Neutralizing Intraspinal Nerve Growth Factor Blocks Autonomic Dysreflexia Caused By Spinal Cord Injury

Characteristics of the neutralizing antibody to NGF

Before testing the Ab to NGF in vivo, we determined an effective concentration (EC) needed to neutralize NGF in an in vitro neurite response assay. Specifically, PC12 cells were assayed for NGF-dependent neurite outgrowth in the presence of serial dilutions of the Ab. Accordingly, we found that a 1:1000 dilution of the anti-NGF Ab prevented neurite outgrowth in PC12 cells in response to 10 ng/ml NGF (Table 1, Fig.1). Likewise, a 1:10,000 dilution prevented neurite outgrowth in response to 1 ng/ml NGF (Table 1). To test the specificity of the Ab, PC12^nnrcells [that do not express trkA receptors (Green et al., 1986)] were generated to express trkB and trkC receptors for brain-derived neurotrophic factor (BDNF) and neurotrophin-3 (NT-3), respectively. Using a similar dose–response curve, we found that no concentration of the Ab could block neurite outgrowth in response to 10 ng/ml of BDNF or NT-3, indicating a specificity of the Ab for NGF.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

PC12 cell neurite response assay. PC12 cells were cultured for 48 hr without NGF treatment (a) or with 10 ng/ml NGF (b), with 10 ng/ml NGF plus anti-NGF Ab (1:1000; c) or with 10 ng/ml NGF plus anti-NGF Ab (1:2000; d).

The anti-NGF Ab was also tested immunocytochemically on sections from injured rat spinal cord to determine if it labels the same cells that we have previously identified as NGF-IR using an affinity-purified antibody (Conner and Varon, 1992; Krenz and Weaver, 1998a). When used for immunocytochemistry, the neutralizing anti-NGF Ab strongly stained macrophages (Fig.2a,c) and Schwann cells in the dorsal root entry zones and around blood vessels (Fig.2b,d). The neutralizing anti-NGF Ab did not bind to cell types other than those we previously identified with an affinity-purified antibody to NGF (Krenz and Weaver, 1998a), thereby demonstrating lack of nonspecific binding to spinal cord cells.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Digital photomicrographs of horizontal sections of spinal cord at the injury site in the thoracic segment immunostained using the anti-NGF Ab. Sections were taken from a rat 7 d after SCT at T4. Panels a and c depict NGF-IR macrophages at the transection site. Panels b andd illustrate NGF–IR Schwann cells at the dorsal root entry zone subpial rim and near an arteriole. The boxeson panels a and b delineate the region depicted in panels c and d, respectively. Scale bars: a, b, 100 μm;c, d, 50 μm. Arrowspoint to examples of immunoreactive cells.

Intrathecal delivery of antibody to NGF blocks afferent fiber sprouting

We infused the anti-NGF Ab into the entire thoracolumbar intrathecal space caudal to a SCT at T4 to determine whether neutralizing intraspinal NGF for 2 weeks prevents CGRP-IR afferent fiber sprouting. Immunocytochemical detection of the distribution of the anti-NGF Ab in the spinal cord in a rat revealed that it diffused into all the thoracolumbar segments. It was distributed throughout the white and gray matter of the cord in T6–9 and T10–13 and, in L1–5, it penetrated ∼200 μm from the dorsal surface, extending from lamina I–VII (Fig. 3a). No immunofluorescence was found in the spinal cord of an intact, untreated rat (Fig. 3b) or the cerebellum of the anti-NGF Ab-treated rat (Fig. 3c).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Digital photomicrographs demonstrating that the neutralizing antibody (anti-NGF Ab) penetrated the spinal cord after intrathecal administration of the anti-NGF Ab to a cord-injured rat for 2 weeks. The spinal cord had been transected at the fourth thoracic segment (T4). The presence of the anti-NGF Ab (made in rabbit) was visualized by fluorescence immunocytochemistry using an anti-rabbit antibody. Immunoreactivity to the anti-NGF Ab in the dorsal two-thirds of a cord section at T12 of the injured rat is shown in panela. Lack of immunoreactivity in T12 of an intact, untreated rat (b) and in the cerebellum of the anti-NGF Ab-treated rat (c) demonstrates that the labeling is specific. Scale bar, 100 μm (refers to all panels).

Although CGRP-IR fibers and terminals are heavily distributed in laminae I and II of the dorsal horn, only laminae III–V were analyzed because interneurons that likely receive input from primary afferent neurons to mediate spinal sympathetic reflexes are located in these laminae (Joshi et al., 1995; Clarke et al., 1998). The area of these fibers in laminae III–V was quantified by measuring their area within a standard region of the dorsal horn in transverse sections of spinal cord. This area increased markedly at 2 weeks after SCT (Fig.4) as we reported previously (Krenz and Weaver, 1998b), and the increase was blocked by intrathecal infusion of the anti-NGF Ab. In the spinal cord of intact rats, CGRP-IR fibers and terminals in the dorsal horn extended sparsely into laminae III-V (Fig.4a,d). In the SCT animals that received only intrathecal nonimmune rabbit IgG, the area of CGRP-IR fibers in laminae III–V of all cord portions was significantly (50–75%) larger than in the corresponding cord segments in the intact rats (Figs.4b,e, 5). In the SCT rats receiving anti-NGF Ab treatment, this enlargement of the afferent arbor was completely blocked (Figs.4c,f, 5). The area of CGRP-IR fibers in the cord caudal to the transection after anti-NGF Ab was significantly less than in corresponding cord portions of animals receiving nonimmune IgG and was the same as that in intact rats (Fig. 5). Anti-NGF Ab treatment did not block the enlargement of the afferent arbor in spinal segments rostral to the transection (Fig. 5). This was the anticipated result because the anti-NGF Ab should not have reached cord segments rostral to the SCT.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Digital photomicrographs of CGRP-IR fibers in the dorsal horn of control intact rats (a, d), rats 2 weeks after cord injury that received an intrathecal infusion of nonimmune rabbit IgG (b,e), and rats 2 weeks after cord injury that received intrathecal infusion of anti-NGF Ab (c, f). Tissue sections are from T6. Panelsd–f are magnifications of laminae III-V from panelsa–c. The boxes on panelsa–c indicate the regions shown in the higher magnification photomicrographs. Scale bars: a–c, 100 μm; d—f, 50 μm.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

Mean areas of CGRP-IR fibers in laminae III–IV of the thoracic (T) and lumbar (L) spinal segments in rats with no spinal cord injury (intact) and in rats 2 weeks after spinal cord transection (SCT). After SCT, rats received nonimmune rabbit IgG or anti-NGF Ab intrathecally. The CGRP-IR areas in rats treated with rabbit IgG increased significantly (asterisk) compared with intact controls; the CGRP-IR areas in rats treated with anti-NGF Ab were significantly smaller than areas in the same cord segments of rats treated with rabbit IgG (open diamond); the CGRP-IR areas in T1–3 and L1–5 of some spinal rats were larger than the area in T6–9 within the same treatment group (open cross). Each treatment group contained four rats. In the rabbit IgG–treated group, samples could not be obtained for T10-L5 in one rat because of catheter-induced damage to the cord.

After the 2 week incubation of the Ab in the implanted minipump, the effective concentration was redetermined. The same dose–response assay used to assess the Ab before in vivo incubation revealed that a 1:5000 dilution of the Ab was still effective in blocking the neurite response to 1 ng/ml NGF (Table 1). Thus, the EC₅₀ had decreased by, at most, 50% during thein vivo treatment paradigm.

Limited intrathecal delivery of antibody to NGF has local effects on afferent sprouting

A second group of SCT rats was tested to determine whether a local application of anti-NGF Ab in only one portion of the cord caudal to the transection would affect sprouting only in that region of the cord. In this experiment, an animal served as its own control because afferent sprouting in cord regions receiving Ab could be compared to sprouting in regions that did not receive antibody. This design avoided possible interanimal variability. Two weeks after SCT, rats received an intrathecal infusion of anti-NGF Ab or aCSF delivered only to four or five segments of the cord caudal to the transection. Within individual rats, the portions of cord superfused with anti-NGF Ab contained a smaller area of CGRP-IR fibers in laminae III–V than portions that were not superfused (Table2). In addition, in portions of the cord superfused with anti-NGF Ab, laminae III–V had a decreased area of CGRP-IR fibers compared to the area in the same spinal cord portions in rats superfused with aCSF (Table 2). The area of CGRP-IR in caudal portions receiving aCSF did not differ from caudal portions receiving no superfusion of aCSF (Table 2). The areas of CGRP-IR fibers in the segments rostral to the transection were similar in aCSF-treated (10,750 ± 1114 μm²) and anti-NGF Ab-treated (9950 ± 1181 μm²) spinal rats. Therefore, the effects of intrathecal anti-NGF Ab appeared to be caused by a local spinal action of the antibody.

Subcutaneous delivery of antibody to NGF does not block afferent sprouting

In the experiments described above, afferent sprouting occurred unabated rostral to the SCT, and in untreated caudal regions, suggesting that the anti-NGF Ab had acted in the cord rather than by leaking into the circulation to neutralize peripheral NGF. These results did not address another important question, the potential role of peripherally produced NGF in sprouting of central afferent arbors after SCT. Therefore, anti-NGF Ab was delivered subcutaneously to hindlimb targets of the lumbar sensory nerves for 2 weeks, and central sprouting of CGRP-IR afferent fibers from the limb was assessed. This method of anti-NGF Ab delivery effectively blocks peripheral nerve sprouting triggered by peripheral sources of NGF (Diamond et al., 1992). Unilateral subcutaneous infusion of anti-NGF Ab into the right hindlimb had no effect on primary afferent fiber sprouting in the ipsilateral L4–5 spinal cord in rats 2 weeks after SCT. These lower lumbar segments contain terminations of small-diameter afferent neurons innervating the hindlimb. No difference in area of CGRP-IR afferent arbors was observed between left (6915 ± 624 μm²) and right (7301 ± 455 μm²) dorsal horns of these lumbar spinal segments in anti-NGF Ab-treated rats. Likewise, no difference in the dorsal horn area of CGRP-IR was found between animals receiving anti-NGF Ab (above) and animals receiving saline (left: 9520 ± 1752 μm², right: 7738 ± 1474 μm²).

Intrathecal delivery of antibody to NGF blocks autonomic dysreflexia

To evaluate whether anti-NGF Ab treatment would ameliorate autonomic dysreflexia, arterial pressure and heart rate responses to colon distension were assessed 2 weeks after SCT at T4. This experimental stimulus reliably produces well-characterized autonomic dysreflexia in rats after SCT at T4 (Maiorov et al., 1997a). This study was performed in the same rats in which anti-NGF Ab or nonimmune IgG was delivered for 2 weeks to the entire cord caudal to the transection, and afferent fiber sprouting was evaluated (Figs. 4, 5). Baseline arterial pressure in the two groups of rats was similar, whereas heart rate was lower by ∼45 beats/min in the anti-NGF Ab-treated rats (Table 3, Fig.6). Colon distension initiated a marked increase in the mean arterial pressure of ∼30 mmHg in the nonimmune IgG-treated rats (Table 3, Fig. 6). In contrast, the increase in arterial pressure was significantly reduced to 17 mmHg in the animals treated with anti-NGF Ab. Reflex decreases in heart rate were similar in both groups (57–59 beats/min). The increased arterial pressure in the nonimmune IgG-treated rats outlasted the colon distension stimulus by ∼1 min, whereas the response tended to be abbreviated in the anti-NGF Ab-treated rats (Table 3). The diminished pressor responses to colon distension in the anti-NGF Ab-treated animals paralleled the lack of afferent fiber sprouting in their spinal cords (Fig. 5). Conversely, the undiminished afferent fiber sprouting in the nonimmune IgG-treated rats (Fig. 5) corresponded with the full development of autonomic dysreflexia that we have described previously (Maiorov et al., 1997a). Analyzing the changes in arterial pressure in both groups of rats relative to the area of their T6–9 afferent arbor demonstrated significant correlation (r = 0.80). The correlation was made to the afferent fiber areas in the T6–9 segments, because these segments contain the major population of preganglionic neurons controlling the splanchnic visceral circulation, an important constricting vascular bed in the episodic hypertension (Mathias and Frankel, 1993; Krassioukov and Weaver, 1995).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 6.

Pulsatile arterial pressure (AP), mean arterial pressure (MAP), and heart rate (HR) measurements in two rats 2 weeks after cord transection that received nonimmune IgG or anti-NGF Ab. Colon distension for 1 min (between arrows) stimulated an increase in AP and MAP and a decrease in HR.

Validation of the model for autonomic dysreflexia

The results of this study provide important support for the model that NGF-mediated sprouting of the small-diameter afferent fibers is largely responsible for the episodic hypertension that occurs after cord injury. Several predictions were substantiated. Superfusion of anti-NGF Ab over the spinal cord caudal to a transection injury prevented the central sprouting of these primary afferent fibers throughout the thoracic and lumbar spinal cord. In contrast, injection of anti-NGF Ab into the subcutaneous tissue of the hindlimb had no effect on the sprouting after cord injury. The intraspinal sequestering of NGF had an important functional outcome, significantly reducing the development of autonomic dysreflexia.

NGF can stimulate sprouting in the CNS

Other experiments have shown that the presence of NGF in the spinal cord can lead to sprouting of CGRP-IR fibers. NGF-expressing fibroblasts implanted into the cord induce sprouting of these fibers (Tuszynski et al., 1994, 1996; Grill et al., 1997b). Mice that express a NGF transgene in oligodendrocytes have ectopic CGRP-IR fibers in the white matter of their spinal cords (Ma et al., 1995). Conversely, a study of pain mechanisms after cord injury also showed, like our observation, that treatment with an antibody to NGF prevented CGRP-IR fibers from sprouting in the dorsal horn after spinal hemisection at T13 (Christensen and Hulsebosch, 1997). CGRP-IR afferent fibers sprout in the dorsal horn in this model of cord injury-induced allodynia and hyperalgesia. However, this study did not describe a physiological outcome of that blockade of afferent sprouting.

The effect of anti-NGF antibody on autonomic dysreflexia

The blockade of sprouting by intrathecal anti-NGF Ab was associated with an altered physiological response to SCT. Autonomic dysreflexia, which is well developed in rats 2 weeks after SCT, was markedly reduced by this treatment. The increase in mean arterial pressure in response to colon distension in anti-NGF Ab-treated animals was reduced by 43% compared to nonimmune IgG-treated animals. Our study suggests a causal relationship between small-diameter CGRP-IR afferent sprouting and autonomic dysreflexia, because the suppression of both by anti-NGF Ab treatment was well correlated. Other observations also imply a causal relationship between afferent sprouting and dysreflexia. First, the time course of development of small-diameter afferent sprouting is similar to that of autonomic dysreflexia (Krassioukov and Weaver, 1995; Maiorov et al., 1997a). Second, the magnitude of a spinal sympathetic reflex is known to increase when the afferent input to the spinal cord is increased (Koizumi et al., 1970; Sato and Schmidt, 1973).

Although intrathecal anti-NGF Ab treatment significantly reduced the hypertensive response, the treatment did not entirely eliminate dysreflexia. This incomplete blockade of dysreflexia, despite complete blockade of afferent sprouting, indicates that other mechanisms, independent of the actions of NGF on small-diameter afferent neurons, contribute to the development of autonomic dysreflexia. Indeed, possible actions of NGF on other spinal cord cells may not have been completely blocked. In addition, sprouting of myelinated afferent fibers unresponsive to NGF, sprouting of interneurons, alterations in central or peripheral neurotransmission, and changes in baroreceptor reflexes could also contribute to exaggerated reflexes (Lee et al., 1994). Finally, colon distension could initiate changes in gastrointestinal motility, indirectly activating vagal afferent fibers projecting to the brainstem (Janig, 1996). This activation could lead to vagally-mediated effects on the heart and to changes in sympathetic outflow in cord segments rostral to the injury site (Pittam et al., 1988). If such vago-vagal or vago-sympathetic reflexes were excitatory after cord injury, they could have contributed to the increases in arterial pressure during autonomic dysreflexia and would not have been blocked by our anti-NGF treatment. After an incomplete injury of the cord, such supraspinally mediated reflexes might even impact on the segments caudal to the injury.

The partial blockade accomplished in our study is actually an ideal effect, because total blockade of dysreflexia is not the desired outcome of treatment in humans. The symptoms of minor episodes of dysreflexia, such as modest changes in arterial pressure, heart rate slowing, and sweating are the few indicators to cord-injured people that sensory input from their lower body has changed, either because of a normal physiological process or to disordered function (Mathias and Frankel, 1992). Partial blockade of dysreflexia would reduce the severity of hypertension that renders it incapacitating and life-threatening while permitting cord-injured people some remaining awareness of changes ongoing in their body below their injury.

Some additional effects were seen in rats treated with intrathecal anti-NGF Ab. Animals receiving anti-NGF Ab did not develop the elevated heart rates typically observed after a T4 SCT (Krassioukov and Weaver, 1995; Maiorov et al., 1997a). The nonimmune IgG-treated rats had increased rates (>500 bpm), but those treated with antibody to NGF did not. Neurons controlling heart rate (near segment T2) have lost no input with a T4 injury and were not exposed to anti-NGF Ab. We speculate that the increased heart rate in the spinal rat is, in part, a secondary response associated with the vigorous ongoing spinal reflexes present in these rats throughout the day. These reflexes may affect adrenal preganglionic sympathetic neurons, promoting the release of circulating epinephrine that would increase heart rate. In six of the seven anti-NGF Ab-treated rats, we observed a reduction in skeletal muscle and bladder sphincter spasticity during the 2 week survival time. Typically, SCT rats develop hyper-reflexive responses within this time that include bladder sphincter spasms and limb muscle contractions during manual compression of the urinary bladder to produce voiding. In the anti-NGF Ab-treated SCT rats, less hindlimb kicking occurred during bladder compression, and the bladder sphincter was not in spasm. The dampened reflexes in the anti-NGF Ab-treated rats may not have stimulated the release of epinephrine adequately to increase heart rate.

The heart rate changes during dysreflexia in our T4 cord transection model would be predicted to result from activation of the baroreceptor reflex. The vigorous bradycardia, despite a smaller increase in arterial pressure that occurred in the rats treated with antibody, could suggest sensitization of some component of the baroreceptor reflex, a change that we have not observed previously (Maiorov et al., 1997a). Such an effect would not likely be caused by a direct action of anti-NGF because it would not have reached any neurons involved in this reflex circuit. Instead it might relate to a mechanism secondary to the antibody treatment such as a greater potential for reflex bradycardia when adrenergic drive to the heart is lower. Alternatively, a vago-vagal reflex initiated by the colon distension itself may have played a greater role in initiating bradycardia when arterial pressure responses were smaller.

Conclusion

We have provided the first demonstration that blocking the activity of NGF in the injured spinal cord with a neutralizing antibody not only blocks the maladaptive sprouting of small-diameter primary afferent neurons in the dorsal horn, but also greatly reduces the life-threatening condition autonomic dysreflexia. Together our data reveal a likely mechanism for the etiology of autonomic dysreflexia and a strategy for its prevention. This blockade of intraspinal NGF probably will not adversely affect regeneration or recovery of cord function. Regeneration of the injured cord likely can be promoted by a combination of other neurotrophic agents such as fibroblast growth factor, NT-3, and BDNF (Cheng et al., 1996; Grill et al., 1997a;Menei et al., 1998). Because delivery of growth factors to the injured cord may have deleterious effects such as stimulating the growth of the CGRP-IR afferent fibers (Tuszynski et al., 1996; Grill et al., 1997b), in addition to beneficial effects, regeneration paradigms must take all possibilities into account.

Blocking the sprouting of the small-diameter, CGRP-IR afferent fibers by developing treatments to neutralize the intraspinal effects of NGF could impede the development of autonomic dysreflexia and markedly improve the quality of life of the cord-injured person. Prevention of the condition would be a more satisfying solution than the treatment of its symptoms with their unpredictable occurrence and severity. Moreover, because primary afferent neurons also play a role in other disabling conditions that develop after cord injury, a treatment to reduce autonomic dysreflexia would likely also block muscle spasticity, urinary bladder dysfunction, and some types of chronic pain (Botterell et al., 1953; McGuire and Brady, 1979; Cohen et al., 1988; Xu et al., 1992; Kruse et al., 1993; Middleton et al., 1996; Yezierski, 1996).