Physiological IgM Class Catalytic Antibodies Selective for Transthyretin Amyloid*

Background: Some antibodies express serine protease activity. Transthyretin misfolding causes accumulation of pathogenic amyloid. Results: Constitutively produced IgM but not IgG class antibodies selectively hydrolyzed and dissolved misfolded transthyretin without hydrolyzing physiologically folded transthyretin. Some IgMs were oligoreactive with amyloids and superantigens. Conclusion: Catalytic IgMs may clear misfolded TTR and delay amyloidosis. Significance: The innate antibody repertoire is a source of selective catabodies to toxic proteins. Peptide bond-hydrolyzing catalytic antibodies (catabodies) could degrade toxic proteins, but acquired immunity principles have not provided evidence for beneficial catabodies. Transthyretin (TTR) forms misfolded β-sheet aggregates responsible for age-associated amyloidosis. We describe nucleophilic catabodies from healthy humans without amyloidosis that degraded misfolded TTR (misTTR) without reactivity to the physiological tetrameric TTR (phyTTR). IgM class B cell receptors specifically recognized the electrophilic analog of misTTR but not phyTTR. IgM but not IgG class antibodies hydrolyzed the particulate and soluble misTTR species. No misTTR-IgM binding was detected. The IgMs accounted for essentially all of the misTTR hydrolytic activity of unfractionated human serum. The IgMs did not degrade non-amyloidogenic, non-superantigenic proteins. Individual monoclonal IgMs (mIgMs) expressed variable misTTR hydrolytic rates and differing oligoreactivity directed to amyloid β peptide and microbial superantigen proteins. A subset of the mIgMs was monoreactive for misTTR. Excess misTTR was dissolved by a hydrolytic mIgM. The studies reveal a novel antibody property, the innate ability of IgMs to selectively degrade and dissolve toxic misTTR species as a first line immune function.

Peptide bond-hydrolyzing catalytic antibodies (catabodies) could degrade toxic proteins, but acquired immunity principles have not provided evidence for beneficial catabodies. Transthyretin (TTR) forms misfolded ␤-sheet aggregates responsible for age-associated amyloidosis. We describe nucleophilic catabodies from healthy humans without amyloidosis that degraded misfolded TTR (misTTR) without reactivity to the physiological tetrameric TTR (phyTTR). IgM class B cell receptors specifically recognized the electrophilic analog of misTTR but not phyTTR. IgM but not IgG class antibodies hydrolyzed the particulate and soluble misTTR species. No misTTR-IgM binding was detected. The IgMs accounted for essentially all of the misTTR hydrolytic activity of unfractionated human serum. The IgMs did not degrade non-amyloidogenic, non-superantigenic proteins. Individual monoclonal IgMs (mIgMs) expressed variable misTTR hydrolytic rates and differing oligoreactivity directed to amyloid ␤ peptide and microbial superantigen proteins. A subset of the mIgMs was monoreactive for misTTR. Excess misTTR was dissolved by a hydrolytic mIgM. The studies reveal a novel antibody property, the innate ability of IgMs to selectively degrade and dissolve toxic misTTR species as a first line immune function.
Higher organisms possess a structurally diverse innate repertoire of antibody variable domains (V domains) 2 that was shaped by evolutionary selection pressures imposed by foreign antigens and self-antigens over millions of years. The diversity is supplemented by somatic randomization of the V domains by acquired immunity processes, which enables selection of V domains with noncovalent antigen binding activity over a few weeks upon contact with an antigen. Catalytic antibodies (catabodies) with a promiscuous peptide bond hydrolytic activity, on the other hand, are produced as an innate property of B cells, requiring no exposure to the antigen (1,2). The proteolytic activity was traced to serine protease-like sites in the germ line antibody V-regions, suggesting an ancient origin of the catabodies (3)(4)(5)(6). Humans produce catabodies selective for amyloid ␤ peptide (A␤) (7). Immune tolerance mechanisms in fetal and adult life are thought to suppress synthesis of autoantibodies to essential self-proteins. When produced, subsets of autoantibodies express autoantigen-selective catalytic activities (8 -10). Provided that catabodies discriminate between individual peptide antigens, they hold potential for specific and permanent destruction of toxic self-antigens.
Many proteins can misfold into the amyloid state (11). Soluble transthyretin (TTR) is found in large amounts in blood (ϳ0.3 mg/ml) in its physiological, homotetrameric state (phyTTR) as a carrier protein for thyroid hormones and the retinol-retinol binding protein complex (12). Misfolded ␤-sheet-containing aggregates are generated from TTR (collectively designated misTTR states; Table 1), a process involving formation of a TTR monomer, destabilization of the phyTTR dimer-dimer interface, and growth of soluble misTTR oligo-mers into fibrillar amyloid (13)(14)(15). Small misTTR amounts are probably formed continuously as basal byproducts of alternate TTR folding pathways. Initial evidence for the existence of blood-borne misTTR species is available (16). Dysregulated protein misfolding during normal aging processes causes increased TTR amyloid deposition in several tissues (12). TTR amyloid deposition in the tenosynovium is associated with carpal tunnel syndrome in middle-aged humans (17). TTR amyloidosis in the heart, lung, stomach, kidney, and other tissues causes organ failure in senile systemic amyloidosis (SSA), a disease affecting Ͼ10% humans over age 70 years (12, 18 -20). In addition, mutations in the TTR gene cause early onset familial TTR amyloidosis (21). Amyloid deposits cause architectural and functional tissue damage, and soluble misTTR aggregates exert direct cytotoxic effects. No approved pharmacological treatment for SSA is available, but anti-transthyretin RNAi reagents are undergoing clinical trials (22), and a small molecule drug that interacts with the thyroxine binding site of TTR may help relieve symptoms of polyneuropathy in familial TTR amyloidosis (23). Reversibly binding antibodies that can clear amyloids by phagocytosis of the immune complexes have been proposed as potential therapeutic reagents (24 -26). We report IgM class catabodies produced by healthy humans that hydrolyze misTTR but not phyTTR. The catabodies can be conceived as homeostatic mediators that destroy TTR amyloid and its precursor misfolded TTR forms without interfering in the physiological functions of TTR.

MATERIALS AND METHODS
Antibodies-Human studies were approved by the Committee for the Protection of Human Subjects, University of Texas Health Science Center (Houston, TX). Individual or pooled polyclonal IgM, IgG, and IgA class antibodies were purified from sera of 12 healthy humans without amyloidosis or autoimmune disease (33 Ϯ 7 years in age) by acid elution (pH 2.7) from columns of immobilized anti-human IgM antibodies, Protein G, and anti-human IgA antibodies (1,27). The antibody preparations were free of detectable non-antibody proteins judged by SDS-gel electrophoresis and immunoblotting with specific antibodies to IgM, IgG, and IgA. For evaluation of aging effects, IgMs were purified from non-aged humans (Ͻ35 years in age) or aged humans (Ͼ70 years) of either gender without amyloidosis or autoimmune disease. The monoclonal IgM (mIgM) panel purified from Waldenström macroglobulinemia patients was described (our laboratory codes 1800-1804, 1806, 1809-1811, 1813, 1814, 1816-1819, and Yvo) (2). Unfractionated human serum was pooled from 10 healthy humans (36 Ϯ 5 years in age). To prepare antibody-free serum, the pooled serum was adsorbed sequentially on the anti-IgM, Protein G, and anti-IgA columns (residual IgM, IgA, and IgG estimated by ELISA were 0.05, 0.007, and 0.003%, respectively) (28). FPLC gel filtration of the unfractionated serum pool (70 l) was on a Superose-6 column (GE Healthcare; flow rate, 0.1 ml/min) in 10 mM sodium phosphate, pH 7.4, 137 mM NaCl, 2.7 mM KCl (PBS) containing 0.1 mM CHAPS. The A␤-hydrolyzing recombinant catalytic antibody fragment (clone 2E6) was purified as described (29). Nominal molecular mass was computed by comparison with protein markers (a reference 900-kDa mIgM and 1.4 -670-kDa markers from Bio-Rad). Total protein was measured by the Micro BCA TM method (Thermo Fisher Scientific). Cell surface IgMs were analyzed using the purified peripheral blood B cells from a 25-year-old human subject without amyloidosis (B cell negative selection kit, Miltenyi (Auburn, CA); viability, 90 -95%; Ͼ95% purity determined by staining with phycoerythrin-conjugated mouse antibody to human CD19 (BD Pharmingen) and flow cytometry).
The phyTTR and misTTR states were dissociated into monomers (14 kDa) by boiling test samples (10 min) after mixing with a one-fifth volume of 10% SDS in electrophoresis The biotinylated, electrophilic analogs of misTTR (E-misTTR) and phyTTR (E-phyTTR) were prepared as described (33) (6.0 and 0.1 mol of phosphonate and biotin, respectively, per mol of misTTR; 3.9 and 1.5 mol of phosphonate groups and biotin, respectively, per mol of phyTTR). The starting material for E-misTTR preparation was the particulate misTTR in PBS/ CHAPS recovered by centrifugation (30 min, 75,000 ϫ g).
Proteolysis-All tests were done in duplicate or as specified. Non-aggregated or preaggregated 125 I-TTR (100 nM monomer equivalents) was incubated with purified antibodies in PBS/ CHAPS (20 l). The boiled or non-boiled reaction mixtures were analyzed by SDS-gel electrophoresis (8,930 -17,110 cpm/ lane). Depletion of the 1-mer TTR band (14 kDa) and appearance of lower mass products were quantified by densitometry. Antibody (Ab) hydrolytic activity was computed in units of nM misTTR hydrolyzed/g of antibody/h ((A nM ϫ (1 Ϫ [14 kDa Ab ]/[14 kDa Dil ]))/(g of Ab in the reaction mixture/Ab treatment time in h), where A is the initial misTTR concentration (total TTR concentration ϫ % misTTR content) if nonboiled samples were analyzed or the total TTR concentration (misTTR ϩ phyTTR) if boiled samples were analyzed, and [14 kDa Ab ] and [14 kDa Dil ] are the 14 kDa band intensities after antibody and diluent treatment, respectively). For data reported in this format, variations of the misTTR hydrolysis rate up to 2.4-fold are predicted because of varying misTTR content in different preaggregated TTR preparations (29 -70%).
For cell binding studies, large E-misTTR particles were removed by centrifugation (400 ϫ g, 10 min), and the E-misTTR did not contain flow cytometrically detectable particles. B cells were pretreated with mouse serum (1% in PBS, 10 min) to saturate Fc receptors and with the Endogenous Avidin/ Biotin Blocking kit to saturate cellular biotin as recommended by the manufacturer (Abcam, Cambridge, MA). The cells (1.5 ϫ 10 5 ) were incubated with E-phyTTR or E-misTTR (0.3 M) in PBS for 1 h at 4°C to minimize B cell receptor (BCR) internalization. Competition studies were done by preincubating B cells (1 h, 4°C) with control diluent or non-biotinylated misTTR, phyTTR, or ovalbumin (3 M). Bound biotinylated probes were detected with streptavidin conjugated to phycoerythrin-Cy7 (PECy7-SA; 8 g/ml). IgM ϩ cells were identified with antibody to human IgM conjugated to FITC (2 g/ml; BD Pharmingen). Flow cytometry was done using an LSRFortessa X-20 flow cytometer (BD Biosciences) and FlowJo software (about 10,000 cells/analysis). In control incubations, B cells were stained with FITC-labeled, isotype-matched antibody instead of the anti-IgM antibody or were treated with diluent instead of the biotinylated E-misTTR followed by staining with PECy7-SA. The E-misTTR binding cells were visualized by confocal microscopy as described (1) (multiple acquisitions, thickness 0.15 m; deconvolution by 15 iterations; FITC, ex ϭ 488 nm, em ϭ 525 nm; 4,6-diamino-2-phenylindole, ex ϭ 350 nm, em ϭ 470 nm; PECy7, ex ϭ 488 nm, em ϭ 767 nm).
Amyloid Dissolution-Preaggregated non-radioactive TTR (7.1 M, 100 g/ml) was incubated in duplicate with diluent or mIgMs (120 g/ml) in PBS containing 0.1 mM CHAPS at 37°C. The dissolution reaction was monitored from turbidity or ThT binding measurements.
Statistical Analysis-Errors are S.D. values. p values were from the unpaired two-tailed Student's t test. Correlations were determined by Pearson's two-tailed test.

125
I-TTR Substrate-Wild type human TTR radiolabeled with 125 I ( 125 I-TTR) was applied to determine hydrolysis of the phyTTR tetramers and misTTR aggregates. Non-aggregated 125 I-TTR preparations behaved as predicted from studies on phyTTR tetramers (30 -32). The non-aggregated 125 I-TTR (a) did not bind appreciably to ThT (Fig. 1A), a probe for ␤-sheet structures; (b) formed minimal particulate aggregates detected by turbidimetry (Fig. 1B); and (c) dissociated from the 4-mer into the 1-mer state upon boiling in SDS (14 kDa band in Fig. ). The 4-mer phyTTR species is dominant in non-boiled, non-aggregated TTR, whereas the 1-mer derived from soluble and insoluble misTTR is present in large amounts in the non-boiled, preaggregated TTR. misTTR content of the preaggregated and non-aggregated 125 I-TTR preparation, respectively, is 70 and Ͻ0.5%. D, TTR distribution in particulate and soluble fractions. Nearly equivalent radioactivity amounts from preaggregated 125 I-TTR were recovered in the pellet (particulate fraction) and supernatant (soluble fraction) obtained by centrifugation. E, misTTR distribution in particulate and soluble TTR fractions. The non-boiled particulate and soluble fractions prepared from preaggregated 125 I-TTR contained the radioactive 1-mer band derived from misTTR as the majority and minority species, respectively. Plotted values computed as 100 ϫ (1-mer band intensity/sum of 1-mer and 4-mer band intensities). Inset, lane 1 shows preaggregated 125 I-TTR without separation into particulate and soluble fractions (radioactivity in 1-mer and 4-mer bands, respectively, 43 and 57%). The 1-mer band is dominant in particulate 125 I-TTR (95% of total radioactivity; lane 2) and is also present in smaller amounts in soluble 125 I-TTR (27% of total radioactivity; lane 3). Error bars, S.D.
1C, lane 1; Ͻ1% tetramer content) but remained in the 4-mer 54-kDa state in non-boiled SDS solutions (Fig. 1C, lane 2; Ͻ1% 1-mer content). Incubation of 125 I-TTR at acid pH generated misTTR aggregates that bound ThT and formed a turbid suspension ( Fig. 1, A and B). Unlike the 4-mer phyTTR state, the misTTR aggregates were dissociated into 1-mers in SDS without prior boiling (Fig. 1C, compare lanes 2 and 4). The relative 4-mer/1-mer band intensity in non-boiled, preaggregated 125 I-TTR, therefore, is an index of the phyTTR/misTTR ratio (misTTR content of six preaggregated 125 I-TTR preparations from densitometry, 49 Ϯ 15%). The particulate fraction recovered from a preaggregated 125 I-TTR preparation (ϳ50% of total radioactivity; Fig. 1D) was composed almost exclusively of misTTR species (Ͼ95% 1-mers in Fig. 1E, lane 2). The 1-mers were also present at low levels in the supernatant (soluble fraction) of preaggregated but not nonaggregated TTR (compare lane 3, Fig. 1E with lane 2, Fig. 1C), consistent with the existence of soluble misTTR states (38,39). Small radioactivity amounts were found at positions corresponding to the TTR dimer (nominal mass 32 kDa) and unidentified impurities in the starting TTR (20 and 26 kDa). ThT fluorescence and turbidity measurements indicated comparable aggregation of 125 I-TTR and non-radiolabeled TTR, suggesting unaltered aggregation properties of the radiolabeled protein. The results show that the phyTTR and misTTR states are distinguishable by electrophoresis.
misTTR but Not phyTTR Hydrolysis by Polyclonal IgMs (pIgMs)-We tested the hydrolytic activity of electrophoretically homogeneous polyclonal IgM, IgG, and IgA pooled from the blood of humans without TTR amyloidosis. The pIgMs hydrolyzed preaggregated 125 I-TTR, indicated by depletion of the 1-mer TTR band (14 kDa) in SDS-treated, boiled reaction mixtures ( Fig. 2A). The hydrolytic activity was detected in every individual pIgM preparation from 12 healthy humans (Table 2) but not IgG or IgA from the same humans. Treatment of nonaggregated 125 I-TTR with the pIgMs did not deplete the 1-mer band derived from phyTTR in boiled reaction mixtures, indi- . From densitometry, 85% of the 1-mer band derived from misTTR was depleted by pIgM treatment. The 10and 7-kDa product fragments are observed. The 13-kDa product is not visible because of poor silver stainability. C, hydrolysis of soluble and particulate misTTR species by pIgM. The soluble and particulate fractions of preaggregated 125 I-TTR (misTTR content, 37 and Ͼ95%, respectively) were treated for 18 h with diluent or polyclonal IgM (pIgM) and IgG (pIgG) purified from the same human serum pool (130 g/ml). Non-boiled reaction mixtures were analyzed. D, misTTR selectivity. The 1-mer band (14 kDa) derived from misTTR but not the 4-mer phyTTR band (54 kDa) of preaggregated 125 I-TTR (100 nM; misTTR content, 40%) was depleted as a function of time by treatment with pooled pIgM (130 g/ml). Inset, 4-mer, 1-mer, and 10 kDa band cut-outs from electrophoresis gels; non-boiled reaction mixtures. Error bars, S.D.
cating that the pIgMs do not degrade phyTTR. Depletion of the 1-mer band in boiled reaction mixtures of the preaggregated substrate, therefore, measures misTTR hydrolysis without interference from phyTTR hydrolysis. The pIgMs also degraded preaggregated TTR without the 125 I label (Fig. 2B), eliminating radioactivity-related experimental artifacts. Selective recognition of soluble and particulate misTTR species by the IgM was confirmed in tests of non-boiled reaction mixtures. phyTTR does not dissociate into 1-mers without boiling, and the 1-mer band is derived exclusively from the misTTR species. The 1-mer band derived from misTTR but not the 4-mer phyTTR band was progressively depleted with increasing duration of pIgM treatment (Fig. 2, C and D). Product fragments with mass smaller than the 1-mer TTR band were evident (13 and 10 kDa bands, along with the fainter 7 kDa band; Fig. 2A). Accumulation of the 10-kDa product in the initial reaction phase and its subsequent disappearance with further pIgM treatment suggested reuse of this misTTR fragment for further digestion cycles (Fig. 2D). The appearance of three radioactive products implies hydrolysis of at least two peptide bonds. Because the radioactivity from a single intact TTR molecule is distributed into multiple products, the product bands are less intense than the intact TTR band. The assay method does not detect small products that migrate out of the gel or products that lack a radiolabeled Tyr residue.
Formation of IgM-misTTR immune complexes by the pooled pIgM (130 g/ml) was undetectable (binding of soluble and particulate misTTR species, respectively: 1.9 Ϯ 2.4 and 0.7 Ϯ 0.7%). At an equivalent concentration, the same pIgM preparation hydrolyzed 45 Ϯ 1% of soluble misTTR and 77 Ϯ 1% of particulate misTTR, respectively. The results indicate hydrolysis of misTTR but not phyTTR by IgMs, the first class of antibodies produced by B cells. In contrast, the IgGs and IgAs usually associated with acquired immune responses were devoid of misTTR hydrolytic activity.
Polyclonal IgM Hydrolytic Properties-Mean misTTR hydrolysis by pIgM preparations from humans Ͼ70 years in age was only slightly lower than the level from humans Ͻ35 years in age, and the difference did not reach statistical significance (p ϭ 0.101; Fig. 3A). misTTR hydrolysis was detected at low pIgM concentrations (10 g of IgM/ml, ϳ200-fold lower than the blood IgM concentration). The pIgM hydrolytic rate was comparable with MMP9 (Fig. 3B), an enzyme suggested to limit tissue deposition of TTR amyloid (36). The blood IgM concentration, however, exceeds that of MMP9 (3.2 nM) and other non-IgM proteases (Ͻ1 nM) by about 3 log orders (40,41). misTTR is generated from wild type phyTTR slowly at physio-logical pH (42), and saturation of the IgM by trace misTTR amounts in blood (16) is unlikely. From the observed rates, 50% degradation of 1 nM misTTR at the blood IgM and MMP9 concentrations is predicted to occur in 24 min and 28 days, respectively. Additional findings supported an important role of the IgM: (a) the hydrolytic activity of unfractionated human serum was lost nearly completely upon antibody removal by immunoadsorption (Fig. 3C); (b) the TTR product fragment profiles for purified pIgMs and unfractionated serum were identical (13, 10, and 7 kDa bands; Fig. 3C, inset); and (c) only the 900-kDa pIgM fraction separated from other serum components by FPLC gel filtration expressed hydrolytic activity (Fig. 3D).
pIgM-catalyzed misTTR hydrolysis was inhibited by synthetic electrophilic phosphonate 1, a compound that binds covalently to the nucleophilic sites of serine proteases (Fig. 3E). Synthetic inhibitors of metalloproteases, acid proteases, and cysteine proteases were ineffective. The inhibitor profile is consistent with previous findings that catabodies use a nucleophilic proteolytic mechanism (1,43). Non-antibody proteases are sensitive to endogenous enzyme inhibitors present in serum, and the enzymatic reaction may also be inhibited by amyloid binding proteins (e.g. SAP) (36). Including antibody-free human serum or purified SAP in the reaction mixtures did not reduce the misTTR hydrolytic activity of pIgM (compare with control albumin treatment; Fig. 3E). Moreover, because misTTR was hydrolyzed by unfractionated IgMs without removal of any serum components, the IgMs are insensitive to serum protease inhibitors. We cannot exclude the possibility of a serum cofactor that enhances IgM catalysis, because the hydrolytic activity of unfractionated serum per unit IgM mass exceeded that of pure pIgM from the same serum by 4.5-fold (7.2 and 1.6 nM/g of IgM/h, respectively; this is a valid comparison because nearly all of the serum hydrolytic activity is due to the IgMs). Catabody-activating cofactors were described previously (44,45). Alternatively, because catabodies are sensitive to conformational perturbations (29), the acid elution procedure employed for IgM purification may attenuate the hydrolytic activity.
The pIgM hydrolytic activity was saturable with increasing misTTR concentration (Table 3). In addition to misTTR, the pIgMs hydrolyzed small peptide substrates, a reaction that reports the catalytic subsite properties independent of noncovalent epitope binding (1,46). The apparent K m for misTTR was 92-fold lower than that of EAR-AMC, a small peptide hydrolyzed by IgMs at the Arg-AMC bond, suggesting selective IgM-misTTR recognition (K m approximates K d , the equilibrium dissociation constant). Diffusional restrictions known to decelerate enzymatic conversion of particulate substrates compared with soluble substrates (47,48) probably limit the rate of particulate misTTR hydrolysis by the IgMs. The pIgM preparations are mixtures of individual IgM subsets, and misTTR is a mixture of multiple misTTR aggregation states. The apparent kinetic constants, therefore, are average values for hydrolysis of multiple misTTR species by multiple IgMs.
misTTR-directed B Cells and mIgMs-The reactivity of mIgMs expressed as BCRs in healthy humans often indicates initial antibody specificity prior to contact with external

-Mer TTR hydrolytic activity of IgMs
IgM preparations with hydrolytic activity Ͼ0.1 nM/h/g were designated "positive." Polyclonal IgMs were from non-aged humans (Ͻ35 years). The hydrolytic activities of IgMs from male and female humans were statistically indistinguishable (n ϭ 9 and 3, respectively; p Ͼ 0.05). Monoclonal IgMs were from patients with Waldenström macroglobulinemia.

No. of IgMs
Hydrolytic activity immunogens. The E-misTTR analog contains electrophilic phosphonates that bind covalently to nucleophilic BCR sites responsible for catalysis, permitting identification of misTTRselective B cells without loss of the bound probe due to peptide-bound hydrolysis. Flow cytometry identified a discrete IgM ϩ B cell population with saturable binding activity for E-misTTR that was inhibited by competitor misTTR but not the irrelevant protein ovalbumin (Fig. 4A). The intensity of E-misTTR staining differed over 2 log orders, suggesting the variable E-misTTR reactivity of individual B cells. Saturable E-phyTTR binding by the B cells was negligible, indicating selective recognition of the E-misTTR probe (Fig. 4B). Most of the E-misTTR binding sites were colocalized with cell surface IgM, consistent with BCR recognition of the misTTR (Fig. 4C, image 3 is a merged rendition). A panel of mIgMs secreted by cancerous B cells from patients with Waldenström macroglobulinemia without TTR amyloidosis was applied as a model for individual IgMs present in where S 0 is substrate concentration at time 0, E 0 is the enzyme concentration at time 0, k is the second order rate constant derived from the equilibrium binding and catalytic constants, and t is reaction time. Inset, SDS gel showing reduced 1-mer misTTR band following MMP9 treatment (lane 2) compared with diluent treatment (lane 1). C, pIgM immunoadsorption. misTTR hydrolysis by unfractionated human serum and Ab-free serum was measured as in A. The protein content of unfractionated serum (diluted 1:100) and the antibody-free serum was adjusted to be identical (1.14 mg/ml). IgM concentrations in hydrolysis assays using unfractionated serum and antibody-free serum were 20 and Ͻ0.01 g/ml, respectively. Data are expressed as percentage depletion of the 1-mer misTTR band compared with the control substrate reaction mixture treated with diluent (100% value ϭ 22.5 Ϯ 5.6 nM). Inset, electrophoresis showed identical product fragments (13, 10, and 7 kDa bands) generated by digesting preaggregated 125 I-TTR with unfractionated serum (lane 3, 1:100 dilution) and purified IgM (lane 4, 120 g/ml). Lane 1, substrate treated with diluent. Lane 2, substrate treated with antibody-free serum. D, distribution of hydrolytic activity of unfractionated serum. Only the IgM-containing 900-kDa fraction (retention volume 8 -11 ml) obtained by FPLC gel filtration of human serum (blue trace) displayed detectable misTTR hydrolytic activity assayed as in A (red bars). IgM concentration in the assay, 20 g/ml (equivalent to IgM concentration in 1% unfractionated serum). Elution of reference purified IgM (black trace) is coincident with elution of the serum hydrolytic activity. Non-IgM serum fractions were tested as two pools at a concentration corresponding to 1% unfractionated serum equivalents (retention volume 0 -8.0 and 11.0 -22.0 ml; concentrated using a 3-kDa ultrafilter). Inset, reducing SDS gels of 900-kDa serum IgM fraction showing anti-antibody-stained 70-kDa heavy chain band (lane 1) and anti-/ antibody-stained 25-kDa light chain band (lane 2). Lanes 3 and 4, respectively, are similarly stained heavy and light chain bands of the reference IgM, respectively. H, heavy chain; L, light chain. E, protease inhibitor effects. misTTR hydrolysis was measured using IgM pretreated (1 h) with diluent or the indicated inhibitors of metalloproteases, cysteine proteases, acid proteases, and serine proteases. The structure of phosphonate 1 is shown. SAP is an amyloid-binding protein. Residual hydrolytic activity of inhibitor-treated IgMs was computed as a percentage of IgM activity after treatment with diluent (100% value ϭ 24 Ϯ 4 nM TTR). Hydrolysis was measured using boiled reaction mixtures of 100 nM preaggregated 125 I-TTR incubated with 60 g IgM/ml for 18 h. F, unimpeded hydrolysis of preaggregated 125 I-TTR by pIgM in the presence of Ab-free serum. Plotted are 1-mer misTTR hydrolytic activities expressed as a percentage of the activity of purified (130 g/ml) alone (90 Ϯ 3 nM). Reaction conditions were as in A. Error bars, S.D.  Table 2). The hydrolytic rates were highly variable (undetectable to 2.98 nM/g of IgM/h), consistent with the prediction of differing catalysis by individual catabodies with structurally distinct V domains. Like the pIgMs, the 12 misTTR-hydrolyzing mIgMs hydrolyzed misTTR but not phyTTR, shown by (a) the depleted 1-mer band of preaggregated 125 I-TTR but not non-aggregated 125 I-TTR in boiled reaction mixtures and (b) the depleted 1-mer TTR band originating from misTTR and the undepleted 4-mer phyTTR band in non-boiled reaction mixtures of pre-aggregated 125 I-TTR (Fig. 4, D and E, shows the example of mIgM 1802). Moreover, the mIgMs generated the same 13-, 10-, and 7-kDa TTR product fragments observed for pIgMs (Fig. 4C, inset). The consis-tent selectivity and product fragment profiles suggest a shared structural basis for mIgM and pIgM catalysis. From mutagenesis (43), epitope mapping (49), and protease inhibitor studies (50), we proposed the split site catalysis model (Fig. 5A), involving discrimination between structurally distinct antigen epitopes by the noncovalent binding site prior to peptide bond hydrolysis at the catalytic subsite. At saturating concentrations, the short peptide EAR-AMC substrate is hydrolyzed detectably by the catalytic subsite, a test for catalysis occurring without selective epitope recognition (1,46). As reported (7), titration of mIgM Yvo with the phosphonate 1 inhibitor indicated 10.2 EAR-AMC hydrolytic sites/IgM molecule (Fig. 5B, inset; theoretical number of catalytic sites/IgM, 10). When tested as an alternate substrate, EAR-AMC inhibited the misTTR hydrolytic activity of mIgM Yvo (Fig. 5B). Ala-Ala-Ala-AMC (AAA-AMC), a peptide that is not hydrolyzed by the mIgM, did not inhibit the misTTR hydrolytic activity. It may be concluded that misTTR and EAR-AMC hydrolysis occur at the same mIgM catalytic site. In addition, EAR-AMC and misTTR hydrolysis by the panel of 16 mIgMs was significantly correlated (p ϭ 0.005; Fig. 5C), consistent with regulation of misTTR 3 The panel is assumed to be a random collection of mIgMs. There is no known relationship between B cell carcinogenicity and IgM catalytic activity or antigenic specificity. Studies to be published elsewhere confirmed secre-  hydrolysis by the catalytic subsite efficiency. Importantly, the concentration of misTTR required for detection of hydrolysis was ϳ3 log orders lower than the EAR-AMC substrate, consistent with selective misTTR recognition. Certain mIgMs with low EAR-AMC hydrolytic activity hydrolyzed misTTR robustly (e.g. IgMs 1811 and 1814 falling outside the regression confidence limits in Fig. 5C), suggesting that hydrolysis can be accelerated due to selective misTTR recognition.
Are the Catabodies Polyreactive?-The complementarity-determining regions (CDRs) of some polyreactive IgMs produced without immunogen stimulation can bind epitopes with no discernible structural similarity (51), a phenomenon that may originate from CDR flexibility and the induced fit binding mechanism. Consistent with previous reports (2,7), the pIgMs and mIgMs did not hydrolyze non-amyloidogenic, non-superantigen proteins from eukaryotic and prokaryotic organisms ( Fig. 6A; no hydrolysis of albumin, extracellular epidermal growth factor receptor fragment, ovalbumin, transferrin, collagen adhesion protein, or LukS). Specific catalysis requires spatial coordination between the catabody noncovalent and catalytic subsites (2). The lack of indiscriminate pIgM proteolysis may be explained by uncoordinated CDR-based, polyreactive binding activity and catalytic subsite activity.
misTTR Dissolution-Treatment of a 10-fold molar excess of preaggregated TTR with the hydrolytic mIgM 1802 but not non-hydrolytic IgM 1819 resulted in time-dependent reduction of turbidity to the nearly basal value of non-aggregated TTR (Fig. 7A). ThT binding by preaggregated TTR, an indicator of ␤-sheet content, was also reduced to the nearly basal value by the hydrolytic but not the non-hydrolytic mIgM (Fig. 7B). The observed dissolution rate is diffusion-restricted by the particulate substrate character, and the dissolution process probably proceeds through a series of hydrolytic reactions, eventually converting particulate misTTR to soluble products. The mIgMs (130 g/ml) did not form detectable immune complexes (percentage of soluble and particulate 125 I-misTTR binding: hydrolytic mIgM 1802, 1.9 Ϯ 1.6 and 0.3 Ϯ 0.1%, respectively; non-hydrolytic mIgM 1819, 1.0 Ϯ 2.6 and 1.6 Ϯ 1.6%, respectively). Because the hydrolytic mIgM 1802 has no misTTR binding activity and, except for the V domains, is structurally identical to the non-hydrolytic mIgM, the dissolution reaction must be due to misTTR digestion.

DISCUSSION
Small misTTR amounts are probably produced as normal byproducts of alternate TTR folding pathways. With advancing age, such misTTR species are thought to serve as seeds for pathological TTR amyloid formation. Our findings suggest physiological production of misTTR-selective IgM catabodies as a first line defense function against TTR amyloidosis. The IgMs hydrolyzed soluble and particulate misTTR but not the nonpathogenic phyTTR tetramer found at micromolar concentrations in blood, suggesting selectivity for an aggregation-induced misTTR neoepitope that is absent in phyTTR. The IgMs, therefore, can degrade misTTR with no danger of saturation by the more abundant phyTTR species. There was no evidence for IgM degradation of non-amyloid, non-superantigen proteins (2,7). Two misTTR-hydrolyzing mIgMs did not hydrolyze A␤ or microbial superantigens, and other mIgMs recognized these substrates at varying levels. Saturation of oligoreactive IgMs with irrelevant amyloids may limit the misTTR clearance capacity, but this is a threat only in patients with non-TTR amyloid disease. Likewise, superantigens accumulate only during microbial infections, and impairment of misTTR clearance is not predicted in the non-infected state. The selectivity properties, therefore, support a role for the IgMs in defense against TTR amyloidosis.
Further evidence for the beneficial IgM role includes the following: (a) preaggregated TTR was dissolved completely by a hydrolytic mIgM; (b) misTTR was degraded at IgM concentrations ϳ200-fold lower than the blood IgM concentration, and virtually the entire misTTR degradative capacity of serum was attributable to IgMs; (c) the blood IgM concentration exceeds that of non-antibody proteases by several log orders, and pIgMs hydrolyzed misTTR at a rate comparable with that of a nonantibody protease (MMP9); (d) the pIgM hydrolytic activity was refractory to serum inhibitors of non-antibody proteases; and (e) pIgM from different healthy adults hydrolyzed misTTR within a narrow rate range, suggesting stable maintenance of the misTTR hydrolytic activity under disease-free conditions. The beneficial IgM function may extend beyond clearance of blood-borne misTTR. IgMs permeate most anatomic bloodtissue barriers. Lymph from many tissues, a common surrogate for the tissue interstitial fluid, contains IgM in the mg/ml range (59 -61), indicating the feasibility of tissue misTTR clearance.
Notwithstanding the favorable catabody properties, pathological TTR amyloid accumulates in SSA patients, indicating that the catabody clearance capacity may be overwhelmed by independent age-associated factors that accelerate protein misfolding (e.g. the reaction of amyloidogenic proteins with advanced glycation and lipid peroxidation end products) (62,63). In addition, we cannot exclude reduced catabody formation as a possible factor in TTR amyloidosis. Age-associated shifts in the structural and functional properties of antibody V domains are documented (64,65), individual mIgMs hydro-lyzed misTTR at widely variable rates, and aged humans display increased susceptibility to infections (66). We observed a small, non-significant tendency toward reduced misTTR hydrolysis in healthy, aged humans. Future studies on patients with SSA are required to assess whether immune senescence is a predisposing factor in amyloidosis. In addition, long term follow-up of humans and mice with genetic antibody deficiencies (67,68) may be useful to define the contribution of the immune system in defense against amyloidosis. The catalytic function may also be conceived to diminish the inflammatory consequences of antibody-antigen interactions. Reversible binding at the Fab region of conventional antibodies stimulates Fc region-mediated activation of complement and phagocytosis, exemplified by the inflammatory effects of A␤-binding antibodies (69). Rapid catabodies form only transient immune complexes and then release the digested antigen products, reducing the probability of harmful immune complex effects.
According to the acquired immunity paradigm, IgGs with specific binding activity directed to individual target epitopes are produced within a few weeks by the immunogen-driven maturation of antibody light and heavy chain variable domain sequences (V L and V H domains, respectively). Acquired immunity rules did not yield catabodies capable of complex peptide bond hydrolytic reactions (70). Moreover, the rules provide no basis to anticipate the strict catabody selectivity for misTTR. Because misTTR-directed IgMs are produced by healthy humans without TTR amyloidosis, they are probably produced constitutively without a requirement for acquired immunity processes. Catabody synthesis can occur without immunogendriven V domain maturation because the serine protease-like nucleophilic sites are encoded by heritable germ line V region genes (3)(4)(5)(6). A germ line V L gene residue is an essential constituent of the misTTR-hydrolyzing mIgM Yvo catalytic triad (Ser 110 , IMGT numbering), and the two remaining catalytic residues are evidently acquired during V-(D)-J gene recombination, which occurs prior to contact with the immunogen (Arg 112 and Glu 113 at the V L -J and V H D-J junctions, respectively (71)). Furthermore, the IgMs are the first antibody class produced by B cells, whereas IgGs generated at the terminal acquired immunity stage are poorly hydrolytic. The antibody C domain scaffold can influence antigen binding at the remote V domains (72). C domain swapping studies indicated loss of catalytic activity at the IgM 3 IgG class switch step due to the unfavorable effect of the IgG C domains on the V domain catalytic site (28). Detailed information about factors regulating IgM constitutive synthesis is not available. However, IgMs are synthesized at stable levels by germ-free animals (73), and their constitutive production may involve an autonomous B cell developmental program and BCR-independent lymphocyte growth stimulators (74).
In adult humans, IgGs are the dominant blood-borne antibodies, found at 5-6-fold greater concentrations than IgMs. Many IgGs express specific antigen binding activity, consistent with maturation of their V domains by an immunogen-driven somatic mutation process. Such antigen-specific IgGs are the basis of protective responses induced by microbes and vaccines, and numerous antigen-binding monoclonal IgGs are approved for therapeutic use (75). Catalytic IgG preparations from patients with autoimmune disease are reported to hydrolyze self-antigens (8, 76 -78), but dysfunctional B cell reactivity in autoimmune disease may contribute to increased catalytic IgG synthesis (8), and the hydrolytic rate of the IgGs is usually lower than that of IgMs (79,80). Monoclonal IgGs from routine immunization protocols display minimal or no catalytic activity, but their dissociated light subunit frequently hydrolyzes peptide bonds (81)(82)(83). Other IgGs generated by healthy humans express low affinity, polyreactive antigen binding profiles (often designated "natural" IgGs) that may originate from the occurrence of class switch recombination in bystander B cells without sufficient mutations in the V domains required to accomplish specific antigen binding (84,85). The polyreactivity was suggested to contribute in the therapeutic effect of pooled IgGs from healthy humans in patients with bacterial infections and certain autoimmune diseases (commonly known as intravenous immune globulin or IVIG preparations) (86). The catalytic activity of such IVIG preparations is marginal compared with IgMs (27). Also, the IgMs in the present study display catalytic specificities that are readily distinguishable from the polyreactive IgG binding pattern.
Protein-protein recognition is dictated by sequence-dependent chemical interactions, topographical shape complementarity, and local flexibility of the reacting sites. The amyloidsuperantigen link is obvious from mIgMs with oligoreactivity for both protein classes (e.g. mIgM Yvo). Comparison of linear sequences did not indicate a homologous epitope that is shared by TTR, A␤, Protein A, and gp120 (Fig. 8A). Our innate catabody model conceives an initial selectivity-conferring binding step that places the substrate peptide backbone in spatial register with the catabody hydrolytic subsite (8,43). The oligoreac-tivity may originate from substrate interactions at a single cross-reactive catabody site or multiple catabody sites with unique ligand preferences. Antibody V domains are described to express multiple ligand-interacting sites outside the classical CDR-based antigen binding pocket (87). For instance, the framework region-based V H domain sites that recognize Protein A and gp120 are not identical (54,55), and yet another V L domain site recognizes the superantigen Protein L (88). Because misTTR hydrolysis by mIgM Yvo was not inhibited competitively by Protein A and gp120, non-identical sites are probably involved in hydrolysis of misTTR versus superantigens. In contrast, A␤ inhibited the hydrolysis of misTTR, suggesting amyloid substrate recognition at a shared IgM site. A precedent for the TTR/A␤ cross-reactive site is available; reversibly binding antibodies can recognize the ␤-strand conformational epitope found in amyloid proteins with differing sequence (52,53).
Humans produce no more than 100,000 non-antibody proteins, including all known enzymes and receptors with specificity for a broad range of ligands. The recombined IgM repertoire derived from the germ line V, D, and J segment configurations is far larger (from germ line gene sequences available in IgBLAST, ϳ4 ϫ 10 9 V L -V H domain pairs derived by recombination of 152 V L with 19 J genes and recombination of 273 V H genes with 34 D and 13 J genes; this estimate excludes further repertoire expansion occurring during the junctional diversification process). We suggest the human germ line IgM repertoire as a source of catabodies with functionally useful selectivity for individual substrates developed by Darwinian immune evolution (Fig. 8B). Like the misTTR-selective catabodies, the innate character of catabodies to A␤ is evident from superior A␤ hydrolysis by IgMs compared with IgGs and by production of A␤ binding and hydrolyzing antibodies by healthy humans without amyloidosis (7,89). Jawed fish, the first extant organisms with an immune system orthologous to humans, produce TTR and A␤ with sequences similar to their human counterparts (Fig. 8C). Because tissue amyloid deposition has no known benefit in eukaryotes, the catabodies may have evolved to retard amyloidosis. Amyloids of 27 proteins can accumulate to pathogenic levels, causing premature senescence and disease (11). Proteins accumulate in the amyloid state within minutes to days in vitro, suggesting the existence of functionally important mechanisms that maintain amyloids at non-toxic levels in vivo. Amyloidosis occurring prior to reproduction will jeopardize survival of the species. Increased misfolded protein levels can be detected years prior to the appearance of disease symptoms (e.g. toxic A␤ oligomers in middle-aged humans (90) and particulate A␤ deposits in reproducing monkeys (91)). Moreover, young humans are susceptible to amyloidosis in conjunction with cardiovascular disease, infections, cancer, and exposure to environmental toxins (92-94) (e.g. atrial natriuretic peptide amyloidosis in congestive heart failure (95)). These arguments support the potential role of innate amyloid-clearing catabodies in furnishing an evolutionary survival advantage. Catabodies to the superantigens HIV gp120 (2, 96), S. aureus Efb (97), and S. aureus Protein A (present report) may serve as innate defense mediators against microbes. Superantigens, however, downregulate acquired B-and T-cell immunity, and the survival advantage for the microbe gained by evading host acquired immunity (56) may trump the protective function of anti-superantigen catabodies.
In summary, our findings indicate the physiological production of misTTR-selective catabodies. Unlike other approaches to suppression of TTR amyloidosis, the catabodies remove misTTR without interfering in the essential functions of the properly folded TTR state. The misTTR-hydrolyzing IgMs, therefore, hold potential for developing a safe treatment for TTR amyloidosis. Several monoclonal IgMs have been approved in the United States for human tests, and pentamer IgM-containing polyclonal antibodies are approved for human therapy in Europe (98). Converting pentamer IgM to the monomer IgM state does not compromise the catalytic activity (1), and monomer IgMs can be considered as candidate therapeutic reagents. Small amounts of IgM monomers are found physio-logically in humans (99), and lower organisms produce larger monomer IgM amounts (100). In addition, highly catalytic single V L domains and paired V L domains can be engineered (8,29) as candidate drugs from the misTTR-hydrolyzing IgMs.