Epitope mapping of novel monoclonal antibodies to human angiotensin I‐converting enzyme

Abstract

Angiotensin I‐converting enzyme (ACE, CD143) plays a crucial role in blood pressure regulation, vascular remodeling, and immunity. A wide spectrum of mAbs to different epitopes on the N and C domains of human ACE have been generated and used to study different aspects of ACE biology, including establishing a novel approach–conformational fingerprinting. Here we characterized a novel set of 14 mAbs, developed against human seminal fluid ACE. The epitopes for these novel mAbs were defined using recombinant ACE constructs with truncated N and C domains, species cross‐reactivity, ACE mutagenesis, and competition with the previously mapped anti‐ACE mAbs. Nine mAbs recognized regions on the N domain, and 5 mAbs–on the C domain of ACE. The epitopes for most of these novel mAbs partially overlap with epitopes mapped onto ACE by the previously generated mAbs, whereas mAb 8H1 recognized yet unmapped region on the C domain where three ACE mutations associated with Alzheimer's disease are localized and is a marker for ACE mutation T877M. mAb 2H4 could be considered as a specific marker for ACE in dendritic cells. This novel set of mAbs can identify even subtle changes in human ACE conformation caused by tissue‐specific glycosylation of ACE or mutations, and can detect human somatic and testicular ACE in biological fluids and tissues. Furthermore, the high reactivity of these novel mAbs provides an opportunity to study changes in the pattern of ACE expression or glycosylation in different tissues, cells, and diseases, such as sarcoidosis and Alzheimer's disease.

1. INTRODUCTION

Angiotensin I‐converting enzyme (ACE, CD143) is zinc metallopeptidase with broad substrate specificity ¹ which plays an important role in diverse processes such as blood pressure regulation, the development of vascular pathology and remodeling, and immunity. ² , ³ , ⁴ ACE is constitutively expressed on the surface of endothelial cells, absorptive epithelial and neuroepithelial cells, and cells of the immune system, such as activated macrophages and dendritic cells. ² ACE has also been defined as a marker of hematopoietic stem cells. ⁵ Two ACE isoforms are expressed namely somatic ACE, which contains homologous N‐ and C‐terminal domains, ⁶ and the testis‐specific form, denoted testicular ACE (tACE). The latter is solely expressed in germ cells during spermatogenesis and comprised of only the C‐terminal domain with a unique 36‐residue sequence at its N‐terminus. ⁷ Disruption of the ACE gene in mice resulted in a dramatic decrease in male fertility and kidney function. ³ , ⁸ , ⁹ ACE was assigned as a CD marker, designated CD143. ¹⁰ , ¹¹

Besides its well‐known role in blood pressure regulation, local changes in ACE expression are also the hallmark of a wide range of diseases, summarized in Reference 12, including sarcoidosis, tissue repair after myocardial infarction and in human atherosclerotic plaques, in the kidneys and lungs of patients with interstitial fibrosis, diabetes mellitus, acute pyelonephritis, different stages of acute renal failure, and primary pulmonary hypertension.

We previously generated and characterized a panel of 16 mAbs against conformational epitopes on the both the N and C domains of ACE. This set of mAbs have been extensively mapped, which allowed us to establish the novel approach for ACE characterization, denoted conformational fingerprinting, ¹³ , ¹⁴ , ¹⁵ , ¹⁶ to facilitate the study of various aspects of ACE structure and biology as well as ACE‐related pathologies, such as sarcoidosis, ¹³ uremia, ¹⁴ Gaucher's disease, ¹⁷ prostate and lung cancers. ¹⁸ , ¹⁹ Moreover, epitope mapping of these mAbs helped to identify several important functional mutations of ACE‐reviewed in Reference 20. The structure–function relationship of ACE, which has been refined using ACE conformational fingerprinting, has helped to identify regions involved in specific functions. ²¹ For example, mAb 3A5 induces shedding and has an anti‐catalytic effect, while mAb 1B3 allows for the detection of an ACE mutation in the juxtamembrane stalk region, and mAbs 6A12/1G12 can be used to detect ACE inhibitors in patient blood samples. ¹⁶ The study on the effect of mAbs on ACE ectodomain shedding or dimerization in reverse micelles have revealed a region in the N domain that suppresses ACE shedding, and may be involved in dimerization. ²²

The characterization of mAb epitopes to ACE at the molecular level is clearly an invaluable tool for investigating ACE structure and function. In this study, we generated 14 novel mAbs against human somatic ACE, purified from seminal fluid (i.e., originating from the epithelial cells of the human prostate and epididymis ¹⁵ ). We characterized and mapped these mAbs to epitopes on the N and C domains of human somatic ACE. Some of the epitopes for these novel mAbs partially overlap with epitopes previously defined, and have discernable functional activities, therefore widening the repertoire of mAbs available for studying different aspects of ACE biology.

2. RESULTS AND DISCUSSION

We previously demonstrated that the pattern of mAbs binding to the conformational epitopes on the surface of ACE is an extremely sensitive marker of local conformational changes in ACE. Denoted “conformational fingerprinting”, this method can be used to detect conformational changes due to mutations, denaturation, inhibitor binding, and the presence of the membrane anchors. Previously, we demonstrated that the conformation of ACE is cell‐ and tissue‐specific, and likely caused by differences in post‐translational modifications, including different glycosylation, by studying ACE from endothelial cells, macrophages, and dendritic cells of sarcoid granulomas. ACE contains 17 potential N‐linked glycosylation sites of which 15 can be glycosylated, most likely with complex‐type, fucosylated biantennary glycans. ²³ The exact glycan structure, however, is cell‐ and tissue‐specific. The ACE fingerprint could thus potentially identify the cells from which ACE originates. ¹³ , ¹⁶ We also previously proved that the conformation is tissue‐specific by comparing the conformational fingerprint of lung and seminal fluid ACEs using a set of 17 mAbs to epitopes of human ACE. Patterns of binding to ACE isolated from the lungs and seminal fluid were dramatically different, ¹⁵ reflecting differences in the local conformations of these ACE isoforms, likely due to different patterns of ACE glycosylation in the lung endothelial cells and in the epithelial cells of the epididymis/prostate. The differences in local conformations of ACE could be the basis for the generation of mAbs that are capable of distinguishing tissue‐specific ACE isoforms. Human seminal fluid contains 50‐fold more ACE than blood, and it likely originates from the glandular epithelial cells of the epididymis and prostate, which express significant amounts of somatic ACE. ¹⁵ Therefore, in the present study, human lung homogenate was used as a source of ACE from lung endothelial cells and seminal fluid was used as a source of prostate and epididymal ACE from epithelial cells.

2.1. Immunization and screening of hybridoma clones

To generate a new set of mAbs that are specific to the human somatic prostate ACE, we immunized mice with human somatic ACE purified from seminal fluid according to a previously published method. ¹⁵ To obtain mAbs that specifically recognize prostate ACE, we performed double screening of the hybridoma clones using a plate precipitation assay and ELISA. When catalytically active human ACE is adsorbed to the plastic surface of a 96‐well microtiter plate, it undergoes partial denaturation. ²⁴ Therefore, an ELISA assay will identify the hybridoma clones producing mAbs to both conformational and sequential epitopes of ACE. The plate precipitation assay, on the other hand, can identify hybridoma clones producing mAbs that bind only to catalytically active ACE. ²⁵ Unlike ELISA, the plate precipitation assay does not interfere with heterophilic mAbs which could be detected by ELISA. ²⁶

In the first hybridization, out of 670 post‐fusion primary cell populations growing in 96‐well plates, the presence of antibodies to seminal fluid ACE was found in 91 wells. Of these, four stable populations producing mAbs to ACE were further expanded and stable clones of population 2H4 were chosen for further characterization. In the second and third hybridizations, out of the 500 primary cell populations reactive to ACE from seminal fluid ACE (and ACE from human lungs) more than 20 stable cell populations were chosen for further investigation. Of these, six stable populations were successfully cloned, while seven others existed as stable populations without cloning.

2.2. Initial characterization of novel mAbs to human somatic ACE

Characterization and epitope mapping of 14 novel mAbs (IgG1 isotype with a κ chain) that bind to ACE from seminal fluid (ACE SF), and from human lung homogenate, was carried out. We initially compared binding of these 14 mAbs to human somatic two‐domain ACE purified from seminal fluid (Figure 1). During screening of culture medium from primary hybridoma cell populations, we also precipitated the truncated N and C domains of ACE, which allowed us to present an ability to bind ACE separately for mAbs to the N and C domains (Figure 1a). The ability of the novel mAbs to bind ACE was compared with that of mAb 9B9, the strongest mAb from our previous set of mAbs to human ACE, ²¹ and with that of several other mAbs ‐ i2H5, ²⁹ 1G12, ³⁰ and 5F1, ³¹ and the C domain 3F10 (directed to the immuno‐dominant epitope on the C domain), and 1E10 which is anti‐catalytic for the C domain active center. ³²

FIGURE 1.

ACE activity precipitation by mAbs to ACE. (a) Purified ACE from human seminal fluid (ACE SF), adjusted to a final enzymatic activity of 5–10 mU/ml using ZPHL as a substrate, was used as a source of catalytically active ACE and incubated in a microtiter plate coated with mAbs via a goat‐anti‐mouse IgG bridge. Precipitated ACE activity was quantified by spectrofluorometric assay with ZPHL. (b) Purified ACE from human lung homogenate and ACE SF were precipitated with mAbs as in A, and presented as ACE SF/ACE lung binding ratio. Lung and SF ACE activities were equilibrated using ZPHL as a substrate prior to precipitation. Precipitation of ACE activity by mAb 9B9 in this particular graph was normalized by ACE activity in both preparations –in order to validate further normalization of ACE precipitation by other mAbs via normalization with mAb 9B9. (c) Pooled human serum from healthy blood donors (Interstate Blood Bank, Inc, Memphis, TN) diluted 1/5 with PBS and equilibrated to ACE activity in ACE SF (with ZPHL as a substrate) was precipitated by mAbs as in (a), and presented as the human serum ACE/ACE SF ratio. (d) ACE activity precipitation by tested mAbs was quantified (as in A) in the culture medium of CHO cells expressing human recombinant truncated N‐ and C‐domain constructs as well as soluble somatic ACE without the transmembrane anchor, denoted WTΔ, ²⁷ , ²⁸ equilibrated to ACE activity (with ZPHL as a substrate) and presented as N‐ or C‐domain/somatic ACE ratio. Data are presented as the mean percentage of ACE activity precipitated by the indicated mAbs, determined in three independent experiments (which did not differ by more than 10%) relative to that precipitated by mAb 9B9. ²¹ , ²⁸ mAb 9B9 is indicated by a green bar, orange bars indicate an increase of ACE precipitation of more than 20% over 100%, brown bars indicate an increase of more than 50%, red bars indicate an increase of more than two‐fold, yellow bars indicate a decrease in ACE precipitation of more than 20% over 100%, blue bars indicate a decrease of more than 50%

Monoclonal antibody 5G8 (directed to an epitope on the C domain) precipitated purified seminal fluid ACE even more effectively than our strongest published mAb, 9B9 (Figure 1a). The binding of the novel set of 14 mAbs to human lung ACE is presented in Figure 1b as the ACE SF/ACE lung ratio. mAb 5G8 demonstrated only modest preferential binding to seminal fluid ACE in comparison to lung ACE which is similar to the binding of mAb 3F10 from a previous mAb set. ¹⁵ In contrast, mAbs with epitopes on the N domain (6H6 and 2D1) showed preference for ACE purified from lung, similar to what was found previously for mAb 5F1. ³¹ These results likely confirm that there are differences in glycosylation of ACE isolated from the seminal fluid and lung. ¹⁵ Moreover, this suggests that most of these 14 mAbs originated from different clones.

The ratio of ACE activity precipitation from human serum to that from seminal fluid is shown in Figure 1c. Of the 14 novel mAbs tested, six mAbs to the N domain and one mAb to the C domain (3C10), demonstrated a decreased ability to precipitate ACE from human serum. There are two possible reasons to explain this decrease: (1) the six mAbs to the N domain may have epitopes overlapping with mAbs 1G12/6A12 which have dramatically decreased binding to serum/plasma ACE due to competition with bilirubin for binding to a region on the N domain close to R532 ³³ ; (2) these seven mAbs may be very sensitive to the presence of sialic acid on the glycans of the corresponding glycosylation sites, which are present in their epitopes, as was demonstrated for mAb 5F1 on the N domain ³¹ and mAb 1E10 on the C domain. ³² Plasma ACE is more sialylated than its “parent” lung enzyme because of the elimination of less sialylated proteins from the blood by liver lectins. ³⁴ The grouping of mAbs based on this characteristic again suggested that the 14 mAbs originated from different hybridoma clones.

We next measured the precipitation of recombinant human somatic ACE, two domain ACE, and truncated N and C domains, all expressed in CHO cells (Figure 1d) and showed that the pattern of ACE activity precipitation for WTΔ ACE (soluble ACE without transmembrane anchor ²⁷ ), also differed (for four mAbs) from the pattern observed for lung ACE (Figure S1A), confirming that glycosylation of ACE (as well as the pattern of mAb binding) is both cell‐ and tissue‐specific. ¹³ , ¹⁵ , ¹⁶ Paradoxically, it was found ³⁵ that ACE activity was increased in the cerebrospinal fluid of patients with Alzheimer's diseases (AD), which was confirmed by several studies of ACE activity in the AD brain, while ACE immunoreactive protein in the CSF of patients with AD was significantly decreased. In the Miners et al. study, ACE immunoreactive protein was quantified by R&D ACE assay, where a monoclonal antibody specific for human ACE had been pre‐coated onto a microplate and to precipitate ACE from CSF or brain homogenates. The characteristics of the particular mAb were not included in the publication; therefore, we can hypothesize that this paradoxical result was obtained because this particular mAb was sensitive to significant changes in local ACE glycosylation in AD‐associated neurons.

In the present study, precipitation of recombinant human ACEs (somatic, N and C domains) by each mAb was calculated as the ratio of ACE activity precipitation for separate N or C domains to that of two‐domain soluble somatic ACE (WTΔ) (Figure 1d). We chose somatic ACE expressed in the same host cells because the pattern of ACE glycosylation will be the same for two‐domain ACE and the separate truncated N and C domain constructs. This ratio increased significantly for mAbs 1B12, 2D7, 2D1 and 5F1 having epitopes on the N domain of ACE and for mAbs 1E10 (from the previous series ³² ) and 3C10 (in the present set), which have epitopes on the C domain. The increased single domain/somatic ACE binding ratio suggests that the C‐domain globule in two‐domain somatic ACE significantly masked the epitopes for mAbs 1B12 and 2D7 (and perhaps epitopes for 2D1 and 5F1) on the N domain, while the N‐domain globule dramatically shielded binding of the C‐domain mAb 3C10 in the present set (and the C‐domain mAb 1E10 in the previous series) (Figure 1d).

Interestingly, mAb 6C8 had a significantly decreased N domain/somatic ACE ratio, similar to that of mAb i2H5, which is anti‐catalytic towards the N domain active center. ²⁹ Therefore, we estimated ACE activity in soluble recombinant human ACE expressed in CHO cells (WTΔ ACE) with two substrates and calculated the ZPHL/HHL ratio for ACE precipitated by all tested mAbs (Figure S1B). This ratio allowed us to detect mAbs with anti‐catalytic activity. ²⁹ However, the effect of mAb binding on the ZPHL/HHL ratio (i.e., on the anticatalytic activity) is different for mAbs bound to ACE in solution and mAbs that are immobilized on a plate via goat‐anti‐mouse bridge. ¹⁴ , ³⁶ Because the ZPHL/HHL ratio was not significantly lower for mAb 6C8 (Figure S1B), it is possible that the observed significant decrease in N domain/somatic ACE binding ratio (Figure 1d) is not due to the anti‐catalytic activity of mAb 6C8 towards the N domain active center, but rather due to local conformational changes in the epitope for 6C8 on the N domain after assembling with C domain globule in somatic, two‐domain ACE. Similarly, the observed decrease in the C domain/somatic ACE ratio for mAbs 4C12 and 1A2 (Figure 1d) could be caused by local conformational changes in the epitopes for mAbs 4C12 and 1A2 on the C domain after assembling with N domain in somatic, two domain ACE. Furthermore, since mAbs 4C12 and 1A2 behaved identically in all tests, these two mAbs must have originated from the same hybridoma clone. Therefore, in all further tests we used only the mAb 4C12.

2.3. Epitope mapping of novel mAbs to the N domain of human ACE

To obtain further information on the localization of the epitopes of the novel mAbs to the N domain of ACE we examined the cross‐reactivity of eight novel N domain mAbs with ACE from chimpanzee serum and macaque rhesus (Macaca mulatta) lung homogenate, and compared it to that of the well‐characterized mAb 9B9 ²¹ and mAbs with overlapping epitopes (i2H5/1G12/6A12) ²⁹ , ³⁰ (Figure 2a,b). These eight novel mAbs demonstrated four different binding patterns: (1) 5B3 behaved similar to 9B9; (2) 6C8 and 6H6 were similar to 6A12; (3) 4H11 was similar to 4F4, and close to 2D7 and 1B12; (4) mAb 2D1 had a different pattern from the other tested antibodies (Figure 2a,b). We also tested the precipitation by these mAbs of ACE activity from three mutants of the truncated N domain, which were used previously for epitope mapping of mAbs i2H5, 1G12, and 6A12 ²⁹ , ³⁰ (Figure 2c–e). The epitopes of mAbs 6C8 and 6H6 overlap with that of mAb 6A12, but are not identical to it. Also, the epitopes of mAbs 6C8 and 6H6 are not identical to each other, because they have a three‐fold difference in affinity (Figure 1a), originated from different hybridizations, and the binding affinity of 6H6 to its epitope on the N domain was similar for the truncated N domain and two‐domain somatic ACE, while the binding affinity of mAb 6C8 was significantly lower for the single N domain ACE (Figure 1d).

FIGURE 2.

ACE activity precipitation by mAbs to the N domain of ACE. Chimpanzee serum (a) or macaque rhesus lung homogenate (b) were used as sources of catalytically active ACEs. The culture medium of CHO cells expressing the mutated forms of the truncated N domain ³⁰ was used as a source of N domain mutants (c–e). Human serum, lung homogenate, and culture medium of CHO cells expressing truncated N domain were used as controls. mAbs 9B9 ²¹ and i2H5/1G12/6A12 ³⁰ were used as controls for mapped mAbs to the N domain of ACE. These different ACE preparations (adjusted to a final enzymatic activity of 5–10 mU/ml using ZPHL as a substrate) were incubated in a microtiter plate coated with each of the tested mAbs and precipitated ACE activity was quantified as in Figure 1. Data are presented as the mean percentage of ACE activity precipitated by the indicated mAbs, determined over 2–3 independent experiments (which did not differ by more than 10%). Coloring is as in Figure 1

To define the epitope for mAb 5B3 (which behaved similarly to mAb 9B9 in Figure 2) we tested the binding of this mAb to mutant ACEs that were previously used for epitope mapping of mAbs 9B9 and 3G8. ²¹ The binding pattern of mAb 5B3 to 14 mutants of the N domain was very similar, but not identical to that of mAb 3G8 (Figure 3a,b). Since both mAb 5B3 and mAb 3G8 bind to the C‐domain chimera, where the first 141 residues resemble N domain, ³⁷ the stretches of the epitopes of 5B3 (as well as 3G8, but not for 9B9) are localized completely N‐terminally.

FIGURE 3.

ACE activity precipitation by novel mAbs to the N domain of ACE. The culture medium of CHO cells expressing mutated forms of the truncated N domain was used a source of N domain mutants. ²¹ , ³⁰ The culture medium of CHO cells expressing the truncated wild‐type N domain was used as a control. mAbs 3G8 ²¹ and 5F1 ³¹ were used for comparison (previously mapped mAbs to the N domain of ACE) to the novel mAbs 5B3 and 2D1. (a), (b) mAb 3G8 versus mAb 5B3. (c), (d) mAb 5F1 versus mAb 2D1. (e) mAb 6C8 versus mAb 6H6. The arrows emphasize differences in the effect of mutation on precipitation by the previously mapped and novel mAbs. These different mutants of the N domain of human ACE (adjusted to a final enzymatic activity of 5–10 mU/ml using ZPHL as a substrate) were incubated in a microtiter plate coated with tested mAbs and precipitated ACE activity was quantified. Data are presented as the mean percentage of ACE activity precipitated by the indicated mAbs, determined over 2–3 independent experiments (which did not differ by more than 10%). Coloring is as in Figure 1

Interestingly, binding of mAb 6C8 did not change with any of these 14 mutations of the truncated N domain, while binding of mAb 6H6 decreased with seven of them (Figure 3e). Thus, a diagrammatic representation of the epitopes of the novel mAbs 6C8 and 6H6 (and mAb 5B3) is presented as in Figure S2. The binding of mAbs 6C8 and 6H6 was completely absent in the R532Q mutant of the N domain (Figure 2e) similar to mAb 6A12 from a previously published series), widening our repertoire for the detection of ACE mutations that mimic sarcoidosis by dramatically increasing ACE shedding through prevention of bilirubin binding to the N domain of ACE and changes in the ACE conformation. ²⁰ , ³³ Also, the binding affinity of these two mAbs is sensitive to the presence of ACE inhibitors (~200% compared to the control (p > 0.05), but less so than mAb 6A12 (~400% compared to the control ³⁰ ), which is a consequence of bilirubin dissociation from the N domain of blood ACE after ACE inhibitor binding. ³³

To define the epitope for mAb 2D1, we analyzed the binding affinity of mAb 2D1 for nine mutants of the truncated N domain, and compared it with that of mAb 5F1 (from the previous series of mAbs ³¹ ). Despite significant differences in the binding of 2D1 and 5F1 to five different mutants (Figure 3c), the pattern of binding for these two mAbs suggested that their epitopes overlap, which was confirmed by competition experiments (data not shown). Both mAbs 2D1 and 5F1 are very sensitive to the presence of sialic acids on glycans, because their binding to serum/plasma ACE (where glycans are enriched with sialic acids ³¹ , ³⁴ was dramatically reduced by up to two‐fold (Figure 1c) in comparison to their binding to ACE from tissue or recombinant human ACE. A schematic of the epitopes of the novel mAb 2D1 is presented in Figure S3.

Epitope mapping of mAb 2H4 (from the first hybridization) was performed separately (S1). The novel mAb 2H4 recognized an epitope overlapping with that of mAb 9B9 (Figure S4), and partially with mAb 5B3 (Figure S5). The novel mAbs 5B3 and 2H4 could thus widen the repertoire of available mAbs to study ACE shedding, since 10‐fold more 5B3 bound to ACE than 3G8 which was the only mAb that decreased ACE shedding in vitro in CHO‐ACE cells, ²¹ , ³⁸ and reduced dimerization in reverse micelles. ²² In addition, mAbs 5B3 and 2H4 may be able to detect ACE originating from sarcoid granulomas, in human blood, since mAb 3G8 ¹³ had increased binding to blood ACE from patients with active sarcoidosis (which originated from activated macrophages and endothelial cells) in comparison to blood ACE from healthy donors ¹³ which originated from lung endothelial cells, the main source of ACE in the blood. ³⁹

It was shown previously that mAb 5F1 is sensitive to ACE dimerization and can also be used to detect ACE mutations which mimic sarcoidosis ²⁰ by increasing ACE shedding. ⁴⁰ Since the epitope of the novel mAb 2D1 (Figure S3) overlaps with that of 5F1, this mAb may increase our ability to study ACE dimerization and shedding‐related mutations.

To localize the epitopes for mAbs 4H11, 4F4, 2D7 and 1B12 (Figure 2), we measured their cross‐reactivity with ACE from the sera of an additional nine species: rats, mice, rabbits, hamsters, sheep, cats, mini‐pigs, guinea pigs, goats and dogs. The novel mAbs 1B12 and 2D7 demonstrated significant, and almost identical, cross‐reactivity to ACE from sheep, rabbits, mice, rats and hamsters (Figure S6), while mAbs 4H11 and 4F4 did not recognize any of these (see also Table 1). Analysis of the pattern of cross‐reactivity of mAbs 1B12 and 2D7, protein sequence alignment of tested ACEs (Figure S6), and competition of all four mAbs with mAbs i2H5 or 9B9 (Figure S7A, S7B) were used to map the epitopes of these four mAbs (Figure S8). An extended view of the 2D7 epitope is presented in Figure S9. This epitope contains A232 which is interesting since the A232S substitution is associated with Alzheimer's disease. ⁴¹ This substitution may be another example of a transport deficient (i.e., Loss‐of‐Function, LoF) ACE mutation (similar to R1069Q ⁴² ), in which case carriers of this mutation should have decreased ACE expression, and thus decreased blood ACE levels. Alternatively, the close proximity of this mutation to the entrance of N domain active site cleft might cause impaired cleavage of N‐domain selective substrates, such as Aβ 1–42, ⁴³ in carriers of this mutation. Therefore, future studies should be undertaken to measure the effect of AD‐associated ACE mutations on the expression of ACE and its cleavage of substrates such as Aβ 1–42.

TABLE 1.

Characteristics of the novel set of mAbs to ACE

							Species cross–reactivity, % from human ACE
Domain specificity	mAb	Antigenic region	Overlap with previous mAb	Glycosylation site in the epitope	ACE SF/ lung binding ratio, %	KD nM	Chimp	M Rh	Goat	Sheep	Cat	Rat
N domain	9B9	1		Asn 9/25/82	95	1.6	83	12	‐	‐	15	28
	2H4	1	9B9	Asn 9/25/82	103	2.9	68	0	306	333	‐	‐
	5B3	1	3G8	Asn 9/25	112	2.3	75	4	‐	‐	‐	‐
	1B12	2	I2H5/1G12	Asn 289/416	96	NT	59	76	72	41	‐	3
	2D7	2	I2H5/1G12	Asn 289/416	90	6.0	65	79	60	33	‐	3
	4F4	2	I2H5/1G12	Asn 289/416	88	NT	66	77	‐	‐	‐	‐
	4H11	2	I2H5/1G12	Asn 289/416	98	NT	71	105	‐	‐	‐	‐
	6C8	2	I2H5/1G12	Asn 45/289/416	95	2.5	0	2	‐	‐	‐	‐
	6H6	2/3	6A12/5F1	Asn 25/45/289	59	124	1	1	‐	‐	‐	‐
	2D1	3	5F1	Asn 45/117	38	42	5	136	‐	‐	‐	‐
C domain	3C10	4	1E10	Asn 666/685	103	2.3	73	38	18	10	11	‐
	5G8	5	1B8/3F10	Asn 731	137	0.4	95	22	‐	‐	‐	‐
	8H1	5	1B8/3F10	Asn 731	110	0.3	65	63	‐	‐	‐	‐
	4C12	5	1B3/3F10	Asn 731/1196	104	NT	106	105	‐	‐	1.0	0.5

Note: The strongest mAb from the previous set (9B9) was included for comparison (green coloring). The properties of the previous set of mAbs to ACE were previously published. ¹³ , ¹⁶ , ²¹ , ³² Antigenic regions were defined according to the epitope overlaps. Coloring of some of the species cross‐reactivity values as in Figure 1. SF, seminal fluid. Hyphens indicate that no binding was observed. Chimp: chimpanzee; M.Rh., Macaque Rhesus. Determination of the binding affinity constant (K_D) of mAbs to purified human lung ACE was performed by Surface Plasmon Resonance (SPR) in triplicates with SD less than 10%. NT, not tested.

2.4. Epitope mapping of novel mAbs to the C domain of human ACE

We examined the cross‐reactivity of five novel C domain mAbs with ACE from macaque rhesus (Macaca mulatta) lung homogenate, and sera from 10 different species: chimpanzees, rats, mice, rabbits, hamsters, sheep, cats, goats, mini‐pigs, guinea pigs and dogs, to obtain further information regarding C‐domain epitope localization. The cross‐reactivity was compared with that of well‐characterized mAbs to the immunodominant C domain epitopes (mAbs 1B8/3F10), as well as the anti‐catalytic mAb 1E10. ³² The novel C domain mAb 3C10 was different to the other four novel mAbs, based on its serum ACE/seminal fluid ACE ratio (Figure 1c). We examined the cross reactivity of mAb 3C10 with ACE from different species relative to 1E10, performed sequence alignment (Figure S10), and compared it with the cross reactivity of mAb 1E10. Despite different patterns of cross‐reactivity for mAbs 3C10 and 1E10 (Figure S10), these mAbs recognized highly overlapping epitopes on the C domain (Figure S11) which was confirmed by a competition experiment (data not shown). Furthermore, different amino acid residues are crucial for the binding of these mAbs to ACE: D616 and K677 for mAb 1E10 ³² and motifs 621SK622 and 681NQL683 for mAb 3C10. It has been reported that mAb 1E10 can discriminate between testicular ACE and somatic ACE in solution ³² and, especially, between ACE on the surface of spermatozoa and the surface of somatic cells. ⁴⁴ Since the epitope for mAb 3C10 overlaps with that of mAb 1E10, and 3C10 binds with a four‐fold stronger affinity than mAb 1E10, this mAb might have potential as an immunocontraceptive drug. Furthermore, mAb 3C10 recognized denatured ACE by Western blotting (data not shown) which indicates that this mAb recognized at least one sequential stretch of its epitope.

The cross‐reactivity of the remaining four mAbs to the C domain of ACE, and their corresponding sequence alignments, are shown in Figure S12. Competition experiments (Figure S7C‐E) demonstrated that the epitope of mAb 5G8 has 70% overlap with the epitope for mAb 3F10, while the mAb 4C12 overlaps with mAb 3F10 by 50% (Figure S13). All of these overlapping epitopes contained the glycosylation site Asn731, which is glycosylated quite differently in ACE on the epithelial cells of the prostate and epididymis versus ACE on endothelial cells. ¹⁵ Therefore, the novel mAbs 5G8 and 4C12 could improve our capacity to discriminate between ACE originating from different cells and tissues.

The novel mAb 8H1 recognizes a region on the C domain globule that was not recognized by the previous sets of mAbs, because it did not compete with mAb 3F10 (and its group) (Figure S7C‐E), or with mAbs 1E10 and 3C10 (data not shown). The cross‐reactivity of mAb 8H1, and the observed decrease in binding to macaque rhesus ACE (Figure S12) argued that this decrease is likely due to mutation of two residues of ACE in this region in macaque rhesus ACE – D753E and T877P (Figure S12). While T877P is not on the surface, D753 is, and the substitution of Asp by Glu may thus have an impact on mAb 8H1 binding. However, substitution of T877 by Pro can introduce a turn in the polypeptide chain and thus residues next to T877P (which are on the surface, such as D876, T878 and E879, and perhaps other residues on this helix) may dramatically change the local conformation. The epitope for 8H1 is shown in Figure 4a,b. Interestingly, the epitope for this mAb contains three ACE mutations that were previously associated with Alzheimer's disease: E738K, ⁴¹ , ⁴⁵ T887M ⁴⁵ , ⁴⁶ and N1007K. ⁴¹ Binding of mAb 8H1 was at least five‐fold lower for ACE from a patient with the N1196K/T877M double mutation ²⁰ (Figure 4c), indicating that mAb 8H1 could serve as a marker for the T877M Alzheimer's disease‐associated mutation. Binding of mAb 1B3 to ACE from a patient with the P1199L mutation decreased several folds, confirming previous results. ¹⁶ Decrease of mAb 1A2 binding to ACE from both patients could be considered as an indication that the epitope for 1A2 is sensitive to conformational changes in the C terminal end of soluble ACE due to different mutations in the juxtamembrane stalk region.

FIGURE 4.

Fine epitope mapping of mAb 8H1 on the C domain of ACE. The crystal structure of the of the C domain fragment of human ACE, where the 36 amino acid residues unique to tACE were deleted (PDB 2XY9) is shown using molecular surface (a) and ribbon (b) representations. Key amino acids referred to in the text are denoted using somatic ACE numbering. The ribbon and surface are colored gray, with some amino acid residues colored as follows: asparagines of the putative glycosylation sites are highlighted in green. The C‐terminal end of this C domain fragment (1187–1,200) is marked in light blue. Amino acid residues (837AQH) that are crucial for mAb 1B3 (directed to the C terminal end) and P730 (crucial for mAb 3F10) are marked in red; C734 is the end of the cysteine bridge (C728‐C734). The epitopes for mAb 8H1 to the C domain are shown using circles with a diameter 30 Å, which correspond to approximately 700 Ų of the area covered by this mAb. (c) ACE activity was precipitated by several tested mAbs from citrate plasma of patients with P1199L mutation and from patient with two ACE mutations –N1196K and T887M ²⁰ and compared to that precipitated from control human citrated plasma. Coloring of the bars is as in Figure 1. Data represent the mean ± SD from 2–5 (depending on mAbs) independent experiments (each in triplicate)

It is interesting to note that the AD‐associated mutations on the N and C domain (A232S, E738K, T887M and N1007K) are all located on, or near, hinge regions in these domains. It has previously been shown via molecular dynamics simulations, ⁴⁷ and more recently also proposed by X‐ray crystallography, ⁴⁸ that these flexible loops facilitate conformational change from an open to a closed active site upon ligand binding. Although it needs further investigation, it is possible that these AD‐associated mutations may alter hinging and thus recognition by mAbs 2D7, 5G8, and 8H1. These novel mAbs could likely be particularly useful for the identification of patients who are susceptible to Alzheimer's disease due to LoF ACE mutations. Table 1 summarizes important characteristics of the novel mAbs to ACE. Affinity constants for the tested mAbs that were determined by Surface Plasmon Resonance (Biacore) vary from 0.3 to 124 nM.

2.5. Conformational fingerprinting of ACE in dendritic cells

Previously we demonstrated that some ACE mAbs could be useful for the study of sarcoidosis. ¹³ , ²⁰ To understand the applicability of the novel ACE mAbs in this study for sarcoidosis research and diagnostics, we performed conformational fingerprinting of soluble ACE from dendritic cells.⁴⁹

The pattern of ACE activity precipitation using 25 mAbs (conformational fingerprinting) from immature and mature DC culture medium (i.e., soluble ACE) was compared with that for purified ACE from human lung, a representative of endothelial ACE, ³⁹ and with culture medium from CHO‐ACE cells, representative ACE from epithelial cell ACE. ¹³ It is evident that the local conformation of ACE in DC differs dramatically from that in lung and epithelial cells (Figure 5). Additionally, it is clear that the local conformation of ACE in immature and mature DC is practically identical (Figure 5a,b). mAb 2H4 could be considered as a positive marker for ACE from DC considering its binding to soluble DC ACE was 4‐fold more than to lung ACE.

FIGURE 5.

Conformational fingerprinting of soluble ACE from dendritic cells (DC). (a), (b), Culture medium from immature DC (a), and mature DC (b), was incubated with wells coated by 25 different mAbs to ACE. Precipitated ACE activity was quantified by spectrofluorometric assay with ZPHL. And compared with ACE activity precipitation from purified ACE from human lung. (c), Culture medium from CHO‐ACE cells (clone 2C2) ¹³ was used as a source of ACE from epithelial cells. Data are expressed as the mean percentage of ACE activity precipitated by the indicated mAbs, determined in 2–5 independent experiments (each in triplicates). Results were normalized by added ACE activity in four tested ACE preparations and presented as DC/ lung ACE or CHO‐ACE / lung ACE binding ratios. Coloring as in Figure 1

The binding of all mAbs (except mAb 2H4) to DC ACE was much lower than the binding to endothelial ACE. A probable explanation is the much higher sialylation of DC ACE in comparison to that of endothelial cell ACE. This may be of diagnostic value, because mAbs binding/activity ratio for serum ACE in patients with active sarcoidosis (i.e., having a significant proportion of ACE from granulomas in addition to ACE from lung capillaries) should be significantly lower than in patients without sarcoidosis (or patients with pulmonary only sarcoidosis).

Differences in the mAbs binding pattern (i.e., local ACE conformation) to ACE from epithelial and endothelial cells are also significant (Figure 5c), and epitope specific. Different binding could be due to different glycosylation of potential glycosylation sites ‐Asn 25/82 and Asn 45/117. Conformation fingerprinting of ACE easily identified the potential difference in glycosylation (or maybe only sialylation) of Asn1196 in lung epithelial cells (Figure 5c) and in endothelial cells. However, differences between the binding of certain mAbs to soluble DC ACE and ACE entering the circulation and subsequent filtration by reticuloendothelial system of the liver could be significant. ¹³ Thus, providing the conformational fingerprints of ACE are different for blood samples, this approach could have diagnostic value for sarcoidosis.

3. CONCLUSION

Thus, based on the binding properties of these mAbs to different ACE preparations, constructs, and species we have localized the epitopes for 14 novel mAbs, as summarized schematically on a model of two‐domain somatic ACE (Figure 6).

FIGURE 6.

Schematic diagram of the epitope localization of mAbs on human somatic ACE. The schematic diagram was prepared using the UCSF ChimeraX v1.0 molecular visualization program by orientating the N‐ and C‐domain crystal structures in a potential sACE‐like conformation. The epitopes for mAbs to the N and C domain are shown using molecular surface representations of the ligand‐bound crystal structures of the N domain (PDB 4BXK) and C domain (PDB a fragments of human somatic ACE. The juxtamembrane stalk region, which terminates at Pro1209 in this C‐domain crystal structure, is shown as a loop in the cell membrane (represented by a dashed line) while the inter‐domain linker is represented by a dashed arc. The surfaces are colored in wheat, with the epitopes proposed for each mAb, and the mAb name, given in different colors. Asparagine residues of the putative glycosylation sites are highlighted in lime green. Potential AD‐associated mutations are highlighted in magenta

Generation of mAbs that can recognize conformational epitopes on the N and C domains of ACE provides a tool for studying detailed aspects of ACE biology. In the present study, we characterized 14 novel mAbs recognizing different epitopes on the N and C domains of human ACE. These mAbs are sensitive to even subtle conformational changes of the ACE structure, and thus have the potential to be used as probes for the detection and quantification of different forms of ACE. The novel mAb 5B3, which overlaps with mAb 3G8 (directed to the N domain of ACE) but binds with higher affinity, could be helpful for studying ACE shedding as well as the detection of ACE originating from sarcoid granulomas in the patient plasma/serum, as 3G8 did. The novel mAb 2D1 (which overlaps with mAb 5F1, also directed to the N domain) may increase our ability to study ACE dimerization and shedding‐related mutations. The novel mAb 8H1, which recognizes an epitope on the C domain of ACE, could be helpful for the identification of several ACE mutations that are associated with Alzheimer's disease (at least T877M), while mAb 2H4 could be considered as a marker for ACE sialylation in dendritic cells.

Together with knowledge about their epitope localization, this panel of mAbs provides a powerful tool for studying ACE biology in humans.

4. MATERIALS AND METHODS

4.1. Purification of ACE antigen

Somatic ACE from human seminal fluid and human lung tissue was purified using lisinopril affinity chromatography. ⁵⁰ Purified ACE from seminal fluid was used as an antigen for immunization, and for screening of primary hybridoma cell populations and clones.

4.2. ACE activity assay

The enzymatic activity of human and animal tissues and sera, as well as different forms of recombinant human ACE, were measured using a fluorimetric assay with substrates Hip‐His‐Leu (HHL, 5 mM) and Z‐Phe‐His‐Leu (ZPHL, 2 mM) as described previously. ¹³

4.3. Immunization, fusion, and screening

Mice were immunized five times with pure seminal fluid ACE at 10ug per i.p. injection, as previously described. ¹⁵ The sera were tested using ELISA and a plate precipitation assay against two ACE isoforms: pure seminal fluid ACE and pure lung ACE. Upon reaching antibody titers of 5 × 10⁵‐10⁶ according to the ELISA, mice were boosted and splenocytes were isolated and fused to mouse myeloma cells 653‐Ag8, subclone P₃O₁ in the first hybridization and Sp2/0‐Ag14 in the second and third hybridizations. The resulting populations were grown in 96‐well plates and were subjected to primary screening with seminal fluid ACE and lung ACE in a plate precipitation assay simultaneously. Selected populations were cloned, and supernatants were repeatedly assayed to identify the best clones. mAb Ig class and subclass determination was performed using a Mouse Typer® Sub‐Isotyping Kit (Bio‐Rad Laboratories, Hercules, CA) according to the manufacturer's instructions.

4.4. Plate immunoprecipitation assay

Immunoprecipitation assays were carried out according to Reference 12. Shortly, microtiter (96‐well) plates (high binding, Corning, Corning, NY) were coated with anti‐ACE mAbs via goat anti‐mouse IgG (IMTEK, Moscow, Russia or Invitrogen, Rockford, IL) bridge and incubated with serum samples. After washing the unbound ACE with PBS‐Tween 20, plate‐bound ACE activity was measured by adding a substrate for ACE (ZPHL) directly into each well. Conformational fingerprinting of ACEs with mAbs to ACE was performed as described previously. ¹³

4.5. ACE constructs and stable cell lines

Stable cell lines of Chinese Hamster Ovary (CHO) cells expressing wild type human somatic ACE as well as other human ACE constructs/mutants, were generated according to Reference 25, and mutants of the ACE N domains were generated according to Reference 21.

4.6. Binding analysis by surface plasmon resonance

Measurement of binding affinity constants for the mAbs to ACE was performed at the Biophysics Core of the Research Resource Center of the University of Illinois at Chicago using the Biacore 8 K instrument (Cytiva, Marlborough, MA, USA) and a CM5 chip.

PBS with 0.005% Tween‐20 (PBSTween) was used as a system running buffer. Two approaches were used: (1) Purified human lung ACE was immobilized directly on the surface of sensor chip were coated with purified human lung ACE at the 400–800 response unit (RU) level. mAbs (as a pure mAbs or as a culture medium from hybridoma cells in 10% of FBS) were applied over immobilized ACE by injection various concentrations (0.082 to 50 nM) using a flow rate of 20 μl min⁻¹. (2) Alternatively, sensor chip was coated with affinity‐purified Goat‐anti‐mouse IgG polyclonal antibodies (Invitrogen, Rockford, IL) and then with mAbs to ACE to a density of 1,400–1,600 response units (RU) After that purified human lung ACE was applied by injection in triplicate of various concentrations (0.082 to 50 nM) for 240 sec association and 600 sec dissociation time at 20°C with running buffer. ACE concentration was determined based on the enzymatic activity with ZPHL as a substrate. All experiments were repeated three times. The chip surface was regenerated with 100 mM citric acid in the case of titration by ACE as analyte. The resulting sensorgrams were analyzed using Biacore 8 K Evaluation Software v2.0. Data were referenced with RU values of blank channel and assay buffer alone before fitting the data with 1 to 1 steady‐state affinity equation embedded in the Biacore 8 K Software.

Evaluation Software. Affinity constants were calculated as K_D = k_d/k_a.

4.7. Generation of dendritic cells and collection of culture medium

The generation of dendritic cells was performed from blood of 10 healthy individuals as described in Reference 51. Informed consent was obtained from all donors according to the Declaration of Helsinki (World Media Association 2000). The protocol for a research project has been approved by a suitably constituted Ethics Committee of Institute of Fundamental and Clinical Immunology, Institute of Traumatology and Orthopedics, Federal Neurosurgical Center, and Institute of Cytology and Genetics, within which the work was undertaken. Peripheral blood mononuclear cells (MNCs) were obtained by density gradient centrifugation (Ficoll‐Paque; Sigma‐Aldrich, Germany) of heparinized whole blood samples and DCs were generated by culturing of plastic‐adherent MNC fraction in RPMI‐1640 medium supplemented with 2.5% fetal calf serum in the presence of rhGM‐CSF and IL‐4 for 4 days at 37°C and 5% CO₂ atmosphere. The maturation of DCs was induced by further exposure with a standard lipopolysaccharide concentration (10 μg/ml) for an additional 24 h. ⁵¹

AUTHOR CONTRIBUTIONS

Isolda Popova: Investigation; methodology. Lizelle Lubbe: Formal analysis; investigation; validation; visualization; writing‐review & editing. Pavel Petukhov: Methodology; visualization. Gavriil Kalantarov: Investigation; methodology. Ilya Trakht: Formal analysis; funding acquisition; methodology; resources. Elena Chernykh: Funding acquisition; methodology; resources. Olga Leplina: Methodology; resources. Alex Lyubimov: Funding acquisition; project administration; resources. Joe Garcia: Funding acquisition; supervision. Steven Dudek: Funding acquisition; resources. Edward Sturrock: Funding acquisition; supervision; visualization; writing‐review & editing. Sergei Danilov: Conceptualization; data curation; formal analysis; investigation; methodology; project administration; resources; supervision; validation; visualization; writing‐original draft; writing‐review & editing.

CONFLICT OF INTEREST

Dr. Garcia is Founder and CEO of Aqualung Therapeutics Corp. (Tucson, AZ).

Supporting information

Appendix S1: Supplementary Information

Figure S1 Effect of mAbs on the recognition of recombinant soluble ACE from CHO cells

A. Precipitation of ACE activity in culture medium from recombinant soluble ACE (WTΔ) relative to purified ACE from human lung homogenate presented as WTΔ/ACE lung binding ratio, as in Figure 1. ACE activity in culture medium/purified solution was equilibrated using ZPHL as a substrate prior to plate precipitation assays. B. ACE activity precipitated from the culture medium of CHO cells expressing human somatic ACE without transmembrane anchor (WTΔ) by each given mAb was determined fluorometrically directly in the wells (as in Fig. 1) with two substrates for ACE, ZPHL and HHL. Data are expressed as the ratio of the rates of hydrolysis of the two substrates (ZPHL/HHL ratio) by ACE bound to mAbs in comparison with the mean value obtained for all 20 mAbs. * ‐ p < 0.05 in comparison with mean value for the whole set of mAbs. Data represent the mean ± SD from 2–3 independent experiments (each in triplicate). Coloring of the bars is as in Figure 1.

Figure S2 Fine epitope mapping of mAb 6C8 and 6H6 on the N domain of ACE.

The crystal structure of the of the N domain fragment of human ACE, where 7 amino acid residues determining putative glycosylation sites were mutated (PDB 3NXQ) is shown using ribbon and molecular surface representations. Key amino acids referred to in the text are denoted using somatic ACE numbering. The ribbon and surface are colored gray, with some amino acid residues colored as follows: asparagines of the putative glycosylation sites are highlighted in green. Amino acid residues (R407 and R532) crucial for mAb 6A12, mAbs 6C8 and 6H6 are marked in red. The epitopes for mAbs 6C8 and 6H6 to the N domain (and mAb 6A12 from a previously published mAb series ³⁰ ) are shown using circles with a diameter of 30 Å, which correspond to an area of approximately 700Ų covered by these mAbs.

Figure S3 Fine epitope mapping of mAb 2D1 on the N domain of ACE.

The crystal structure of the N domain fragment of human ACE (PDB 3NXQ) is shown as in Figure S2. Amino acid residues crucial for mAb 2D1 binding to ACE are marked in red. The epitopes for mAb 2D1 (and the previously mapped mAb 5F1 ³¹ ) are shown on the N domain using circles as in Figure S2.

Figure S4 Fine epitope mapping of mAb 2H4 on the N domain of ACE.

A. ACE purified from human lung homogenate was incubated with 100‐fold excess of tested mAbs (or with PBS‐as a control) and was then incubated in a plate coated with mAb 2H4. ACE activity precipitated by mAb 2H4 was measured directly in the wells as in Figure S1. B. ACE activity of N domain fragment mutants precipitated by mAbs 2H4 was compared to that precipitated by mAb 9B9 as in Figure S1. Coloring of the bars is as in Figure 1. Data represented the mean ± SD from 2–3 independent experiments (each in triplicate). C. The crystal structure of the N domain fragment of human ACE (PDB 3NXQ) is shown using a molecular surface representation. Key amino acids referred to in the text are denoted using somatic ACE numbering. The surface is colored gray, with some amino acid residues colored as follows: asparagines of the putative glycosylation sites are highlighted in green. Amino acid residues crucial for mAb 2H4 binding to ACE are marked in red. The epitopes for mAb 2H4 on the N domain (and for the control mAb 9B9 ²¹ ) are shown using circles with a diameter of 30 Å, which correspond to an area of approximately 700Ų covered by these mAbs.

Figure S5 Fine epitope mapping of mAb 5B3 on the N domain of ACE.

The crystal structure of the of the N domain fragment of human ACE, (PDB 3NXQ) is shown using ribbon and molecular surface representations. Key amino acids referred to in the text are denoted using somatic ACE numbering. The ribbon and surface are gray, with some amino acid residues colored as follows: asparagines of the putative glycosylation sites are highlighted in green. Amino acid residues crucial for mAb 5B3 binding to ACE are marked in red. The epitopes for mAb 5B3 the N domain (and control mAb 3G8 ²¹ ) are shown using circles with a diameter of 30 Å, which correspond to an area of approximately 700Ų covered by these mAbs.

Figure S6 Cross‐reactivity of mAbs 1B12 and 2D7 with mammalian ACEs. mAbs 1B12 and 2D7 precipitated ACE activity from purified seminal fluid ACE significantly (see also Figure 1(a)). ACE activity was also precipitated from lung homogenate of macaque rhesus and sera from 10 species (chimp, sheep, rabbit, hamster, cat, mini‐pig, guinea pig, goat, rat, and mouse). These two mAbs precipitated measurable ACE activity from sera or lung homogenate of 8 species, expressed as a % of that of human serum (or human lung homogenate in the case of macaque rhesus lung homogenate). Excerpts from the alignment of the N domains of mammalian ACE to which mAbs 1B12 and 2D7 cross‐reacted are highlighted as follows: 7 sites of N‐glycosylation and their corresponding asparagines are highlighted in green. Amino acid residues that were primate‐specific are highlighted in yellow. Amino acid residues that were changed in macaque rhesus are highlighted in magenta. Amino acid residues in sheep and rabbit ACE which could be responsible for the significant decrease in mAbs binding, are highlighted in blue. Amino acid residues in hamster, rat, and mouse, which could be responsible for the dramatic decrease in mAbs binding, are highlighted in grey. Amino acid residues in human ACE which were likely responsible for mAbs 1B12 and 2D7 binding, are highlighted in red. The sequence of the first 52 amino acid residues of sheep ACE were incorrect in the database (in our opinion) and are presented here as Xs). Residues identical to human ACE are represented by dashes. Alignments were performed using ClustalW.

Figure S7 Competition between novel mAbs and mAbs with mapped epitopes for ACE.

ACE purified from seminal fluid (5–10 mU/ml with ZPHL as a substrate) was pre‐incubated with 100‐fold molar excess of tested (competing) mAbs and then added to plates coated with the indicated mAbs (A‐E). Precipitated ACE activity was quantified directly in the wells with ZPHL as a substrate and expressed as a % of the amount of precipitated ACE activity without competing mAbs. Data are presented as the mean percentage of ACE activity precipitated by the indicated mAbs, determined over 2–3 independent experiments (which did not differ by more than 10%). Coloring is as in Figure 1.

Figure S8 Fine epitope mapping of mAbs 2D7/1B12/4H11/4F4 on the N domain of ACE.

The crystal structure of the of the N domain fragment of human ACE, where 7 amino acid residues determining putative glycosylation sites were mutated (PDB 3NXQ) is shown using ribbon and molecular surface representations. Key amino acids referred to in the text are denoted using somatic ACE numbering. The ribbon and surface are colored gray, with some amino acid residues colored as follows: asparagines of the putative glycosylation sites are highlighted in green. Amino acid residues crucial for mAb 2D7 binding to ACE are highlighted in red. The epitopes for mAbs 2D7/4H11/4F4 on the N domain (and for control mAb i2H5 ²⁹ ) are shown using circles with a diameter of 30 Å, which corresponds to an area of approximately 700Ų covered by these mAbs.

Figure S9 Extended version of 2D7 epitope.

The crystal structure of the of the N domain fragment of human ACE, where 7 amino acid residues determining putative glycosylation sites were mutated (PDB 3NXQ) is shown and colored as in Figure S8. Amino acid residues crucial for mAb 2D7 binding to ACE are highlighted in red. The position of the ACE mutation (A232S) associated with Alzheimer's disease ⁴¹ is shown by arrow. The epitopes for mAbs 2D7 (on the N domain (and for control mAb i2H5) are shown using circles with diameter of 30 Å.

Figure S10 Cross‐reactivity of mAb 3C10 (and 1E10) with mammalian ACEs. mAbs 3C10 (and 1E10) precipitated ACE activity from purified ACE preparation from seminal fluid significantly (90 and 74% compared to that precipitated by mAb 9B9, respectively, − see also Figure 1(a)). ACE activity was also precipitated from lung homogenate of macaque rhesus and sera from 10 species (chimp, sheep, rabbit, hamster, cat, mini‐pig, guinea pig, goat, rat and mouse). These two mAbs precipitated measurable ACE activity from serum or lung homogenate of 7 species, expressed as a % of that precipitated from human serum (or human lung homogenate in the case of macaque rhesus lung homogenate) (these values are given in bold). Excerpts from the alignment of the N domains of mammalian ACEs to which mAbs 3C10 (and 1E10) cross‐reacted are highlighted as follows: 4 sites of N‐glycosylation and the corresponding numbers of their asparagines are highlighted in green. Amino acid residues that were human‐specific are highlighted in yellow, and those changed in macaque rhesus, are highlighted in magenta. Amino acid residues in human ACE which could be crucial for 3C10 binding are highlighted in red. Residues identical to human ACE are represented by dashes. Alignments were performed using ClustalW.

Figure S11 Fine epitope mapping of mAb 3C10 on the C domain of ACE.

The crystal structure of the of the C domain fragment of human ACE, where the 36 amino acid residues unique to tACE were deleted (PDB 2XY9) is shown using ribbon and molecular surface representations. Key amino acids referred to in the text are denoted using somatic ACE numbering. The ribbon and surface are colored gray, with some amino acid residues colored as follows: asparagines of the putative glycosylation sites are highlighted in green. Amino acid residues crucial for mAb 3C10 binding to ACE are highlighted in red. The epitope for mAb 3C10 on the C domain (and for control mAb 1E10 ³² ) are shown using circles with a diameter of 30 Å, which corresponds to an area of approximately 700Ų covered by these mAbs.

Figure S12 Cross‐reactivity of mAbs 4C12, 5G8 and 8H1 with mammalian ACEs. mAbs 4C12, 8H1 and 5G8 precipitated ACE activity from purified seminal fluid ACE significantly (91, 78 and 117%, respectively, relative to that precipitated by mAb 9B9 ‐ see also Figure 1(a)). ACE activity was also precipitated from the lung homogenate of macaque rhesus and sera from 10 species (chimp, sheep, rabbit, hamster, cat, mini‐pig, guinea pig, goat, rat and mouse). These mAbs precipitated measurable ACE activity from serum and lung homogenate of indicated species, expressed as a % of that precipitated from human serum (or human lung homogenate in the case of macaque rhesus lung homogenate) (these values are given in bold). Excerpts from the alignment of the C domains of mammalian ACE to which the tested mAbs cross‐reacted, are highlighted as in Figures S5 and S7. The amino acid residues in human ACE which are responsible for mAbs 5G8 and 8H1 binding are highlighted in red, in light blue for mAb 4C12, and in magenta for mAb 3F10. Residues identical to human ACE are represented by dashes. Alignments were performed using ClustalW.

Figure S13 Fine epitope mapping of mAb 5G8 and 4C12 on the C domain of ACE.

The crystal structure of the of the C domain fragment of human ACE, where the 36 amino acid residues unique to tACE were deleted (PDB 2XY9) is shown using ribbon and molecular surface representations. Key amino acids referred to in the text are denoted using somatic ACE numbering. The ribbon and surface are colored gray, with some amino acid residues colored as follows: asparagines of the putative glycosylation sites are highlighted in green. The C terminal end of this C domain fragment (1187–1,200) is marked in light blue. Amino acid residues (837AQH) crucial for mAb 1B3 (directed to the C terminal end) and P730 (crucial for mAb 3F10) binding are marked in red. The C734‐end of the cysteine bridge (C728‐C734) is marked in yellow. Amino acid residues D753 and T877 crucial for mAbs 5G8 and 8H1 are marked in yellow. The epitopes for mAbs 5G8 and 4C12 to the C domain (and for control mAb 3F10 ³² ) are shown using circles with a diameter of 30 Å, which correspond to an area of approximately 700Ų covered by each of these mAbs.

Popova IA, Lubbe L, Petukhov PA, et al. Epitope mapping of novel monoclonal antibodies to human angiotensin I‐converting enzyme. Protein Science. 2021;30:1577–1593. 10.1002/pro.4091