Viral infection is the primary cause of changing B-cell epitopes from CD19 to CD20. The usual explanation is that this is due to EBV, which is practically endemic in humans. The persistence of autoimmune response is something of a theoretical problem.
Part of the answer could come from EBV, a member of the herpes family which inserts its own DNA through pores in the nuclear membrane. Persistent active infection could generate antibodies aimed at something inside that membrane. A retrovirus could also result in antinuclear antibodies, because with its RNA transcribed to DNA and integrated into chromosomes, active infection would also provoke immune attacks on the cellular machinery generating the offending proteins.
Several members of the herpes family of viruses have been partially implicated in this disease: HSV-1, HSV-2, EBV, VZV, HHV-6. HHV-6 is even found integrated into chromosomes, which herpes viruses are not supposed to be able to do. We even have examples of cell lines derived from prostate cancer "constitutively producing" both a gamma retrovirus (JHK) and EBV.
Our problem comes from virologists who started classifying this as contamination 40 years ago, and have never seen fit to question their own views. Even when one such virus comes from humans, as above, they insist the retrovirus must have come from a mouse -- because there are no known human gamma retroviruses. A review of the literature convinced me that type C virions have a marked tendency to appear in cell lines associated with particular classes of cancers, like lymphomas or adenocarcinomas. Beyond simply indicating that the etiology of the cancer was unknown, (thus a likely subject of research,) the appearance of such a C-type retrovirus, and dismissal as a contaminant, seems to be a good predictor of poor future progress on etiology.
I believe we are seeing dual infection by herpes viruses, or possibly others, and a small retrovirus. With typical viral genomes in the herpes family running around 170,000 base pairs, and the suspect retroviruses around 8,100 base pairs, it would be no trick at all to slip the entire retrovirus inside a herpes virus. The retrovirus could be parasitizing a virus which human immune systems tolerate. The exact relationship between the two needs to be elucidated.
As always, these are personal views without the weight of authority.