Activation of human peripheral blood mononuclear cells (PBMC) with phytohemagglutinin (PHA) strongly upregulates expression of CD38 in human T lymphocytes 34. Therefore, to isolate a CD38 cDNA, PHA-activated cynomolgus PBMC were chosen as the source of RNA for amplification by RT-PCR using primers derived from the human CD38 5' and 3' untranslated regions. The 1113 base-pair (bp) insert contained an open reading frame of 906 bp (Figure 1A) that was 95% identical to the human CD38 sequence. The cDNA encodes a 301 amino acid (aa) polypeptide with the typical CD38 type II membrane protein structure, i.e., a short cytosolic tail (residues 1–21), a transmembrane region (residues 22–44), and an ECD (residues 45–301) containing the signature 12-cysteine array. Alignment of the macaque and human CD38 polypeptides showed 92% identity and 94% similarity. There is complete conservation of the IC region while there are five conservative changes in the transmembrane region where macaque CD38 has one more residue than human CD38. Macaque CD38 has four potential N-linked glycosylation sites, as in human CD38; three are co-linear. Furthermore, macaque CD38 shows conservation of the 4 acidic residues (Glu¹⁴⁸, Asp¹⁴⁹, Asp¹⁵⁷, Glu²²⁸) and 2 tryptophans (Trp¹²⁷ and Trp¹⁹¹) that play a critical role in the ADP-ribosyl cyclase/cADPR hydrolase activities of human CD38 35. Lys¹³⁰ is also maintained suggesting that, like human CD38, binding of ATP to this residue may lead to inhibition of the hydrolase activity 36. Likewise, macaque CD38 conserves Arg²⁷¹ which is ADP-ribosylated in human CD38, causing inactivation 37.

Human CD38 has been characterized as a single-copy, 8-exon gene that spans ~70 kb 538 and not only CD38 but ADPR cyclase genes in general are highly conserved from mollusks to humans 3940. In addition, the Southern blot hybridization patterns of macaque and human genomic DNAs digested with EcoRI, BamHI and HindIII and probed with their homologous CD38 cDNA are similar (data not shown), indicating that the structural organization of macaque and human CD38 is highly conserved. This was further demonstrated by the finding that the same primers previously used to amplify the 8 human CD38 exons 41 also amplified 8 CD38 exons from macaque genomic DNA. The putative macaque exons could be perfectly aligned with their human counterparts 5 in number, size and splice site of their exons (Table 1). All intron-exon boundaries conformed to the GT-AG rule, most 5' splice donor and 3' splice acceptor sequences of the 7 introns were identical.

To analyse surface expression and function of macaque CD38, the cDNA insert was subcloned into the pcDNA3.1 expression vector and a similar construct was prepared containing the human CD38 cDNA (see Materials & Methods). DNAs were transfected for heterologous expression in the NIH/3T3 cell line which does not express CD38 43. Given that cross-reactivity of OKT10 anti-human CD38 mAb has been exploited to detect CD38 in rhesus macaque hematopoietic cells 4445, clones expressing cynomolgus CD38 were identified by indirect immunofluorescence (IF) with OKT10 (see below). The stable cell line expressing macaque CD38 was designated NIH/mac38, while NIH/hum38 is the human CD38-expressing cell line.

The ecto-cyclase activity of macaque CD38 was evaluated by incubating the NIH/mac38 clone with nicotinamide guanine dinucleotide (NGD), an NAD analog which is converted by ADP-ribosyl cyclases such as CD38 to cyclic GDP ribose (cGDPR). Unlike cADPR, cGDPR is a stable, fluorescent end-product which can be detected in cell supernatants 46. Increased fluorescence following incubation with NGD was detected in supernatants of macaque and human CD38 transfectants, demonstrating that macaque CD38 is enzymatically active (Figure 2A).

In human red blood cells (RBCs), CD38 is the only source of ecto-cyclase activity 47 which was also found on the surface of cynomolgus macaque RBCs (Figure 2A), suggesting a further parallel with human CD38.

To establish the approximate molecular weight of macaque CD38, SDS-PAGE and Western blot analysis were performed with lysates prepared from NIH/mac38 and NIH/hum38 cells. Blots were probed with the AT1 anti-human CD38 mAb which detected a band of ~42 kDa in both transfectants (Figure 2B).

To raise mouse mAbs against macaque CD38, live NIH/mac38 cells were used for immunization. Four mAbs, KK1B5 (IgG1), KK4E5 (IgG2a), KK6A11 (IgG2a) and KK9H4 (IgG1) were selected for further analyses. All four anti-macaque CD38 mAbs reacted by IF with NIH/mac38 but were negative with the parental cell line (Figure 3). Only two mAbs (KK1B5 and KK9H4) reacted with NIH/hum38.

The anti-macaque CD38 mAbs were further assessed by IF for binding to cynomolgus PBMC and to SL-691 and SL-999, two cynomolgus macaque B lymphoblastoid cell lines. All four mAbs reacted with PBMC from cynomolgus macaques, indicating they recognize native cynomolgus CD38, and they strongly stained cells of the two B cell lines (Table 2).

Additional Western blot analyses were carried out with the anti-macaque CD38 mAbs. MAbs KK1B5 and KK9H4 confirmed detection of a ~42 kDa band in NIH/mac38 cell lysates but also recognized a second band of ~84 kDa. (Figure 2C). The doublet was also identified in SL-999 cynomolgus B cells (data not shown). Instead mAbs KK4E5 and KK6A11 only recognized a band of ~84 kDa, even in reducing conditions. No bands were detected by these mAbs in parental NIH/3T3 cells.

The observation that mAbs KK4E5 and KK6A11 recognize only cynomolgus CD38 while mAbs KK1B5 and KK9H4 also recognize human CD38 indicates that the two mAb subsets are directed towards different epitopes. To evaluate the contribution of disulfides to these epitopes, mAb reactivity was assessed after treating NIH/mac38 with dithiothreitol (DTT). Reduction had no effect on binding of the species-specific mAbs (KK4E5 and KK6A11) but significantly reduced binding of mAbs KK1B5 and KK9H4 (Figure 4A), indicating that the latter recognize a disulfide-requiring conformational epitope of the cynomolgus CD38 ECD, and predicting they should recognize a similar epitope in human CD38. The results (Figure 4B) confirm that treatment of NIH/hum38 with DTT decreased binding of mAbs KK1B5 and KK9H4, indicating that cynomolgus macaque and human CD38 have a conformational epitope in common.

Reciprocal experiments were performed to see if the correlation between cross-reactivity and epitope sensitivity to DTT was also valid for a panel of 6 well-known mAbs raised against human CD38. MAbs IB4, IB6, OKT10, SUN-4B7, AT1 and HB7 were assessed for binding to native and DTT-treated NIH/hum38, and for cross-reactivity with cynomolgus CD38. Binding of mAbs IB4, IB6 and HB7 to human CD38 was unaffected by DTT and none bound cynomolgus CD38 (Figure 4C and 4D). On the contrary, binding of mAbs OKT10, SUN-4B7 and AT1 to human CD38 was significantly reduced by DTT and all three mAbs bound cynomolgus CD38 (Figure 4D) in a DTT-sensitive manner.

The observation that mAbs KK1B5 and KK9H4 (raised against cynomolgus CD38) and mAbs OKT10, AT1 and SUN-4B7 (raised against human CD38) all bind native but not reduced cynomolgus and human CD38 is compatible with their binding the same epitope. A priori knowledge of the human CD38 epitope map previously established that OKT10 binding is abrogated by deletion of one or both of the C-terminal Cys residues (Cys²⁸⁷ and Cys²⁹⁶) predicted to pair in disulfide bond formation 48 and that mAbs OKT10, AT1 and SUN-4B7 mutually compete for binding to the human CD38 ECD 49. This would position the common epitope of cynomolgus and human CD38 in the C-terminal disulfide loop formed by Cys²⁸⁸ and Cys²⁹⁷ in cynomolgus CD38 (Cys²⁸⁷/Cys²⁹⁶ in human CD38).

To test this possibility, we analysed reactivity of these mAbs with the CD38-negative MT2 human T cell line stably transfected with CD38^Δ285, a human CD38 deletion mutant which lacks the 15 C-terminal amino acids and loses OKT10 binding 29. IF analysis demonstrates that the 15 C-terminal amino acids of human CD38 are required for binding of mAbs KK1B5, KK9H4 and OKT10 (Figure 5). In contrast, mAbs SUN-4B7 and AT1 maintained binding to MT2/CD38^Δ285 (data not shown). These data are consistent with the presence of two close conformational epitopes in human and cynomolgus CD38: one identified by mAbs KK1B5, KK9H4 and OKT10 located in the last (6^th) disulfide loop, the other by mAbs SUN-4B7 and AT1 mapping to the penultimate (5^th) C-terminal disulfide loop involving Cys²⁵⁴-Cys²⁷⁵ (Figure 6). Note that human-cynomolgus CD38 amino acid sequence identity in the 5^th loop is 20/22 amino acids, and 10/10 amino acids in the 6^th loop.

PBMC were isolated by Ficoll-Paque PLUS (Amersham Biosciences) centrifugation from whole blood obtained from a female cynomolgus macaque housed according to state regulations at the RBM (Ivrea, Italy). Cells were cultured in medium with 5 μg/ml PHA (Sigma-Aldrich) for 72 h, harvested and resuspended in TRIzol RNA extraction solution (Invitrogen). RT-PCR was carried out using the Titan One Tube RT-PCR system (Roche Diagnostics) with 150 ng total RNA and the gene-specific primers derived from the human CD38 sequence (GenBank accession no. D84278): forward 5'-AGT TTC AGA ACC CAG CCA-3' (corresponding to nt -84 to -66 upstream of the ATG initiation codon); reverse 5'-ATT GAC CTT ATT GTG GAG G-3' (corresponding to nt 102–121 downstream of the TGA stop codon). After RT for 50°C for 30 min, the cDNA was amplified by two rounds of PCR: 5 min at 94°C, followed by 30 (first round) or 35 (second round) cycles at 94°C for 30 s, 54°C for 30 s and 2 min at 72°C. The product was gel purified and cloned into the pGEM-T Easy vector (Promega). The inserts of three recombinant clones were analyzed by automated sequencing (Applied Biosystems).

Cynomolgus macaque (15 samples) and chimpanzee (5 samples) genomic DNAs were obtained as described 59. The eight exons of the cynomolgus CD38 gene were amplified by PCR (45 cycles, annealing temp 52°C) using primers known to amplify human CD38 exons 41 while the CD38 promoter of cynomolgus macaque was amplified (15 cycles annealing at 52°C, followed by 35 cycles with 0.3°C touchdown) with a forward primer from the human CD38 promoter (5'-GAA GAG GCA AGA AAA GCC-3') and reverse primer chosen from macaque CD38 exon 1 (5'-AACTCG CAG TTG GCC ATA-3'). The chimpanzee CD38 promoter was amplified with the human CD38 sequences. The 5' end of cynomolgus CD38 intron 1 was amplified (same conditions used for exon amplification but 57°C annealing and 1.5 M MgCl₂) with forward primer 5'-CCG TCC TGG CAC GAT GCG TCA AG-3' from macaque exon 1, and reverse primer 5'-ACA CCC TCC TCC CCT ACC ACA GG-3' taken from human CD38 intron 1. Amplicons were gel purified and analysed by automated sequencing. Alignments were performed with CLUSTALW (ExPASy, Swiss Institute of Bioinformatics).

NIH/3T3 murine fibroblast and COS-7 monkey kidney cell lines were from the ATCC. Production of SL-691 and SL-999 cynomolgus B lymphoblastoid cell lines was previously described 60. The mutant human CD38^Δ285 plasmid 48 was kindly provided by Dr. Toshiaki Katada (University of Tokyo, Japan) while the MT2^Δ285 transfectant was kindly provided by Dr. Umberto Dianzani (A. Avogadro University of Eastern Piedmont, Novara, Italy) 29. Cells were maintained in RPMI 1640 medium with 10% heat-inactivated FCS, penicillin/streptomycin. The murine anti-human CD38 mAbs AT1 (hybridoma kindly provided by Dr. Jo Hilgers, BioProbe AV, Amstelveen, The Netherlands), SUN-4B7, OKT10, IB4, IB6 and HB7 were produced in-house from hybridoma culture supernatants. CBT3G, a murine anti-human CD3 mAb, was used as IgG control. F(ab')₂ goat anti-mouse Ig-FITC was used as secondary antibody (Jackson ImmunoResearch Laboratories).

cDNAs were cloned into pcDNA3.1/V5-His-TOPO expression vector (Invitrogen). Stable transfected cell lines were produced in NIH/3T3 by electroporation (250 V, 960 μF at 20°C), selected for 3–4 weeks in G418 after which isolated clones were picked and transferred to 96-well plates.

Ecto-GDPR cyclase activity of intact cells was determined as previously described 293061. Briefly, parental and transfected NIH/3T3 were used at 2.5 or 5 × 10⁵ cells/ml in PBS. For RBCs, 420 μl packed volume were brought to 1 ml by addition of NGT buffer (0.15 M NaCl, 5 mM glucose, 10 mM Tris. Cl, pH 7.4). To 1 ml cell suspensions 10 μl 10 mM NGD (Sigma) in 20 mM Tris, pH 7.4 or 10 μl buffer (control) were added. After 30 min at 37°C, supernatants were collected after brief centrifugation. Supernatants were analysed by fluorescence spectrometer set at excitation wavelength 300 nm and emission wavelength 410 nm. Cell lines were measured in triplicate in three independent experiments; RBCs from two different animals were tested once; RBCs from humans were tested in duplicate in two independent experiments.

Cells were lysed in NP-40 buffer (150 mM NaCl, 1.0% NP-40, 50 mM Tris pH 8.0) containing protease inhibitors. Samples (20 μg protein/lane) were analysed by 8 or 10% SDS-PAGE and transferred to PVDF membranes (Bio-Rad Laboratories, Hercules, CA). Membranes were blocked in 5% milk/TBST, incubated for 2–3 h at RT with mAb supernatant with 1% milk, and incubated with horseradish peroxidase-labeled anti-mouse IgG (PerkinElmer) followed by ECL visualization.

BALB/c mice (Charles River Laboratories) were anaesthetized with Avertin i.p. injection and immunized by intrasplenic injection with 300 μl containing 5 × 10⁵ live NIH/mac38 cells in PBS. Eight days later, mice received an i.p. boost of NIH/mac38 cells, and mouse sera tested 8 days later. The spleen of one mouse that responded to immunization was selected for fusion to P3.X63.Ag8.653 murine myeloma cell line. Hybridoma supernatants were screened for reactivity with the immunizing cells and lack of reactivity with the parental cell line. Positive hybridomas were cloned by limiting dilution. MAbs were used in the form of supernatants containing NaN₃.

Surface expression of cynomolgus and human CD38 was determined by IF. Briefly, 2 × 10⁵ cells/sample were incubated with 100 μl mAb supernatant/NaN₃ for 1 h at 4°C, and 30 min at 4°C with FITC-labeled secondary Ab. Background fluorescence was established with control IgG antibody. Cells were analysed immediately either by fluorescence microscope, FACSCalibur flow cytometer (10,000 events acquired) with CellQuest software (Becton Dickinson) or Olympus FV3000 confocal microscope with Nomarski optics for differential interference contrast (DIC) and FluoView 300 software.

Cells were detached with 1 mM EDTA/PBS, washed and resuspended in complete medium at a concentration of 2 × 10⁶ cells/ml. One hundred μl 100 mM DTT stock was added per ml cell suspension, kept for 45 min in a 37°C incubator and washed twice with PBS/BSA/NaN₃ before analysis.