PATHOGENESIS: LAYING THE GROUNDWORK FOR DISCOVERY

Antibodies to SSA/Ro ribonucleoproteins in the maternal sera, often in association with SSB/La, have been almost universally demonstrated when congenital heart block (CHB) develops in utero in the absence of structural abnormalities.^1–4 This model of passively acquired autoimmunity offers an exceptional opportunity to examine the effector arm of immunity and to define the pathogenicity of an autoantibody in mediating tissue injury. A molecular scenario in which maternal anti-SSA/Ro-SSB/La antibodies convincingly contribute to the pathogenesis of cardiac scarring has yet to be formulated. One difficulty in identification of a pathogenic effect of an autoantibody is accounting for the heterogeneity of that effect. CHB is a stellar example in that not only is the injury seemingly rare but the degree of injury varied, with the spectrum inclusive of clinically inconsequential first-degree block as well as third-degree block and an associated cardiomyopathy that is often fatal.

The necessity of anti-SSA/Ro-SSB/La antibodies is supported by their presence in >85% of mothers whose fetuses are identified with conduction abnormalities in a structurally normal heart.³ However, when Brucato and colleagues⁵ prospectively evaluated 118 pregnancies in 100 patients with anti-SSA/Ro antibodies, the frequency of CHB in a fetus was only 1.7%. Although recurrence rates exceed the approximate 2% risk for a mother who has never had an affected child by five- to tenfold, the risk is not 100%. Moreover, the concordance rate in genetically identical twins is also not 100%. Accordingly, it is likely that antibody specificity alone cannot account for cardiac injury and that fetal factor(s) and/or the in utero environment must amplify the effects of the antibody, which may be necessary but insufficient to cause disease. Notably, one mother in the series reported by Brucato and colleagues⁵ (who gave birth to two healthy children) developed complete heart block herself, raising the possibility that her heart had acquired the amplifying “fetal factors.” Clearly, this is a unique situation and one that needs to be further studied because it is likely to contribute important clues to pathogenesis. A direct pathologic consequence to cells by inhibiting function (as in neonatal myasthenia gravis,⁶ or Type II cytotoxic reactivity, as in hemolytic disease of the newborn⁷) would also predict a much higher recurrence rate of CHB in subsequent pregnancies than the observed recurrence rates of 18 to 20%.^8,9

Another challenging aspect of the pathogenicity of this disease is that the candidate target antigens are normally sequestered intracellularly. This suggests several possibilities. First, the proposed target is not correct. Second, there is a cross-reactivity of the true target with an antigen normally found on the cardiac surface. Third, the target becomes available to maternal antibody following a change in the cell (which results in translocation to the membrane).

The sections that follow address several of the challenges posed by trying to fit the clinical clues with pathogenesis. The majority of the clinical information and patient samples are derived from the Research Registry for Neonatal Lupus (RRNL), established in September of 1994.⁸

TARGET AUTOANTIGENS OF THE SSA/Ro-SSB/La SYSTEM: A RECENT EMPHASIS ON Ro52

Antibodies to the 52kD SSA/Ro protein are found in >75% of mothers whose children have CHB.^10–13 Initial epitope mapping of this response revealed an immunodominant region spanning aa169-291 containing the leucine zipper (which was recognized by the majority of the CHB-sera), frequently in the context of HLA-DRB1*0301, DQA1*0501, and DQB1*0201.¹¹ The finer specificity of the anti-Ro52 response has been confirmed and extended with current focus on aa200-239 (p200).¹³ In an initial evaluation consisting of 9 CHB mothers and 26 anti-SSA/Ro positive mothers of healthy children, antibodies to p200 predicted CHB with greater certainty than currently available testing for either 60kD or 52kD SSA/Ro.¹³ Recent studies integrating an in vivo rodent model and in vitro culturing system suggest that anti-p200 antibodies bind neonatal rodent cardiocytes and alter calcium homeostasis.¹⁴

To address both the clinical necessity and sufficiency of this newly identified p200 reactivity in the development of CHB, as well as the reduced risk of CHB reportedly associated with aa176-196 (p176) and aa197-232 (p197),¹³ maternal sera were evaluated from The Research Registry for Neonatal Lupus (RRNL)⁸ and from the prospective study PR Interval and Dexamethasone Evaluation (PRIDE) in CHB.¹⁵ In addition, the PRIDE study provided the opportunity to address whether the level of anti-p200 antibodies positively correlated with length of the Doppler mechanical PR interval (>150 msec corresponds to first-degree block).

The majority of the 156 Ro52-positive sera tested were reactive with _p200 (>3 SD above control), irrespective of clinical status of the child.¹⁶ Mean OD values of p200 did not differ significantly among mothers of children with CHB (0.187 ± 0.363 SD), rash (0.176 ± 0.356 SD), or no manifestation of NL (0.229 ± 0.315 SD). p200 reactivity was found in 80/104 (77%) CHB mothers, 24/30 (80%) rash mothers, and 21/22 (95%) mothers who delivered healthy children and had no previous children with NL (P = NS for all comparisons). Sera from 4 CHB mothers with varied p200 titers (range OD 0.025 to 1.818) bound the surface of nonpermeabilized apoptotic but not healthy human fetal cardiocytes. These observations suggested that antibodies to p200, equivalent to antibodies to full-length SSA/Ro, do not bind the surface of fetal cardiocytes unless those cells become apoptotic (p200 is translocated to the membrane).

For 32 Ro52-positive women completing the PRIDE study (22 no previous child with NL, 7 previous child with CHB, 3 previous child with rash) in whom p200 levels were determined during pregnancy, the correlation between level of p200 (OD range 0.000 to 1.170) and maximal fetal PR interval (range 115 to 168 msec) was not significant (Spearman R = 0.107, P = 0.58).

Our interpretation of these data is that reactivity to p200 is a dominant but not uniform anti-Ro52 response in women whose children have CHB. Because exposure to this antibody specificity was observed with a similar frequency in children without CHB born to mothers with anti-Ro52, additional factors are necessary to convert risk to disease expression.

Eftekhari and colleagues recently reported that antibodies reactive with the serotoninergic 5-hydroxytryptamine (5-HT)_4A receptor, cloned from human adult atrium, also bind 52kD SSA/Ro.¹⁷ Moreover, affinity-purified 5-HT₄ antibodies antagonized the serotonin-induced L-type Ca channel activation in human atrial cells. Two peptides in the C terminus of 52kD SSA/Ro, aa365–382, and aa380–396 were identified that shared some similarity with the 5-HT₄ receptor. The former was recognized by sera from mothers of children with NLS, and was reported to be cross-reactive with peptide aa165-185, derived from the second extracellular loop of the 5-HT₄ receptor. These findings are of particular importance because >75% of sera from mothers whose children have CHB contain antibodies to 52kD SSA/Ro as detected by ELISA, immunoblot, and immunoprecipitation.^12,18

Given the intriguing possibility that antibodies to the 5-HT₄ receptor might represent the hitherto elusive reactivity directly contributing to AV block, we examined whether the 5-HT₄ receptor is a target of the immune response in these mothers.¹⁹ Initial experiments demonstrated mRNA expression of the 5-HT₄ receptor in the human fetal atrium. Electrophysiologic studies established that human fetal atrial cells express functional 5-HT₄ receptors. Sera from 116 mothers enrolled in the RRNL, whose children have CHB, were evaluated: 99 (85%) contained antibodies to SSA/Ro, 84% of which were reactive with the 52kD SSA/Ro component by immunoblot. In sum, none of the 116 sera were reactive with the peptide spanning aa165-185 of the serotoninergic receptor. Rabbit antisera that recognized this peptide did not react with 52kD SSA/Ro. Accordingly, although 5-HT₄ receptors are present and functional in the human fetal heart maternal antibodies to the 5-HT₄ receptor are not necessary for the development of CHB.

Most recently, Eftekhari’s group and ours jointly assessed the role of anti-5HT4 antibodies.²⁰ Sera from 101 anti-SSA/Ro52-positive mothers (of whom 74 had children with CHB and 27 had children without heart block), 8 anti-Ro52-negative mothers who had other anti-Ro/La reactivity and children without heart block, and 18 healthy anti-Ro/La-negative donors were assessed in a single blind test using an ELISA coated with a 5-HT₄ receptor-derived peptide. Also tested were 12 anti-Ro/La-negative mothers, of whom 1 had a child with CHB, 5 had children with structural heart block, and 6 had children who developed heart block after birth. Discrepancies between previous observations in our two groups could be ascribed to small differences in the setup of the assay.

Of the 74 sera from Ro52⁺ mothers of children with CHB, 11 were reactive with the 5-HT₄ peptide. Sera from the Ro52⁻ mother of a child with CHB, one of 6 Ro52⁻ mothers of children with structural HB, 3 of 35 Ro/La⁺ mothers of unaffected children, and 2 of 18 Ro52⁻ controls were also 5-HT₄-positive. Although 5-HT₄ receptor autoantibodies do not have the predictive value of anti-Ro52 autoantibodies, the presence of these antibodies in a minor subset of mothers whose children have CHB suggests an additional risk factor that may contribute to the pathogenesis of disease.

CARDIAC HISTOPATHOLOGY

Histopathologic studies constitute a major basis for formulating hypotheses regarding the pathogenesis of CHB. It appears logical to assume that the time of death relative to initial immune attack may influence the pathologic findings. Evidence of a cellular infiltrate might be present if death occurs close to the time a bradyarrhythmia is first detected, but calcifications and fibrosis may be the sole pathologic finding if death has occurred months later. However, based on data generated in our laboratory (see material following) the inflammatory phase seems to be rather evanescent (in that we have seen extensive fibrosis in a fetus electively terminated almost immediately following the in utero diagnosis).²¹ Moreover, based on serial echocardiograms a fetus can progress from normal sinus rhythm to complete block in a week.²² Although published literature on serial echocardiograms in mothers at high risk of a pregnancy complicated by CHB is limited (currently being addressed by the PRIDE study,¹⁵ discussed previously), it has been the general experience that the first clinically apparent abnormality in cardiac function is bradycardia, and only very rarely myocarditis (i.e., effusions, ventricular dysfunction, and so on). This implies that early inflammation is not clinically detectable and/or that atrioventricular (AV) nodal injury occurs independently of an inflammatory pancarditis.

Specific vulnerability of the conducting system is unexplained.²³ Ho and colleagues described the histopathology of seven hearts with CHB and associated maternal antibodies to the SSA/Ro polypeptide. In all of these hearts there was atrial-axis discontinuity: the AV node was replaced by varying degrees of fibrosis or fatty tissue.²³ The distribution of the distal conducting system was normal.

The diffuse fibroelastosis reported in some of these affected babies is considered to result from dilatation of the cardiac chambers secondary to the compensatory increased stroke volume present in CHB.²⁴ However, Nield and colleagues²⁵ have recently reported 13 CHB patients with endocardial fibroelastosis (EFE)—6 diagnosed in utero and 7 in the postnatal period—despite presumed adequate ventricular pacing of all but one infant. EFE is associated with significant mortality and morbidity: 9 (70%) of these 13 patients died, and 2 (15%) required heart transplants.

Given the importance of histologic data to infer pathogenic mechanisms, medical records of all families enrolled in the RRNL were reviewed to determine the incidence and timing of death, with emphasis on the pathologic findings in the affected fetal hearts.²⁶ Complete autopsy reports were available in 11 cases. The mean time from detection of CHB to autopsy was 11 weeks. Although in three cases there were various lesions of the tricuspid valve, the pathologic descriptions were more suggestive of an imposed injury than a true developmental defect. These included nodularity, dysplasia, hypoplasia and fusion of valve leaflets, and fibrosis. The pulmonary valve was abnormal in two, one of which was described as stenotic dysplastic and the other as nodular and dysplastic. Aortic valve insufficiency and stenosis and hypoplasia of the mitral valve leaflet were observed in one. Endocardial fibroelastosis of the right and left ventricles (RV, LV), with or without calcification, was present in 7.

Chronic changes in the myocardium were documented in 10, and included biventricular hypertrophy and increased RV and LV walls, thickened but hypoplastic RV, and hyperchromatic nuclei of the myocytes. Abnormalities of the AV node or vicinity were noted in 8 with involution, fibrosis, fatty infiltration, or calcification. However, in 2 the AV node per se appeared normal: in one there was calcification in adjacent tissue, and in another there was an atrophic His bundle with replacement by dense focally calcified fibrous tissue and scarring of the left and right bundle branches. Although previously unappreciated, autopsies obtained from the RRNL revealed a high incidence of valvular abnormalities. Although there were sufficient changes in the AV node to account for CHB in most cases, clinical conduction abnormalities may have been secondary to a functional exit block in a normal-appearing node.

SA nodal disease expands the spectrum of conduction dysfunction. However, from a clinical standpoint sinus bradycardia is almost never seen. In the rare instances it has been observed, it has not been sustained.²² The absence of sinus bradycardia is probably due to the presence of other atrial pacers. These studies leave little doubt that the signature lesion of autoantibody-associated CHB is fibrosis, which can clearly extend beyond the conduction system. Consequently, the cascade leading to fibrosis is a major focus of investigation.

ACCOUNTING FOR ACCESSIBILITY OF INTRACELLULAR ANTIGENS TO CIRCULATING MATERNAL AUTOANTIBODIES: APOPTOSIS MAY BE THE CRITICAL LINK

Apoptosis has traditionally been conceptualized from an immunologic point of view as either a means of maintaining B- and T-cell tolerance^27,28 or as a mechanism for providing accessibility of intracellular antigens to induce an immune response.²⁹ Casciola-Rosen and colleagues have demonstrated that autoantigens are clustered in two distinct populations of surface blebs on keratinocytes.²⁹ The larger blebs, so-called apoptotic bodies (derived from the apoptotic nucleus), contain both SSA/Ro and SSB/La proteins with SSB/La detected at the cell surface surrounding large blebs in the later stages of apoptosis. The 52kD protein was not specifically identified but rather deduced because evaluation was done with a patient serum considered “monospecific” for 52kD SSA/Ro antibodies. The smaller blebs, arising from fragmented rough endoplasmic reticulum and ribosomes, contain SSA/Ro (presumably of cytoplasmic origin). SSB/La was not contained in these blebs.

Apoptosis may be relevant in the pathogenesis of NLS. It is a selective process of physiologic cell deletion in embryogenesis and normal tissue turnover and plays an important role in shaping morphologic and functional maturity.³⁰ Apoptosis is a process that affects scattered single cells rather than tracts of contiguous cells. In the normal adult myocardium, apoptosis has been observed only rarely.^31,32 In contrast, apoptosis does occur during the development of the heart. In the 1970s, Pexeider extensively characterized the temporal and spatial distribution of cell death in the hearts of chicken, rat, and human embryos.³³ Major foci included the AV cushions and their zones of fusion, the bulbar cushions and their zones of fusion, and the aortic and pulmonary valves. Albeit much of the cell death was noted in non-myocytes, a focus of myocyte death was apparent in the muscular interventricular septum as it grew toward the AV cushions in mid-gestation. Takeda and colleagues demonstrated apoptosis in mid-gestational rat hearts using terminal deoxynucleotidyl transferase (dUTP) nick end labeling (TUNEL), an in situ technique that detects DNA strand breaks in tissue sections.³⁴ Although not coincident with the precise timing of CHB, it has also been suggested that apoptosis contributes to the postnatal morphogenesis of the SA node, AV node, and His bundle.³⁵ Perhaps a novel view of apoptosis is that it facilitates the placing of cardiac target autoantigens in a location accessible to previously generated maternal autoantibodies. Tissue damage might be a consequence of being in the right place at the wrong time.

Apoptosis in electively terminated human abortuses aged 18 to 24 weeks has been recently assessed by our group.²¹ In fact, there was little detectable apoptosis, but that seen was most prominent in the septal region. We hypothesize that under conditions of physiologic remodeling apoptotic cardiocytes are rapidly cleared, thus accounting for the limited detection.

To investigate the hypothesis that apoptosis indeed facilitates accessibility of SSA/Ro and SSB/La to circulating maternal autoantibodies, cultured human fetal cardiac myocytes were incubated with staurosporine or 2,3-dimethoxy-1,4-naphthoquinone (DMNQ).³⁶ By phase contrast microscopy, morphologic signs of early apoptosis were observed in 40% of the cardiocytes after approximately 4 hours and increased to 95% after 7 hours. The cellular topology of SSA/Ro and SSB/La was evaluated with confocal microscopy and determined in non-apoptotic and apoptotic cardiocytes by indirect immunofluorescence using two previously characterized antisera: one “monospecific” anti-SSB/La and the other recognizing both 52kD and 60kD SSA/Ro with goat anti-human IgG-FITC as secondary antibody.

In non-apoptotic cardiocytes, SSA/Ro was predominantly nuclear with minor cytoplasmic staining and SSB/La was confined to the nucleus. In early apoptotic cardiocytes, condensation of the SSA/Ro- or SSB/La-stained nucleus was observed accompanied in some cells by a “ring” of bright green fluorescence around the periphery. In the later stages of apoptosis, the nuclear SSA/Ro and SSB/La staining became weaker. Blebs could then be seen emerging from the cell surface, stained with both SSA/Ro and SSB/La. Scanning electron microscopy unambiguously confirmed the surface expression of SSA/Ro and SSB/La (as assessed by gold particle labeling of autoantibodies) on cultured human fetal cardiocytes rendered apoptotic.

These earlier published studies have now been extended to include a more in-depth evaluation of structure/function of extrinsic and intrinsic apoptosis pathways in human fetal and adult hearts, and surface accessibility of SSA/Ro-SSB/La Ag to maternal antibodies.³⁷ High levels of Fas-associated death domain protein (FADD) and TNFR-associated death domain protein (TRADD), key components in the apoptotic machinery, were observed in CHB but not normal cardiac tissues. Fetal cardiocytes readily became apoptotic following stimulation with either anti-Fas or TNFα when plated on poly(2-) hydroxyethylmethacrylate (pHEMA), a nonadherent condition. However, these same stimuli did not induce apoptosis in adherent cells. Thus, in fetal cardiocytes adhesion to substrate was pivotal to escaping extrinsic pathway activation, whereas adult cardiocytes did not undergo apoptosis via the extrinsic pathway even in the absence of anchorage. However, adult cardiocytes treated with staurosporine underwent apoptosis, suggesting that these cells do have the machinery to execute apoptosis via the intrinsic pathway.

Utilizing monoclonal antibodies generated from a chicken phage display library, it was demonstrated that Ro52, Ro60, and La48 are surface accessible on fetal cardiocytes regardless of the method used to induce apoptosis. Accessibility appeared to be restricted to select domains because not all antibodies that stained permeabilized cells were reactive with intact apoptotic cells. These studies support extrinsic activation of apoptosis, differentially operative in fetal compared to adult human cardiocytes, as a mechanism linking autoantibody to subsequent injury.

In vivo studies have confirmed the observations made in vitro. Tran and colleagues have demonstrated the translocation of SSB/La in apoptotic cardiocytes in the conduction system of the unmanipulated mouse fetal heart.³⁸ Clustering of SSB/La near the surface of apoptotic bodies occurs in vivo under physiologic conditions. To assess proof of concept and examine whether SSB/La and/or SSA/Ro epitopes on apoptotic cells are accessible for binding by antibodies in vivo, these same investigators have exploited a murine passive transfer model in which the fate of human autoantibodies actively transported across the placenta could be traced in fetal tissues known to have high rates of apoptosis.³⁹

Specifically, BALB/c pregnant mice were injected with human anti-SSA/Ro-SSB/La serum, monospecific anti-Ro60 serum, affinity-purified anti-SSB/La, anti-dsDNA, or normal human serum. Apoptotic cells identified in the fetal conduction tissue (present under normal physiologic conditions of remodeling) showed redistribution of SSB/La from the nucleus to the surface of apoptotic bodies. Fetuses from anti-SSA/Ro-SSB/La Ab-injected mothers showed a striking co-localization of human IgG with apoptotic cells in the atrium, AV node, liver, skin (with particulate epidermal deposition), and newly forming bone. The IgG-apoptotic cell complexes were organ specific and not detected in thymus, lung, or gut. No IgG deposits were identified in fetuses from mothers injected with anti-dsDNA, anti-Ro60, or normal sera. Experiments with affinity-purified anti-SSB/La and anti-SSA/Ro-SSB/La Abs absorbed with SSB/La confirmed the specificity of deposited IgG as anti-SSB/La. That Ro60 was not identified on the apoptotic cells remains puzzling.