap between sectioned ends of rat sciatic nerve within silicon chambers (Rich et al., 1989). However, it might be argued that this is a “pharmacologiCal” effect of high doses of NGF allowing it to bind to receptors of the “real” trophic agent. An example of such concentration-dependent heterologous binding is that of brain-derived growth factor (BDNF; Leibrock et al., 1989) to the high-affinity NGF receptor (Rodriguez-Teb6r et al., 1990). Further support for a role for NGF comes from Bjerre et al. (1974), who showed that fiber regrowth of sympathetic nerve terminals after application of 6-OH-dopamine could be transiently inhibited by anti-NGF antibodies. In addition, collateral reinnervation from sensory neurons can be specifically blocked in the presence of anti-NGF antibodies (Owen et al., 1989). Failure of antibodies to NGF to inhibit regeneration, although they do prevent collateral sprouting, has been reported (Diamond et al., 1987). However, in such experiments the problem of antibody penetration through the blood-nerve barrier has to kept in mind. Nevertheless, it clearly remains to be definitively determined whether lack of NGF or other NCF-like neurotrophic factors, such as brain-derived neurotrophic factor (Leibrock et al., 1989) or NT-3 (see Hohn et al., 1990), is involved in the markedly reduced regeneration of myelinated sensory axons when degeneration does not occur.

Motor Axon Regeneration in C57BUOla

Mice

The success of motor axon regrowth in C57BLIOla micesuggeststhatthereisnothinginherentlyinimical to nerve growth in the undegenerated distal nerve stumps of these mice. The implications are that either motoneurons do not require trophic maintenance to regrow, or they do, but the appropriate material is present even in theabsenceof degeneration. We have found that motor axons regenerating in the sciatic nerves of C57BUOla mice travel almost exclusively with unmyelinated nerve fibers, crossing out of their

own endoneurial sheaths to do so (Brown et al., 1989, J. Physiol., abstract). Perhaps Remak cells are a source of a growth factor for motoneurons. The fact that motor axons are attracted to grow toward nerve stumps (Kuffler, 1986) makes this an interesting possibility. The recent discovery of high levels of ciliary neurotrophic factor in normal rat sciatic nerves and the finding that this growth factor can promote the survival of neonatal motoneurons whose axons have been cut may also turn out to be relevant (Stockli et al., 1989; Sendtner et al., 1990). The success of motor axons, however, makes it unlikely that failure of sensory axons is caused by lack of apo-lipoprotein E (Ignatius et al., 1987).

Retrograde Neuronal Death During fiber regrowth in wild-type animals, the growing tips of the regenerating axons can gain access to the NGF or other neurotrophic factors synthesized in the distal stump. Positive evidence that there is a functional depletion of (macrophagedependent) neurotrophic factors in regenerating neurons of C57BLI Ola mice, compared with wild-type mice, is the loss of some 20% of the myelinated afferent fibers in the mutant nerve proximal to a crush site. The loss of unmyelinated fibers is not dissimilar in all three of the strains we studied (46% lost in the C57BUOla mice, 42% in the C57BU6J mice, and 52% in the BALB/c mice) in spite of their different patterns of NGF secretion. Loss of dorsal root ganglion cells as a whole following axotomy has been described previously (Aldskogius et al., 1985; Himes and Tessler, 1989) and is similar in magnitudetothe lossofaxons in the proximal nerve stump described here. It is likely, therefore, that the short fall of axons in the proximal stump represents death of the whole neuron, rather than a simple withdrawal of the axons.

Conclusions The present results from C57BUOla mice provide clear proof that the macrophage-dependent process of Wallerian degeneration is a necessary prolog for rapid regeneration of most sensory axons. The data raise the possibility that one of the necessary steps may be the production of NGF, which, even in other mouse strains, may not be manufactured in optimal amounts for either complete survival or regrowth of all the axons. It is also clear that there is a degree of heterogeneity among sensory axons in their trophic requirements which is quite apart from the very clear difference now shown to exist between motor and sensory axons. A likely basis for the heterogeneity is varied growth factor requirements, for most axons can grow on a wide range of substrata, a selection of which is available in peripheral nerves. Experimental

Procedures

Animals, Anesthesia, and Surgery When possible, nearly all experiments have compared results from three strains of mice, C57BUOla. C57RU6J, and BALBk. The

Neuron 368

C57BUOla strain was found to possess axons that degenerate very slowly, a trait inherited as a Mendelian dominant (Perry et al., IVVOa). As further investigation has shown no clear evidence that these mice are of the 6 substrain, they will be referred to as C57BUOla (see Perry et al., IVVOa). The C57BU6J strain is genetically related to the C57BUOla strain, but its axons degenerate fast. However, it has been reported that sensory regeneration in this strain is not as good as in some other strains, such as the BALB/c (Xin et al., 1990). The mice were obtained from HarlanOlac, Bicester, Oxfordshire, in the first instance, and subsequently, some were bred in our own animal houses. Anesthesia was carried out with intraperitoneal Avertin (0.2 ml/l0 g weight of 12 g/liter tribromoethanol [Aldrich], originally dissolved in a small volume of tert-amyl alcohol). Sciatic nerves were cut or crushed in the mid-thigh, and the saphenous nerves were cut or crushed with fine (number 5) watchmakers’forceps either immediately below the inguinal ligament or a few millimeters more distally. Instruments were sterilized with 70% alcohol, and skin incisions were closed with fine silk (ep 0.5). No cases of infection or autotomy occurred. Electrophysiological Estimation of Regeneration Saphenous nerves were dissected from about lo-15 m m proximal to the inguinal ligament to below the knee (some IS-20 m m below the inguinal ligament) and studied in vitro at room temperature (22OC-26OC) in physiological saline (135 m M NaCI, 2.5 m M KCI, 10 m M glucose, 1 m M MgCI,, 2.5 m M Car&, 6 m M HEPES [pH 7.41). The proximal end of the nerve, above the crush site, was lifted out of saline onto bipolar silver stimulating electrodes, and the distal end was placed onto a pair of silver recording electrodes. The remainder of the nerve was in the grounded saline. Stimulating pulseswere50orlOO~s in duration for low-threshold myelinated axons and 1 ms for unmyelinated fibers. Maximal voltage available was 9 V. The threshold for the most excitable fibers in normal saphenous nerves was 2 mV). The distance from the crush to the first recording wire was then measured to the nearest millimeter. Finally, the potential 3 m m below the crush site (easily visible from the change in appearance of the nerve at that point) was recorded as a crude measure of the number of axons at that point. To follow sensory regeneration in the sciatic nerve, two methods were used. First, in vitro recordings were made from the nerve to the soleus at varying times after sciatic nerve crush in the thigh while the isolated muscle was stretched. This enabled the reinnervation of muscle spindles to be followed. Second, the whole sciatic nerve from the ankle to the spinal nerve roots was dissected and studied in vitro. Recordings could then be made either from the dorsal or ventral roots while the sciatic nerve was stimulated above or at varying distances below the crush site in the thigh. Alternatively, the roots could be stimulated and recordings could be made from the nerve.

Formvar-coated grids for electron microscopy so that complete countsofaxon numberscould be made. Low power micrographs (1600x magnification) were taken to provide a complete reconstruction from prints that had been enlarged 3 times. The various profiles observed in the photographs were then counted. The categories were intact myelinated axons, unmyelinated axons, and Remak bundles. Macrophage Counts The distal segments of cut sciatic and saphenous nerves were dissected, and a 5 m m portion of the nerve at least 2 m m distal to the cut was embedded in OCT compound (Lamb) and rapidly frozen in isopentane cooled in liquid nitrogen. Sections (IO pm thick) were cut on a cryostat and picked upon gelatinized slides. Sections were fixed with 2% paraformaldehyde in 0.1 M phosphate buffer, and macrophages were labeled with a rabbit antiserum directed against F4/80, a plasma membrane glycoprotein specific for mouse macrophages (Austyn and Gordon, 1981). The details of the characterization of the antibody will be reported elsewhere. The primary antibody was detected by the avidinbiotin-peroxidase method using reagents obtained from Vector Laboratories and diaminobenzidineas thechromogen. Carewas taken to block endogenous peroxidase activity and nonspecific binding. Estimation of mRNA for NCF and Its Receptor The analytical methods have been fully described previously (Heumann and Thoenen, 1986), except that nylon membranes were used for blotting instead of nitrocellulose. In each sample, 20 pg of truncated in vitro transcribed NCF mRNA was used as the recovery standard in the quantitative Northern blots. For each data point, three ~15 mg samples of tissue (three nerves) were analyzed. For the in vivo studies, sciatic nerves were frozen in solid CO? 3 or 5 days after nerve section and degeneration in situ. Control intact nerves from the contralateral side were frozen. To obtain data on events in vitro, sciatic nerves were dissected under sterile conditions and placed in 1.5 ml of Eagle’s basal medium (calcium concentration 1.8 mM; Sigma) without fetal calf serum in sterile tissue culture dishes (Nunclon). These were sealed and placed in an incubator at 37OC in a 5% COJair mixture. Macrophage-conditioned medium was added to some of the samples after 72 hr; these were incubated for an additional 3 hr and then frozen as described above. The macrophageconditioned medium was prepared from cultured resident peritoneal macrophages challenged with LPS overnight. Cells were removed from the medium before use. Acknowledgments We are grateful for support from the MRC, the Wellcome Trust, and the Muscular Dystrophy Group of Great Britain. V. H. P. is a WellcomeSenior Research Fellow, and E. R. L. isan MRC Scholar. Mrs. S. Cahusac is thanked for help with histology, Mrs. I. White for helpwith theelectron microscopy, and Frau Christ1 Lutticken for help with mRNA determinations. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement’ in accordance with 18 USC Section 1734 solely to indicate this fact. Received

Electron Microscopy Nerves were fixed by immersion in 2% paraformaldehyde, 2.5% glutaraldehyde in 0.1 M phosphate buffer for 48 hr, washed in buffer overnight, and postfixed in 1% OsOd in phosphate buffer for 1 hr at 4OC. They were then rewashed in fresh buffer and dehydrated through graded alcohols and propylene oxide. They were embedded in Araldite (EMIX kit ~201, EMSCOPE). Both semithin (0.5 pm) and thin (0.05 pm) sections were cut. The former were stained with toluidine blue and examined under the light microscope. The latter were counterstained with lead citrate and viewed with a Jeol 100 CX electron microscope. Cross sections of the saphenous nerve were placed on

January 1, 1991

References Aldskogius, H., Arvidsson, J., and Grant, G. (1985). The reaction of primary sensory neurons to peripheral nerve injury with particular emphasis on transganglionic changes. Brain Res. Rev. 70, 27-46. Austyn, j. M., and Gordon, S. (1981). F4/80; a monoclonal body directed specifically against the mouse macrophage. J. Immunol. 70, 805-815. Bandtlow,

C., Heumann,

R., Schwab,

M. E., and Thoenen,

antiEur. H.

Macrophage 369

Dependence

of Sensory Axon Regrowth

(1987). Cellular localization of nerve growth in situ hybridization. EMBO J. 6, 891-899.

factor

synthesis

by

sion and possible function of nerve growth factor Schwann cells. Trends Neurosci. 77, 299-304.

receptors

on

Kuffler, D. P. (1986). Isolated satellite cells of a peripheral nerve direct the growth of regenerating frog axons. J. Comp. Neurol.

Benfey, M., and Aguayo, A. j. (1982). Extensive elongation of axons from rat brain into peripheral nerve grafts. Nature 296,150152.

249,57-64.

Beuche, W., and Friede, R. L. (1984).The roleof non-resident in Wallerian degeneration. J. Neurocytol. 73, 767-796.

Langley, J. N., and Anderson, H. K. (1904). The union of different kinds of nerve fibres. J. Physiol. 37, 365-391.

cells

Bjerre B., Bjorklund, A., and Edwards, D. C. (1974). Axonal regeneration of peripheral adrenergic neurons, effects of antiserum to nerve growth factor in mouse. Cell Tissue Res. 748, 441-476. Bisby, M. A., and Chen, S. (1990). Delayed Wallerian degeneration in sciatic nerves of C57BUOla mice is associated with impaired regeneration of sensory axons. Brain Res. 530, 117-120. Cornbrooks, C. J., Carey, D. L., McDonald, J. A., Timpl, R., and Bunge, R. P. (1983). In vivo and in vitro observations on Jaminin production by Schwann cells. Proc. Natl. Acad. Sci. USA80,3850-

3854. Davies, A. M. (1986). The survival and growth prioceptive neurons is promoted by a factor muscle. Dev. Biol. 775, 56-67. Diamond, J., Coughlin, M., Maclntyre, L., heau, B. (1987). Evidence that endogenous is responsible for thacollateral sprouting, tion, of nocioceptive axons in adult rats. USA 84, 6596-6600.

of embryonic propresent in skeletal

Holmes, M., and Vis8 nerve growth factor but not the regeneraProc. Natl. Acad. Sci.

Enfors, P., Henschen,A., Olson, L.,and Persson, H. (1989).Expression of nerve growth factor receptor mRNA is developmentally regulated and increased after axotomy in rat spinal cord motoneurons. Neuron 2, 1605-1613. Hall, S. M. (1986). The effect of inhibiting Schwann cell mitosis on the reinnervation of acellular autographs in the PNS of the mouse. Neuropathol. Appl. Neurobiol. 72, 401-414. Hall, S. M., and Gregson, N. A. (1977). The effects of mitomycin C on the process of regeneration in the mammalian peripheral nervous system. Neuropathol. Appl. Neurobiol. 3, 65-78. Hengerer, B., Lindholm, D., Heumann, R., Ruther, U., Wagner, E. F., and Thoenen, H. (1990). Lesion-induced increase in nerve growth factor mRNA is mediated by c-fos. Proc. Natl. Acad. Sci. USA 87, 3899-3903. Heumann, R., and Thoenen, H. (1986). Comparison between the time course of changes in NCF protein levels and those of its messenger RNA in the cultured rat iris. J. Biol. Chem. 267,92469249. Heumann, R., Korsching, S., Bandtlow, C., and Thoenen, H. (1987a). Changes of nerve growth factor synthesis in nonneuronal cells in response to sciatic nerve transection. J. Cell Biol. 704, 1623-1631. Heurnann, R., Lindholm, D., Bandtlow, C., Meyer, M., Radeke, M. J., Nasko, T. P., Shooter, E., and Thoenen, H. (1987b). Differential regulation of mRNA encoding nerve growth factor and its receptor in rat sciatic nerve during development, degeneration and regeneration, role of macrophages. Proc. Natl. Acad. Sci. USA 84,8735-8739. Heumann, R., Hengerer, B., Lindholm, D., Brown, M., and Perry, H. (1990). Mechanisms leading to increases in nerve growth factor synthesis after peripheral nerve lesion. In Advances in Neural Regeneration Research, F. J. Seil, ed. (New York, NY: Wiley-Liss). Himes, B. T., and Tessler, A. (1989). Death of some dorsal root ganglion neurons and plasticity of others following sciatic nerve section in adult and neonatal rats. J. Comp. Neurol. 284,215-230.

Leibrock, J., Lottspeich, F., Hohn, A., Hofer, M., Hengerer, B., Masiakowski, P., Thoenen, H., and Barde, Y.-A. (1989). Molecular cloning and expression of brain-derived neurotrophic factor. Nature 347, 149-152. Lindholm, D., Heumann, R., Meyer, M., and Thoenen, H. (1987). Interleukin-1 regulates synthesis of nerve growth factor in nonneuronal cells of the rat sciatic nerve. Nature 330, 658-659. Lindholm, D., Hengerer, B., Heumann, R., Carroll, P., and Thoenen, H. (1990). Glucocorticoid hormones negatively regulate nerve growth factor expression in vivo in cultured rat fibroblasts. Eur. J. Neurosci. 2, 795-801. Lundborg, G., Longo, F. M., and Varon, S. (1982). Nerve regeneration model and trophic factors in viva. Brain Res. 232, 157-161. Lunn, E. R., Perry, V. H., Brown, M. C., Rosen, H., and Gordon, S. (1989). Absence of Wallerian degeneration does not hinder regeneration in peripheral nerve. Eur. J. Neurosci. 7, 27-33. Mirsky, R., Jessen, K. R., Schachner, M., and Coridis, C. (1986). Distribution of the adhesion molecules N-CAM and Ll on peripheral neurons and glia in the adult rat. J. Neurocytol. 75, 795-815. Owen, D. J., Logan, A., and Robinson, P. P. (1989).A role for nerve growth factor in collateral reinnervation from sensory nerves in the guinea pig. Brain Res. 476, 248-255. Perry, V. H., Brown, M. C., and Gordon, phage response to central and peripheral ble role for macrophages in regeneration. 1223. Perry, V. H., Lunn, E. R., Brown,

S. (1987). The macronerve injury. A possiJ. Exp. Med. 765,1218-

M. C., Cahusac,

S. (1990a). Evidence that the rate of Wallerian controlled by a single rosci. 2, 408-413.

autosomal

dominant

S., and Gordon, degeneration is gene. Eur. J. Neu-

Perry, V. H., Brown, M. C., Lunn, E. R., Tree, P., and Gordon, S. (1990b). Evidence that very slow Wallerian degeneration is an intrinsic property of the peripheral nerve. Eur. I. Neurosci. 2,

802-808. Politis, J. J., Ederle, K., and Spencer, P. S. (1982).Tropism in nerve cell regeneration in vivo. Attraction of regenerating axons by diffusible factors derived from cells in distal nerve stumps of transected peripheral nerves. Brain Res. 253, 1-12. Ramon y Cajal, S. (1928). Degeneration and Regeneration of the Nervous System, transl. R. M. May (London: Oxford University Press). Rich, K. M., Yip, H. K., Osborne, P. A., Schmidt, R. E., and Johnson, E. M. (1984). Role of nerve growth factor in the adult dorsal root ganglia neuron and its response to injury. J. Comp. Neurol. 230, 110-118. Rich, K. M., Alexander, T. D., Pryor, J. C., and Hollowell, J. P. (1989). Nerve growth factor enhances regeneration through silicone chambers. Exp. Neurol. 705, 162-170. Rodriguez-Tebar, A., Dechant, C., and Barde, Y.-A. (1990). Binding of brain-derived neurotrophic factor to the nerve growth factor receptor. Neuron 4, 487-492. Sandrock, A. W., and Matthew, nerve growth factor promotes nerve. Brain Res. 425, 360-363.

W. D. (1987). Substrate-bound neurite growth in peripheral

Hohn,A., Leibrock, J., Bailey, K.,and BardeY.-A. (1990). Identification and characterization of a novel member of the nerve growth factor/brain-derived neurotrophicfactor family. Nature344,339341.

Sekiya, S., Homma, S., Miyata, Y., and Kuno, M. (1986). Effects of NCF on differentiation of muscle spindles following nerve lesions in neonatal rats. J. Neurosci. 6, 2019-2025.

Ignatius, M. J., Shooter, E. M.. Pitas, R. E., and Mahley, R. W. (1987). Lipoprotein uptake by neuronal growth cones in vitro. Science 236, 959-962.

Sendtner, M., Kreutzberg, G. W., and Thoenen, H. (1990). Ciliary neurotrophic factor prevents the degener,ation of motor neurons after axotomy. Nature 345, 440-441.

Johnson,

Sketelj, J., Bresjanac,

E. M., Taniuchi,

M., and Di Stefano,

P. S. (1988). Expres-

M., and Popovic,

M\. (1989). Rapid growth

Neuron 370

of regenerating axons across the segments of sciatic nerve devoid of Schwann cells. J. Neurosci. Res. 24, 153-162. Stockli, roll, P., cloning, trophic

K. A., Lottspeich, F., Sendtner, M., Masiakowski, P., CarGotz, R., Lindholm, D.,andThoenen, H. (1989). Molecular expression and regional distribution of rat ciliary neurofactor. Nature 342, 920-923.

Taniuchi, M., Clark, H. B., Schweitzer, J. B., and Johnson, M. (1988). Expression of nerve growth factor receptors by Schwann cells of axotomized peripheral nerves. Ultrastructural location, suppression by axonal contact, and binding properties. J. Neurosci. 8, 664-681. Verge, V. M. K., Richardson, P. M., Benoit, R., and Riopelle, R. J. (1989). Histochemical characterization of sensory neurons with high-affinity receptors for nerve growth factor. J. Neurocytol. 78, 583-591. Xin, L., Richardson, P. M., Gervais, F., and Skamene, E. (1990). A deficiency of axonal regeneration in C57BU6J mice. Brain Res. 510. 144-146.