Molecular ~rn~~~ol~~, Vol. 29, No. 9. pp. IOSI-1088, 1992

0161-5890/92 $S.oO+ 0.00 Pergamon Press Ltd

Printed in Great Britain.

IMM~NUD~MINANT STRUCTURES OF HUMAN GROWTH HORMUNE IDENTIFIED BY ~OMO~~~-SCANNING MUTAGENESIS JOSEPHR. WEBER,* CHRISTOPHER NELSON,* BRIANC. CUNNINGHAM,~JAMESA. WELLS? and SHERMANFONG*$ *&!partment of ~MMu~obio~ogyand the f%kpartment of Protein Engineering,

Genentech Inc., 450 Pt. San Bruno Boulevard, South San Francisco, CA 94080, U.S.A.

(First received 3 December 1991; accepted in ~eo~sed~orrn 30 ~u~~~$~ 1992) Attract-Homolog-sunning mutagenesis has been reported to be useful in elucidating the antigenie epitopes recognized by monoclonal antibodies and hGH binding to its receptor. However, little is known about which structures are recognized as immunodom~na~t by murine serum antibodies. Therefore, the previously published series of hGH homologs and additional mutants of human placental lactogen (hPL), porcine growth hormone (pGH), and human prolactin (hPRL) were examined for their interaction with murine serum derived anti-hGH antibodies. As compared to wild-type hGH, nine of the nineteen segment substituted mutants tested showed a significant reduction in binding to anti-hGH sera. These disruptive substitutions mapped to 5 regions on a structural model of hGH: the length of helix 1 (residues 1l-33), the Ioop between the first disulfide bond and helix 2 (residues 54-741, the beginning of helix 3 (residues 109-l 12), the carboxyl half of helix 4 (residues 167-182), and the final carboxyl terminus segment of the molecule (residues 184-191). In terms of the current structurat model, three of the five immunodom~nant regions (the loop between residues 54-74, central portion of helix 4 to the carboxyl terminus and part of the amino terminus region of hefix 1) closely overlaps the hGH receptor binding epitopes.

INTRODUCTION

The investigation of small globular proteins of known primary and tertiary structures as model antigens has provided useful information in the understanding of protein antigenicity and immune recognition (Reichlin, 1975; Benjamin ct al., 1984). One approach to the elucidation of antigenic structures of proteins has been to investigate the binding of antibodies to a panel of evolutionally conserved proteins that differ only slightly in primary structure (Urbanski and Margoliash, 1977a, b). However, this approach has been limited in practice by the availability of suitable homologs of the protein under study. An alternative to naturally occurring homologs is the use of recombinant fusion proteins, sitedirected chemical modification, and site-directed mutagenesis methodologies to probe antigenic structures (Cooper et al., 1987; St. Clair et al., 1988, Dowbenko et al., 1988; Cunningham et ai., 1989). In this report, we have explored the use of a strategy termed homologscanning mutagenesis ~Cunningham et al., 1989) to determine the immunodominant antigenic structures SAuthor to whom correspondence should be addressed: Sherman Fong, Department of Immunobiology, Genentech Inc., 460 Point San Bruno Blvd., South San Francisco, CA 94080, U.S.A. Abbreviations: hGH, human growth hormone; hPL, human prolactin; pGH, porcine growth hormone; hPRL, human placental lactogen; mGH, mouse growth hormone; BBS, borate buffered saline; PM, powdered milk in BBS; pNPP, p-nitrophenyl phosphate.

associated with serum antibody recognition of human growth hormone (hGH). Both competitive binding between two mAb and cross-reactivity to related molecules have been used in defining epitopes on hGH (Ivanyi, 1982; Retegui et ai’., 1982; Surowy et al., 1984; Vita et al., 1986; Strasburger et al., 1989). By these methods, however, the precise locations of the antigenic structures could not be elucidated. More recently, epitopes responsible for binding to a panel of eight mAb, as well as to the cloned liver receptor for hGH, were identified to high resolution by homolog-scanning mutagenesis (Cunningham et al., 1989). Chimeric versions of hGH were produced by systematically substituting short segments of primary sequence with analogous sequences from the homologous hormones human placental lactogen (hPL), porcine growth hormone fpGH), and human prolactin (hPRL). The ability of chimeric proteins thus produced to selectively disrupt binding to only one or a small subset of the mAb, along with other lines of evidence, indicated that the variant proteins had retained the overall tertiary structure of wild-type hGH. Each of the eight mAb epitopes individually formed a patch when mapped upon a structural model of hGH. Collectively, these eight epitopes covered much of the hormone’s surface, and provided a considerably higher resolution map than could have been obtained by an antibody competition approach. In this report, we have applied a similar strategy to identify regions of hGH which compose, or at least have a dramatic influence on, serum polyclonal IgG anti-hGH antibody binding.

JOSEPH R. WEBERet al.

1082

MATERIALS

AND METHODS

mAb 8 determined hGH in ELISA

hGH mutants

and wild-type

ELISA plates (Nunc-Immuno Plates, InterMed, Denmark) were precoated for 1618 hr at 4°C with 2 pg/ml of affinity purified rabbit anti-hGH polyclonal antibodies (rabbits injected with hGH in CFA) in BBS (0.1 M borate buffered saline, pH 8.4) and then blocked for at least 1 hr with 2% PM (powdered milk in BBS). Plates were reacted for 2 hr with mutant and wild-type hGH diluted in 2% PM. Captured antigen was detected by incubating for 2 hr with lOpg/ml mAb 8 in 2% PM followed by 1 hr with AP (alkaline phosphatase)labeled goat anti-mouse IgG (Chemicon, Temecula CA) diluted l/500 in PBS, and then developing with 1 mg/ml pNPP (p-nitrophenyl phosphate; Sigma, St. Louis, MO) in 0.05 M carbonate/bicarbonate buffer, pH 9.8, with 0.001 M MgCl,. After 30 min, absorbance at 405 nm was recorded with a plate reader (Molecular Devices, Palo Alto, CA). All solutions were added to plates at lOOpl/well, except for blocking with 2% PM, which was done at 200 pi/well. Plates were washed between steps with 0.05% Tween 20 in BBS, and all incubations were at room temp unless otherwise indicated.

The mutants listed in Fig. 1 were produced by cassette mutagenesis or restriction selection, using short sequences derived from the homologous hormones hPL, pGH and hPRL, which differ from the hGH sequence by 15, 32 and 77%, respectively. Details of their construction, expression in E. coli, and subsequent characterization have been published (Cunningham et al., 1989), except for the addition of five new mutants obtained and characterized by identical methods. Mouse anti-hGH

equivalency for mutant

serum

Two month old female BALB/c mice were injected S.C. twice weekly for 4 weeks with 100 pl of 250 pg/ml hGH in PBS. For each mouse, serum collected at the end of weeks 3 and 4 was combined after testing positive for anti-hGH IgG in an ELISA. For the five sera tested, end point titers, defined by the last serum dilution producing a signal twice the background, ranged from l/3200 to l/25600 under the conditions of the serum anti-hGH assay described below.

Loop 46-52 1 10 20 30 40 50 FPTIPLSRLFDNAML~HRLHQLAFDTYQEFEEAYIPKEQKYSFLQNPQTSL --AM---S--S--"---QH-----A---K---R----EG_F

hGH mGH pGH 11-33 hPL 12-25 hPRL 12-19 hPRL 22-33 hPL 46-52 pGH 48-52

__________~__“__-Q”-----~---~---~------------------___________~_____-____--~-------------------------------------~-~~--y--------------------------_-----_ ---------------------~--SE--SQ--K--_--~---------------------------------------------------------------~~~~--~ ---------------------------------------------_-~_~~~

Loop 54-74 HELIXLOOP 97-104 60 70 80 90 100 CFSESIPTPSNREETQQKSNLELLRISLLLIQSWLEPVQFLRSVFANSLVYG

hGH mGH hPRL 54-74 pGH 57-73 LBOD t hPRL 88-95 hPRL 97-104

HELIX

hGH mGH hPL 109-112 hPRL 126-136 hPRL 137-145 hPRL 146-152

hGH mGH hPRL 158-159 pGH 167-182 C182A pGH 184-191 * designates t designates

H3 to H4 Connection 110 120 130 140 150 ASDSNVYDLLKDLEEGIQTLMGRLEDGSPRTGQIFKQTYSKFDTNSHND

t t

t

t

160 170 180 190 DALLKNYGLLYCFRKDMDKVETFLRIVQCRS*VEGSCGF _______-__S__K__LH-A--Y--VMK-_RF--RF--S--A----RL-------------------------*------_------------K---H-------V-K---*----------------------------------A--*_-----_____________________---------RF--S--A-

space for alignment mutants not previously

published

Fig. 1. Comparison of the amino acid sequences of hGH, mGH and hGH mutants. The nomenclature for the segment substituted variants gives the homolog the segment is derived from followed by the extreme most amino and carboxyl terminal residues in the segment. For example, in pGH 11-33 a chimeric protein has been produced by replacing residues 11 to 33 of hGH with the analogous sequence from pGH. For point mutants, the wild-type residue is given in single letter code followed by its position in hGH and the mutant residue (amino acid single letter code: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gin; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; Y, Tyr).

Immunodominant

structures

of human

Region of ’ secondary structure

hGH Mutant

Helix 1

pGH 11-33 hPL 12-25 hPRL 12.-19 hPRL 22.-23

Loop 4652

Loop

54-14

1. Binding

mAb 8* Determined equivalency % + SE 81 99 85 91

of polyclonal

hormone

1083

Binding of mouse serum anti-hGH and wild-type hGH in ELISA

The concn of each mutant was adjusted until the mutant 405 nm readings were closely matched to the 405 nm readings produced by 80 ng/ml of wild-type hGH. As a control for molar equivalency in ELISA, the signal obtained for each mutant, when detected with mAb 8, is expressed as a percentage of the signal obtained for wild-type hGH (Table 1). mAb 8 has already been characterized. When titered against wildtype hGH or any of the mutants shown in Fig. 1, the same molar concn was required to produce half-maximal binding (Cunningham et al., 1989). When titered against a naturally occurring deletion variant of hGH (residues 3246 missing), however, more than a thousand fold loss in binding was observed. For the 19 mutants tested, the average equivalency was 94%, with a range from 78% to 107%. The SE given for each mutant (Table 1) was calculated from three separate experiments, and each experiment was done using triplicate wells on the ELISA plate.

Table

growth

antibodies

to mutant

To determine mouse anti-hGH sera binding to hGH mutants, plates were precoated with rabbit antibodies and incubated with mutant and wild-type hGH as described above, and then reacted for 2 hr with a limiting dilution of mouse anti-hGH serum in 2% PM. Mouse antibodies were detected with AP-labeled goat antimouse IgG as described above. Binding to each mutant is expressed as a percent reduction of the binding obtained for wild-type hGH. In a preliminary experiment used to optimize the assay conditions, serum from one mouse was tested at dilutions which gave three-fourths, one-half, or one-fourth of maximal binding to wild-type hGH (dilutions of 1/loo, l/400 and l/ 1600, respectively). Although all three dilutions yielded the same relative binding profile, in terms of which mutants disrupted binding, the greatest

mouse

anti-hGH

IgG to hGH mutants

Mouse anti-hGH IgG binding to mutant hGH’ (expressed as a % reduction in binding to wild-type) MS 261 MS 262 MS 263 MS 264 MS 265 serum serum serum serum serum

MS 261-265 Average % reduction *SD

Percentd incidence (N=5)

rf: 12 k 23 &-3 +_ 18

20 21 6 20

82 16 20 58

74 35 0 57

71 25 34 40

46 66 71 40

hPL 4652 pGH 48-52

91+ 12 105 f 13

0 0

0 0

0 0

0 0

3 0

hPRL 54-74 pGH 57-73

105f9 97 + 6

27 56

9 52

6 49

0 67

2 31

9f 11 51 f 13

20 100

59 f 25 33+20 26 +- 28 43116 1+1 o+o

80 60 60 80 0 0

L80D hPRL

88-95

78 + 5 97+ 13

11 9

15 6

12 7

31 0

1 0

14* 11 4*4

20 0

Loop 97-104

hPRL

97-104

104 + 16

18

17

8

4

8

12f8

40

Helix 3

hPL 109-l 12

11

32

29

31

15

48

H3 to H4 Connection

hPRL hPRL hPRL

126136 137-145 146152

94 + 6 105 + 17 84 + 4

10 1 8

12 9 9

9 8 2

0 14 17

7 8 9

Helix 4

hPRL 158-159 pGH 167.-182 C182A

86 f 3 107 + 15 96 + 3

19 53 33

9 61 27

1 63 12

16 35 49

7 31 23

lOf7 49+ 15 29 + 14

20 100 100

3

30

26

49

11

24& 18

60

Helix 2

C-Terminus Segment

pGH

184191

84k

98+

14

31*

12

8*4 8*5 9f5

80 0 0 0

“The hGH mutants (defined in Fig. 1) are grouped according to general regions of easily identifiable secondary structure (Fig. 2). ‘Detection by mAB 8 was used to confirm that equimolar quantities of mutant and wild-type hGH had been captured by rabbit polyclonal antibodies precoated onto ELISA plates. The signal for each mutant has been expressed as a percentage of the signal for wild-type hGH. ‘For each of five mouse (MS) anti-hGH sera, the ability to bind mutant and wild-type hGH was compared. The signal obtained for binding to each mutant is expressed as a percent reduction in the signal obtained for binding to wild-type hGH. SD (not shown), based on N = 3 within the same experiment, ranged from 0% to 12%, with an average of approximately 3%. dFor each of the five sera tested, a reduction in binding to a particular mutant was considered as a positive incidence if the percent reduction was greater than the SE for the mAb 8 determined equivalency (where less than 100% equivalency was obtained, the difference was added to the SE).

JOSEPH R. WEBERet al.

1084

relative reductions were obtained at the largest dilution tested (data not shown). Since the use of a limiting dilution appears to be an important parameter in this type of assay, each of the five mouse sera tested was used at a dilution, in the range of l/400 to l/3200, which had been determined by titration against wildtype hGH to give approximately one quarter of maximal binding. A reduction in binding to a particular mutant, as compared to wild-type hGH, by polyclonal antibodies was considered significant if the reduction in binding was greater than the SE for mAb 8 equivalency. For example, hPRL 22-33 has an equivalency of 91% + 18, and therefore a reduction in binding would have to be greater than 27% to be rated as a positive incidence (a SE of 18 plus the difference between 91 and 100).

-02-I . . . . / 1.0

..I...

1.5

-

2.0

.

2.5

..I....

*

3 0 - Log

35 ii

.

.

.

.

.

m

4.0

4.5

5.0

ter

1.4

B

1 I2 1

o-

I

o-

Solution phase inhibitions

ELISA plates were precoated for 16-18 hr at 4°C with 2 ,ug/ml of wild-type hGH in BBS, and then quenched for at least 1 hr with 2% PM. The dilution of mouse anti-hGH serum previously determined to give approximately half-maximal binding under the conditions of this assay was incubated for 2 hr either by itself or with serial difutions from 320 to 0,005 pg/ml of either mutant or wild-type hGH, and then added to the plates for 2 hr. Antibody bound to the hGH precoat was detected with AP-labeled goat anti-mouse IgG diluted l/l000 in PBS followed by pNPP, as described above. Volumes, washes and incubation temps were also identical to those already described for normalization of mutant and wild-type hGH. Inhibition was calculated as the percent reduction in absorbance at 405 nm of anti-hGH serum mixed with inhibitor as compared to anti-hGH serum alone. Non-scion binding to the pfaie was controlled by using a prebleed from the same mouse. RESULTS

Antibody

binding projles

The relative binding of polyclonal anti-hGH IgG, in sera from BALB/c mice, to mutant and wild-type hGH is summarized in Table 1. Although the values for reduced binding to mutant hGH are based on testing at a single diiution of serum, a complete titration of mouse serum against mutant and wild-type hGH was done for a limited number of the mutants (Fig. 2), and the results indicated that a 50% reduction in binding in Table 1 correlates with a two- to four-fold lower final titer. In the structural model of hGH shown in Fig. 3, the approximate location of amino acid substitutions resulting in a significant loss of binding by at least three of the five mouse sera have been indicated. Described in order from the amino to the carboxyl terminus, the following general binding patterns were observed: On average, the polyclonal mouse anti-hGH IgG bound considerably less well to all four of the helix 1 mutants, as compared to wild-type hGH. A comparison of mutants hPRL 22--33 and hPRL 12-19 suggests that mutations in the carboxyl half of helix 1 have a greater influence on

c

G a

0,s 0.6 04” a200

-02-1w IO

. . . . e”‘.,.“., 1.5

2.0

25

30

35

4.0

4.5

so

- Log Titer

Fig. 2. Titration of mouse 262 anti-hGH serum against mutant and wild-type hGH captured and presented by rabbit polyclonal antibodies. In all 3 panels, titration af mouse 262 anti-hGH serum against wild-type hGH ( n ) gave an end-point titer af 4.1 (where end-point titer is taken as the negative log of the last dilution to give a signal of at least double the background). Panel A. Titration against hPRL 22-33 (0) gave an end point titer of 3.5. Panel B. Titration against pGH 57-73 (a) gave an end point titer of 3.8. Panel C. Titration against hPRL W-104 (a,) gave an end point titer of 4.1. (Titration of pre-immunization serum is shown with hollow symbols.)

binding by polyclonal antibodies than do mutations on the amino half of helix 1. The twa mutants located in loop region 46-52 were tested, and in both cases no detectable loss in binding was observed. However, when the two mutants in loop region 54-74 were tested, one of them, pGH 57-73, displayed an average reduction in binding of 5 1% for the five mouse sera tested, while the other, hPRL 54-74, gave a positive incidence of reduced binding for only I of the sera. Mutants in the next two regions, helix 2 and

Immunodominant

structures of human growth hormone

1085

Helm : V A

RKE hPRL 12-19 hPRL 22-33

Loop 54-74 0 pGH 57-73 j&&Q x hPL 109-112

k&&&g D 0

pGH 167-182 C182A

C-terminus seamen1 n pGH 184-191

Fig. 3. Structural model of hGH. The folding pattern and overall conformation (Cunningham et al., 1989) are based on the crystal structure of pGH (Abdel-Meguid et al., 1987). The long barrel shaped structures represent alpha helices and disulfide bonds connect residues 53-l 65 and 182-189. Common symbols are used to identify the location of amino acid substitutions which result in a dramatic reduction in binding by mouse polyclonal anti-hGH IgG, and substituted residues which are hidden in this view have been indicated by cross-hatching. The location of residues within a helix were fixed by interpolating between the first and last residue of the helix, and the position of other residues along the polypeptide chain were estimated by comparison to the structure of pGH. A view of the opposite side of the molecule can be seen in Cunningham et al.

loop 97-104, showed only minor losses in binding, but in the following region, helix 3, the mutant hPL 109-l 12 gave a large average reduction in binding (31%). Based on the X-ray crystallography data for pGH, the long connecting sequence between helices 3 and 4 has been described as having no well-defined secondary structure (Abdel-Meguid et al., 1987), so the three mutants with substitutions in this region, hPRL 126136, hPRL 137-145 and hPRL 146152, have been grouped as simply the H3 to H4 connection in Table 1 (it should be noted, however, that hPRL 126136 actually includes 3 substitutions at the very end of helix 3). The substitutions used to make these three mutants did not display any measurable ability to disrupt the polyclonal antibody binding. Near the beginning of helix 4, hPRL 158-l 59 demonstrated only very minor ability to disrupt antibody binding. In marked contrast, pGH 167-182, which has four substitutions spanning from the middle to near the carboxyl terminus of helix 4, caused a dramatic loss of antibody binding for all 5 sera (with an average reduction of close to 50%). A 100% incidence of reduced binding was also seen for C182A, where a disulfide bond near the end of helix 4 was disrupted by substituting an alanine for one of the cysteines. Substitutions in the short carboxyl terminus segment of the molecule (pGH 184-191) were effective in disrupting polyclonal IgG binding in 60% of the mouse sera. Solution phase inhibitions As an alternate mouse polyclonal

method of evaluating the binding of anti-hGH IgG to hGH mutants, sev-

era1 of the mutants were tested for their ability to inhibit a mouse anti-hGH serum (mouse 265 of Table 1) from binding to hGH precoated onto ELISA plates. The advantage of this method is that it eliminates any potential bias in the orientation at which the mutants are presented, because they are free in solution when mixed with the mouse serum. The disadvantage of this method is that it requires almost a thousand fold more protein, since pgg/ml concns of inhibitor are required rather than the ng/ml concns used in the antigen capture and presentation assay employing polyclonal rabbit antibodies. Owing to limited quantities of purified protein, only five of the mutants were tested by the solution phase inhibition method. The prediction was that mutants hPRL 22-33, pGH 57-73 and pGH 167-182 would fall far short of 100% inhibition of binding to the hGH fixed onto the plates, since serum from mouse 265 had already shown a dramatic reduction in binding to these mutants as compared to wild-type hGH. By similar reasoning, mutants hPRL 97-104 and hPRL 137-145 in solution should inhibit the serum anti-hGH almost as well as wild-type hGH in solution. Representative inhibition curves are shown in Fig. 4, and a summary of results is given in Table 2. To facilitate comparison, the relevant data from Table 1, with the addition of SE based on four separate experiments, has been added to Table 2. The value actually reported for each solution phase inhibition is the percent of hGH binding that the mutant failed to inhibit, since this is the value most directly comparable to the percent reduction in binding reported in Table 1. The generally close

JOSEPH R. WEBER et al.

1086

I3

A 120-

1201 IO

I IO-

100

loo-

&ii::_::::1

jy/..-;-li

0 ,-..I 100

IO’

. . . . ..-I I02

-“I 105 ng/ml

Fig. 4. Inhibition of pGH 167-182 (0) anti-hGH IgG from hPRL 97-104 (0)

anti-hGH

. . ..-.I . . . . ..“I I04 I05

. . . .ml I06

anti-hGH polyclonal IgG or wild-type hGH (H), binding to wild-type hGH or wild-type hGH (m),

by mutant or wild-type in solution, were used precoated onto ELISA in solution, were used

hPRL 22-33 pGH 57-73 hPRL 97-104 hPRL 1377145 pGH 167-182

of two methods used to evaluate binding IgG to hGH mutants

% Reduction” average + SE 40 * 7 31 f2 8+7 8&2 3115

.-1-1

. . ..-1 103

I04

. ..- ‘

. . . ..I

I05

106

lnhlbltor

hGH. Panel A. Either mutant to inhibit mouse 265 serum plates. Panel B. Either mutant

to inhibit mouse 265 serum IgG from binding to wild-type hGH precoated onto ELISA plates.

The antigenic structure of hGH has been probed extensively in studies which used mAb competition and cross-reactivity to related molecules to define epitopes. Early studies indicated that there were at least four distinct antigenic sites (Ivanyi, 1982; Retegui et al., 1982) or four major antigenic regions (Surowy et al., 1984). Later research using additional mAb expanded upon the four antigenic site model for hGH and presented evidence that the epitopes corresponding to twelve mAb occupy a large percentage of the surface

hGH Mutant

. . . ...1 I02

ng/ml

DISCUSSION

of mouse 265 anti-hGH

. . ..“.I IO’

Inhlbltor

agreement between the two different methods suggests that any bias in orientation caused by immobilizing hGH with polyclonal rabbit antibodies is not substantial.

Table 2. Comparison

0 100

A% Inhibition” average + SE 39-t 17 24+ 8 11+4 1+1 44 + 6

‘Mutant or wild-type hGH was captured and presented by rabbit polyclonal antibodies fixed to ELISA plates. The signal obtained by binding of mouse 265 anti-hGH IgG to each mutant thus presented is expressed as a percent reduction relative to wild-type hGH. SE are based on four separate experiments. bMutant or wild-type hGH in solution was used to inhibit mouse 265 anti-hGH IgG from binding to wild-type hGH fixed to ELISA plates. The A% inhibition was calculated by subtracting the maximum inhibition by mutant from the maximum inhibition by wild-type hGH. Wild-type hGH in solution inhibited approximately 100% (99% f 1 SE) of binding to the wild-type hGH fixed to the plates. SE are based on four separate experiments.

area of hGH (Vita et al., 1986). More recently, an epitope model for hGH has been proposed consisting of at least ten antigenic sites which could be grouped into five antigenic regions (Strasburger et al., 1989). The precise locations of the antigenic structures were not elucidated. In order to identify immunodominant structures in hGH we have characterized the serum polyclonal IgG antibody response produced in BALB/c mice challenged with hGH. These sera were analyzed for their ability to bind to a panel of 19 hGH mutants which collectively alter 90 of the 191 amino acid residues of wild-type hGH. With the exception of 2 single residue-substitutions, these mutants were produced by exchanging short segments of primary sequence (2-23 residues in length, with 2-15 non-homologous residues) from the homologous hormones hPL, hPRL and pGH. Nine short regions of sequence substitutions were identified as mutations which resulted in a dramatic reduction in binding, as compared to wild-type hGH, by polyclonal IgG. In terms of primary and secondary structure designations, these disruptive substitutions can be localized to five immunodominant regions: the length of helix 1 (pGH 11-33, hPL 12-25, hPRL 12-19 and hPRL 22-33) loop 5474 (pGH 57773), the beginning of helix 3 (hPL 109-l 12), the carboxyl half of helix 4 (pGH 167-182 and Cl 82A) and the short carboxyl terminus segment of the molecule (pGH 184-191). Based on the characterization of the hGH mutants (mAb binding, receptor binding, expression levels in E. coli and far ultraviolet circular dichroic spectra) it is probable that the substituted residues introduce only localized alterations in the three-dimensional structure of hGH, and do not disrupt the overall folding of the molecule (Cunningham et al., 1989). Still, a loss of binding does not prove that one or more of the substituted amino acids are necessarily antibody contact residues, but rather only suggests that they are likely to be spatially close to the most dominant antibody binding sites.

Immunodominant

structures

Interestingly, the immunodominant regions of hGH include the same three discontinuous segments of hGH recognized by the liver receptor: the loop between residues 54-74, the central portion of helix 4 to the carboxyl terminus of the molecule, and part of the amino terminus region of helix 1. The immunodominant structures are not, however, restricted to this patch responsible for receptor binding. Substitutions in the carboxyl half of helix 1 (pGH 22-33) and near the amino terminus of helix 3 (hPL 109-l 12) significantly disrupted binding by polyclonal anti-hGH IgG in four of five sera tested, yet neither of these mutants affect binding to hGH receptor (Cunningham et al., 1989). A comparison between the mouse and human GH sequences (Fig. l), which share 65% sequence homology (Linzer and Talamantes, 1985), failed to identify any clear relationship between the dominant antibody binding regions and non-conserved regions of primary sequence. The non-homologous residues are widely distributed throughout the primary sequence. Although all five of the immunodominant regions we have identified did have several residue differences between mouse and human, regions which did not appear to be important for polyclonal IgG binding also contained numerous residue differences. We have examined the specific relationship between binding activity of mAb isolated against hGH and the murine polyclonal antibodies. Cunningham et al. (1989) reported that two anti-hGH antibodies, mAb 1 and 2, failed to bind to hPRL 97-104. In contrast, the serum binding data reported in Table 1 show that hPRL 97-104 appears to have only minor ability to disrupt binding by some of the mouse sera. Thus, the two mAb likely react with only a minor antigenic site. Another notable difference between the mAb epitope mapping data and the polyclonal antibody mapping is that mutants hPRL 22-33 and pGH 167-182 had a profound influence on binding by mouse polyclonal IgG, but did not disrupt binding to any of the panel of 8 mAb previously tested. Screening a larger number of mAb should produce specificities which could be mapped by these two mutants. Structurally, antigenic determinants of proteins have been categorized as being either short peptide fragments comprising continuous (linear) epitopes or conformational (discontinuous) epitopes comprised of polypeptide chains topographically adjacent in space (Benjamin et al., 1984). It has been suggested that all protein epitopes are discontinuous to some extent, and the importance of topographical surface structures has been supported by X-ray crystallographic results suggesting that even continuous epitopes detected by peptide reactivities with antibodies are only part of larger discontinuous epitopes (Barlow et al., 1986). As illustrated in Fig. 3, there are some altered residues which, although distant in the primary sequence, are brought together into such close physical proximity that they may be influencing the same or overlapping epitopes. The importance of these spatial considerations has already been demonstrated by mAb epitope map-

of human

growth

hormone

1087

ping (Cunningham ef al., 1989) where, for seven of eight mAb tested, binding was disrupted by mutations far apart in the primary sequence, but in close proximity on the tertiary structural model for hGH. The proposed epitopes for two antibodies, mAb 3 and 4, involved spatially close segments of helix 1 and 3, and the proposed epitopes for two other antibodies, mAb 5 and 6, involved spatially close segments of loop 54-74 and helix 4 (or helix 4 plus carboxyl terminus). These same regions also contribute as immunodominant structures affecting the binding of serum anti-hGH antibodies. Characterization of the polyclonal serum antibody reactivity with homolog-scanning mutants could be a valuable tool in determining whether second generation versions of a recombinant protein (with alterations introduced to either enhance or remove a receptor binding activity) have significantly altered antigenic profiles. Moreover, animal models are frequently used to evaluate the biological activity of human proteins, and mAb have been identified that inhibit, and in some cases appear to even enhance the biological activity of human protein hormones (Holder et al., 1985). The methodology we have described should be useful in the characterization of human antibodies directed against pharmaceutically important recombinant proteins. Acknowledgements-We are grateful to Bill Lagrimas for mouse handling expertise, to Wayne Anstine for the hGH drawing, and to Brian Fendly for generously providing mAb.

REFERENCES Abdel-Meguid S. S., Shieh H.-S., Smith W. W., Dayringer B. N., Violand B. N. and Bentle L. .4. (1987) Threedimensional structure of a genetically engineered variant of porcine growth hormone. Proc. natnl. Acad. Sci. U.S.A. 84, 6434. Barlow D. J., Edwards M. S. and Thornton J. M. (1986) Continuous and discontinuous protein antigenic determinants. Nature 322, 747. Benjamin D. C., Berzofsky J. A., East I. J., Gurd F. R. N., Hannum C., Leach S. J., Margoliash E., Michael J. G., Miller A., Prager E. M., Reichlin M., Sercarz E. E., Smith-Gill S. J., Todd P. E. and Wilson A. C. (1984) The antigenic structure of proteins: a reappraisal. A. Rev. Immun. 2, 67. Cooper H. M., Jemmerson R., Hunt D. F., Griffin P. R., Yates III J. R., Shabanowitz J., Zhu N.-Z, and Paterson Y. (1987) Site-directed chemical modification of horse cytochrome c results in changes in antigenicity due to local and long-range conformational perturbations. J. biol. Chem. 262, 1159 1. Cunningham B. C., Jhurani P., Ng P and Wells J. A. (1989) Receptor and antibody epitopes in human growth hormone identified by homolog-scanning mutaganesis. Science 243, 1330. Dowbenko D., Nakamura G., Fennie C., Shimasaki C., Riddle L., Harris R., Gregory T. and Lasky L. (1988) Epitope mapping of the human immunodeficiency virus type 1 gpl20 with monoclonal antibodies. J. Virol. 62, 4703. Holder A. T., Aston R., Preece M. A. and Ivanyi J. (1985) Monoclonal antibody-mediated enhancement of growth hormone activity in vivo. J. Endocr. 107, R9.

1088

JOSEPH R. WEBER et al.

Ivanyi J. (1982) Analysis of monoclonal antibodies to human growth hormone and related proteins. In Monoclonal Hybridoma Antibodies: Techniques and Applications (Edited by J. G. R. Hurrell), p. 59. CRC Press, Cleveland. Linzer D. I. H. and Talamantes F. (1985) Nucleotide sequence of mouse prolactin and growth hormone mRNAs and expression of these mRNAs during pregnancy. J. biol. Chem. 260, 9574.

Reichlin M. (1975) Amino acid substitution and the antigenicity of globular proteins. Adv. Zmmun. 20, 71. Retegui L. A., Milne R. W., Cambiaso C. L. and Masson P. L. (1982) The recognition by monoclonal antibodies of various portions of a major antigenic site of human growth hormone. Molec. Immun. 19, 865. St Clair E. W., Pisetsky D. S., Reich C. F. and Keene J. D. (1988) Analysis of autoantibody binding to different regions of the human La antigen expressed in recombinant fusion proteins. J. Zmmun. 141, 4173. Strasburger C. J., Kostyo J., Vogel T., Barnard G. J. and

Kohen F. (1989) The antigenic epitopes of human growth hormone as mapped by monoclonal antibodies. Endocrinology 124, 1548. Surowy T. K., Bartholomew R. M. and VanderLann W. P. (1984) Antigenic sites of human growth hormone and related molecules detected by monoclonal antibodies to human growth hormone. Molec. Zmmun. 21, 345. Urbanski G. J. and Margoliash E. (1977a) The antigenicity of cytochrome c. In Immunochemistry of Enzymes and their Antibodies (Edited by M. R. J. Salton), p. 203. Wiley, New York. Urbanski G. J. and Margoliash E. (19776) Topographic determinants on cytochrome c. I. The complete antigenic structures of rabbit, mouse and guanaco cytochrome c in rabbits and mice. J. Zmmun. 118, 1170. Vita N., Etcheverrigaray M., Biscayart P. L. and Retegui L. A. (1986) Relative distribution of various antigenic determinants on the human growth hormone surface. Molec. Immun. 23, 619.

Immunodominant structures of human growth hormone identified by homolog-scanning mutagenesis.

Homolog-scanning mutagenesis has been reported to be useful in elucidating the antigenic epitopes recognized by monoclonal antibodies and hGH binding ...
1MB Sizes 0 Downloads 0 Views