Journal o f Protein Chemistry, Vol. I1, No. I, 1992

Probing the Structure-Function Relationship of Nerve Growth Factor Steven Vroegop, 1'3 Douglas Decker, 1 Jessica Hinzmann, 2 Roger Poorman, 2 and Stephen Buxser I Received August 23, 1991

We compared the receptor binding, antigenicity, biological activation, and cell-mediated proteolytic degradation properties of mouse nerve growth factor (mNGF) and human NGF (hNGF). The affinity of hNGF toward human NGF-receptor is greater than that of mNGF, but the affinity of mNGF toward rat NGF-receptor is greater than that of hNGF. Thus, the specificity of the interaction between NGF and its receptor resides both on the N G F and on its receptor. Using a group of anti-NGF monoclonal antibodies that competitively inhibit the binding of NGF to receptor, sites differing between mNGF and hNGF were detected. Together, these results indicate that the sites on hNGF and mNGF, responsible for binding to NGF-receptor, are similar but not identical. In comparing the relative abilities of mNGF and hNGF to stimulate a biological response in PC 12 cells, we observed that mNGF was better at stimulating neurite outgrowth than was hNGF, consistent with the differences observed for receptor binding affinity. However, the EDso for biological activation is approximately 100fold lower than the Kd for receptor occupancy, and, thus, the dose-response curve is not consistent with a simple activation proportional to receptor occupancy. The data are consistent with a model requiring a low-level threshold occupancy of NGF-receptor (Ka = l0 -9 M) in order to stimulate full biological activity. Finally, we observed the degradation of NGF by PC 12 cells. We found that the NGF molecule is significantly degraded via a receptor-mediated uptake mechanism. Together, the data provide insight into regions of the NGF molecule involved in contacts with the receptor leading to formation of the N G F :NGF-receptor complex. Additionally, they establish the link between occupancy of receptor and biological activation and the requirement for receptor-mediated uptake in order to degrade NGF proteolytically in cultured PCI2 cells. KEY WORDS: Nerve growth factor (NGF); structure/function comparison; NGF-receptor; neurite outgrowth; biological activation.

binding is critical to understanding the requirements for initiation of biological activity. This process can be greatly aided by the use o f closely related but nonidentical forms of the hormone such as those arising naturally by genetic divergence, particularly forms that differ detectably in binding or biological activity. Antigenicity o f the hormone provides another clue to surfaces on the hormone that m a y be involved in hormone :receptor interactions, since antigenic sites and receptor binding sites must be present on the surface of the native protein. If antibodies can be obtained that recognize receptor binding sites, then

1. I N T R O D U C T I O N Several approaches contribute mining the structure-function tein hormone or growth factor. o f particular regions o f the

significantly to deterrelationship of a proFirst, the contribution hormone to receptor

t Cell Biology, The Upjohn Company, Kalamazoo, Michigan 49001. 2Biopolymer Chemistry, The Upjohn Company, Kalamazoo, Michigan 49001. 3To whom all correspondence should be addressed. 71

0277-8033/92/020f)-0071$06.50/0 © 1992 Plenum Publishing Corporation

72 they provide a tool for further characterizing a region indispensable for function on the surface of the hormone. Second, the relationship between receptor binding and biological activity provides an essential assessment of hormone function and indicates how the structural differences are translated functionally. Finally, an important mechanism for regulation of protein hormone signals is degradation of the protein to attenuate the biological activity. Understanding how degradation occurs is needed to fully understand the functional cycle of the protein. We studied these properties for the neurotrophic factor nerve growth factor (NGF). The primary structure of N G F among mammalian species differs by about 10% (Meier et al., 1986). Until recently, analyses of nerve growth factor were restricted because of the limited supply of N G F from sources other than male mouse submaxillary glands. However, in one study, a sufficient quantity of N G F from bovine seminal vesicles was purified to compare the biological and antigenic properties of bovine N G F (bNGF) with mNGF (Harper et al., 1983). No consistent difference between the biological activities of N G F from these two sources was reported in spite of significantly different antigenic properties of mNGF and bNGF. This was interpreted as evidence for strict conservation of structure in the regions of the molecules responsible for biological activity. Recently, the cloning, expression, and purification of biologically active human N G F (hNGF) was accomplished (Buxser et al., 1991). This provided an opportunity to perform a detailed comparison of the binding and biological activation properties of N G F from these two species. The binding properties of mouse N G F (mNGF) have been studied extensively using two sources of receptor from established cultured cell lines: rat PC12 pheochromocytoma cells (Woodruff and Neet, 1986; Schechter and Bothwell, 1981; Bernd and Greene, 1984; Landreth and Shooter, 1980; Buxser et al., 1990a) and A875 human melanoma cells (Fabricant et al., 1977; Buxser et al., 1983a; Buxser et al., 1983b; Grob and Bothwell, 1983; Grob et al., 1985). It is now known that the primary structure of NGF-receptor present in PC12 cells from rat (Radeke et al., 1987) differs by approximately 10% from that of NGFreceptor in human melanoma cells (Johnson et al., 1986). Whether these structural differences lead to different binding properties or biological activation has not been completely explored. Furthermore, if species specificity does exist, it is not known whether the source of the specificity resides with the hormone

Vroegop et

aL

and/or with the receptor. It is important to examine the species specificity in order to determine if the active site on N G F is highly conserved between species. If species specificity does not exist, then the structure of N G F at the active site should be conserved between species. Conversely, a high degree of species specificity implies divergence of the active site. A priori, the best way to test species specificity would be to determine the relative biological activities. However, biological activation does not directly reflect the molecular details of the hormone :receptor interaction, since binding is optimal at concentrations approximately 100-fold greater than those required to stimulate neurite outgrowth. Thus, it is important to determine not only the relative biological activities of human and mouse NGF, but also the relationship between occupancy of receptor and activation of biological activity. We examined both binding to receptor and activation of neurite outgrowth to understand this relationship from the perspective of hNGF and mNGF. Another way to investigate the surface structures of similar but nonidentical forms of N G F is to examine the relative reactivity of antibodies on different forms of the growth factor. Fortunately, we recently developed and characterized a family of monoclonal antibodies that competitively inhibit binding of N G F to NGF-receptor (Vroegop et al., 1992). Competitive inhibition indicates interaction between the antibody and N G F on the same face of the molecule that associates with NGF-receptor. Thus, this approach also allowed us to probe regions on the surface of N G F involved in association with NGF-receptor. Since one of the more important properties of NGF in vivo is its ability not only to stimulate differentiation but to maintain the differentiated state (Levi-Montalcini, 1987), an important question to address is, how is the action of N G F extinguished? Additionally, differences in biological activity between hNGF and mNGF could be attributed either to differences in the ability to bind to the receptor or to differences in the lability of the forms. Internalization followed by proteolytic degradation is a mechanism anticipated for degeneration of the signal produced by a protein growth factor. Therefore, we observed the degradation of NGF by target cells. In sum, the results presented below compare the binding, biological activation, antigenic and proteolyric degradation properties of mNGF and hNGF. They provide insight into sites on NGF that may constitute the receptor recognition site.

Comparison of Human and Mouse Nerve Growth Factor 2. MATERIALS AND METHODS Preparation of microsomal membranes and radioiodination of N G F were described in detail previously (Buxser et al., 1983a). We did note that radioiodination of human N G F resulted in J25I-labeled N G F with lower specific radioactivity than we observed using equivalent amounts of mouse NGF. In some experiments, this resulted in fewer counts bound and may contribute to greater standard deviations for means in experiments using ~25I-human NGF. MAb were raised against mNGF and purified before use, as described previously (Vroegop et aL, 1992). 2.1. Sources of mNGF and hNGF Murine flNGF was purified from submaxillary glands of male mice according to a previously published technique (MoNey et al., 1976). The gene for human N G F was cloned into a Baculovirus vector and expressed in virus-infected SF-9 insect cells. Human N G F was purified from the culture supernatants of the infected cells using a single-step affinity chromatography procedure. The cloning, expression, and purification were recently described in detail (Buxser et al., 1991). 2.2. Generation of Monoclonal Antibodies and Assays for Their Use The generation and characterization of the monoclonal antibodies used and the affinity of each of the MAb for mNGF was precisely determined, as described elsewhere (Vroegop et al., 1992).

73 assays described in the present report result in determination of apparent KdS for mNGF and hNGF. In contrast to the apparent /(as, determination of the true KdS for binding require determination ,of the Ka of binding to 125I-NGF concomitantly. However, determination of the apparent KdS are sufficient to compare the relative binding affinity of the MAb for mNGF and hNGF and, therefore, only apparent KdS are reported below.

2.4. Binding Assays Binding of t25I-NGF was performed as described previously (Buxser et al., 1990a). Briefly, 200/zl samples contained 0.1 nM 12SI-NGF, 25 ]zg of A875 microsomal membranes, or 50/~g of PCI2 microsoreal membranes, and a concentration of unlabeled mNGF or hNGF, as indicated, in Krebs'-Ringer solution buffered with 20 mM HEPES (pH 7.3) and containing 0.1% bovine serum albumin. All binding assays for the following experiments were performed using 400/~1, fine-tipped microfuge tubes. After incubation for 45-60 rain at room temperature, the tubes were microfuged at 10,000 g for 2 rain, frozen in a dry ice-ethanol bath, and the tips, containing the membrane pellets, were cut off into separate test tubes for counting in a gamma spectrophotometer.

2.5. Data Analysis The data for binding assays were analyzed using nonlinear least-squares analysis (Yamaoka et aL, 1981) by fitting the following equation to 'the data: cpm Bound =

2.3. Inhibition of t2Sl-mNGF Binding to MAb by Unlabeled mNGF or hNGF The radioimmunoassay (RIA) was described previously, in detail (Vroegop et al., 1992). Briefly, MAb was adhered to wells in a 96-well microtiter plate and washed to remove nonadherent MAb. The wells were blocked and a solution resulting in a final concentration of 32 pM J2SI-NGF was added along with a concentration of unlabeled mNGF or hNGF ranging from 0.1 pM to ! 0 nM. The plates were incubated overnight at room temperature and washed four times. Individual wells were cut into test tubes for counting in a gamma spectrophotometer. A saturating concentration of radiolabeled mNGF was present with respect to MAb in the RIAs. Therefore, the

BoK~ Kd + [NGF] + Nonspecific binding

(l)

where/3o is the total cpm bound, and Ka is the equilibrium dissociation constant for unlabeled NGF. The lines shown are theoretical curves calculated by substituting the values for the experimentally determined parameters into the equation to calculate the expected cpm bound over the range of concentrations used in the experiment, in no case was a statistically significant improvement in the fit of the data obtained when an equation with two classes of binding sites was used. Theoretical curves for data fit to other equations were determined similarly (i.e., determining the parameters from the data and calculating theoretical values based on the fitted parameters),

74

Vroegop et aL

2.6. Biological Activity

with an equal volume of ice-cold 10% trichloroacetic acid. Pellets were washed with 10% ice-cold trichloroacetic acid and counted in a gamma spectrophotometer. For each concentration of NGF, the equation of monoexponential decay was fitted to the individual values for total cpm precipitated at each time:

Biological activity was measured by observing the regeneration ofneurites 24 hr after replating PCI2 cells on fresh plates using cells predifferentiated for 68 days with NGF, as described by Black and Greene (1982). Culture medium for biological assays consisted of Delbecco's modified Eagle's medium (DME) supplemented with 2% fetal bovine serum (FBS) and 1% ITS + (Collaborative Research, Inc., Bedford, Massachusetts). Predifferentiation of cells was done in DME supplemented with 10% horse serum, 5% FBS, and 50 ng/ml mNGF. The ceils were washed extensively before replating to break off neurites and to remove the mNGF.

cpm precipitated

where Co is the total cpm of 125I-NGF added at time 0, t is the time in days, and k~xp is the experimental rate constant in units of day -1

3. RESULTS We examined the relative abilities of hNGF and mNGF to compete for binding to NGF-receptor on A875 microsomal membrane preparations using both 125I-labeled mNGF and J25I-labeled hNGF (Fig. 1). We consistently observed that hNGF had a four- to sixfold higher affinity toward human NGF-receptor than did mNGF whether measured by using ~25IhNGF or ~25I-mNGF (Table I). It is also important to note that the binding obeyed single-site Langmuir binding isotherms regardless of the type of labeled and unlabgled N G F used. Thus, even if species in addition to the full-length form are present, such as the des-8 form o f m N G F that appears in preparations

2.7. Determination of the Rate of Proteolytic Degradation of mNGF PC12 cell cultures were prepared, identical to the cultures used to determine neurite outgrowth. Cells were incubated throughout the experiment in the presence of 2% fetal bovine serum. At time =0, 2.5 pM '25I-mNGE was added along with unlabeled mNGF sufficient to bring the final concentration of N G F to the molarity indicated. The cultures were incubated at 37°C for 7 days. Each day the culture supernatant from individual wells was removed and precipitated 12000-

A

11000-

(2)

= Co e -c'k"p)

',:=l

B

10000c-

9000"

O m

150010" 7000-

I,

7000-

£9 z

6000"

~

6000-

I

5000"

± 50oo-

LL

, oo-

4000 -

E CL

,,3000"

°'°*-

r,_) 2000-

.

1000" 0

v vwonl

0.0001

0.001

| vl,v,o,!

0.01

v ,.,.r,~

0.1

,

v iw=

I

1

NGF, nM

. i

~ , H,..I

10

~

' IOO0-

v I 1,,,,~

100

,

i i

0

v , ,.vml

O.OOOl 0.001

i i i . ~

0.01

i

, ,vl~

0.1

,

v ,11~

i

1

i v v.l~l

10

,

1 ,vv~

,

,,

100

NGF, nM

Fig. 1. Competition for binding of 'zSI-mNGF (A) or ~251-hNGF (B) by unlabeled mNGF or hNGF on A875 membranes. Twenty-five micrograms of microsomal membrane from A875 cells were incubated with 0.1 nM '25I-NGF in the presence of the concentration of unlabeled mNGF (open circles) or hNGF (solid circles) indicated for 45 min at room temperature, Specific binding is shown as the mean (±SE) of three replicate samples. The lines shown are theoretical curves based on the parameters generated from fitting Eq. 1 to the data. Nonspecific binding was determined as a fitted parameter and was 2557 cpm and 3051 cpm for (A) and (B), respectively.

Comparison of Human and Mouse Nerve Growth Factor

75

Table

radioiodination process, the different forms have identical receptor binding characteristics. We also compared the binding of hNGF and mNGF to the receptor present in microsomal membranes prepared from PCl2 cells. In contrast to the results using human NGF-receptor, mNGF has a four- to fivefold greater affinity for rat NGF-receptor than does hNGF (Fig. 2). Using ~2SI-hNGF, the standard deviations of the pooled data sets were too targe to distinguish the relative affinities accurately (Table I). However, this does not alter the fundamental conclusion that hNGF and mNGF recognize human NGF-receptor and murine NGF-receptor with different affinities. As noted above for binding to human NGF-receptor, binding to rat NGF-receptor obeys single-site Langmuir binding isotherms consistent with the presence of a single set of binding properties for N G F whether a single or more-than-one chemical form of N G F is present. Previously, we generated six MAb to mNGF and characterized their properties in detail (Vroegop et al., 1992). Here we examined the ability of each of the anti-mNGF MAb to bind to m N G F and hNGF. Three of the MAb--M4, M 12, and M 14A---recognize mNGF with high affinity, but recognize hNGF only weakly (Fig. 3). The relative Kds for binding to

I. Relative Kds for Binding of mNGF or hNGF to NGFReceptor in Membrane Fractions from A875 or PCI2 Cells

12SI-NGF

NGF competitor

Receptor

Kd (nM)

N"

Mouse Mouse Human Human Mouse Mouse Human Human

Mouse Human Mouse Human Mouse Human Mouse Human

A875 A875 A875 A875 PC 12 PCI2 PCI2 PC12

1.58i0.48'* 0.31 + 0.07 1.70±0.33" 0.41 + 0.08 0.064 i 0 . 0 0 9 *b 0.29 • 0.08 (0.71 +0.41) 0.26~0.05

t0 8 3 4 2 3 2 2

*(p