Typing the Histogenetic Origin of the Tumor Cells of Lymphocyte-rich Classical Hodgkin’s Lymphoma in Relation to Tumor Cells of Classical and Lymphocyte-predominance Hodgkin’s Lymphoma¹

HL⁴ is characterized by few HRS cells in a background of lymphocytes, plasma cells, histiocytes, and eosinophils. On the basis of histology and the immunophenotype of the HRS cells, two subgroups, c and lp HL, are distinguished (1, 2).

In lpHL, which accounts for ∼5% of all of the HLs, the infiltrated tissue is dominated by small lymphocytes, which are mostly B cells and show a nodular or diffuse growth pattern. The neoplastic L&H cells express CD20 and CD45, but are negative for CD30 and CD15, and are not infected by EBV (1, 3). In contrast, in cHL the HRS cells express CD30, and are usually CD15-positive, CD45-negative, mostly CD20-negative, and in ∼40% of cases EBV-infected. In addition, BCL6, CD138, Oct2, and BOB1 are also differentially expressed in the HRS cells of cHL and the L&H cells of lpHL (4, 5, 6, 7). BCL6, a zinc-finger transcriptional repressor, is expressed in GC B and T cells, and in the L&H cells of lpHL but only rarely in HRS cells of cHL. Conversely, CD138, a proteoglycan belonging to the syndecan family, is expressed in plasma cells and in most cases of cHL in the majority of the HRS cells but not in L&H cells (8, 9, 10). The transcription factors Oct2 and BOB1 are involved in the transcription of Ig genes, and usually expressed in all of the L&H cells in all of the cases of lpHL but not or only partially in the HRS cells of cHL (6, 7).

The origin of the HRS cells of cHL has been enigmatic for a long time, as they express markers of different hematopoietic lineages. Only recently the analyses of Ig and T-cell receptor loci of single HRS cells revealed that the cells represent monoclonal populations of tumor cells, and in the vast majority of cases derive from B cells and only in a few cases from T cells (11, 12, 13, 14, 15). The rearranged Ig genes are usually mutated, and in about a quarter of cases originally potentially functional Ig rearrangements have acquired obviously crippling mutations destroying the coding capacity (12, 13, 16). Furthermore, the pattern of somatic mutations (the ratio of R versus S mutations in FRs) indicated that the cells were only partially selected for the expression of a functional antigen receptor during the GC reaction (12, 17). As GC B cells with unfavorable or crippling mutations are usually quickly eliminated by apoptosis, and as only a fraction of unfavorable mutations that cause apoptosis of GC B cells are easily identifiable as crippling mutations, HRS cells as a rule may represent preapoptotic GC B cells rescued by a transforming event (12, 16).

For the L&H cells of lpHL, the expression of several B-cell markers suggested a B-cell origin, and single-cell studies indeed revealed that the L&H cells carried rearranged Ig genes (11, 18, 19, 20). Although the Ig rearrangements in both cHL and lpHL are mutated, the mutation patterns are different. In contrast to cHL, in most lpHL cases analyzed intraclonal diversity was observed and the rearranged Ig genes carried no crippling mutations (with 1 exception in 17 informative cases) and were selected for expression of a functional antigen receptor (18, 19, 21). These findings indicate that the L&H cells of lpHL are derived from mutating GC B cells.

During a review of 388 cases diagnosed previously as lpHL the European Task Force on lymphoma project on lpHL recognized that about a quarter of the cases showed a histology resembling lpHL, whereas the immunophenotype of the HRS cells was the one of classical HRS cells, namely CD30- and CD15-positive, and CD20-negative (22). In addition, in about half of these cases the HRS cells were EBV-infected. This new subgroup was named lrcHL and is grouped to the classical forms of HL. However, retrospective analysis of clinical data revealed a clinical behavior resembling those of lpHL and distinct from cHL (used from here on to define classical HL excluding lrcHL; Ref. 23). As the origin of the HRS cells of cHL and the L&H cells of lpHL from two different types of GC B cells was recognized by the pattern of somatic mutations in their rearranged Iggenes (16), and thus far only a single Ig gene rearrangement from one case of lrcHL was analyzed (24), we amplified and sequenced the rearranged Ig genes of HRS cells from seven cases of lrcHL to determine their cellular origin and their relationship to HRS cells of cHL and L&H cells of lpHL on a molecular basis.

All of the lymph node biopsies were taken for diagnostic purposes. Clinical data of the seven patients are summarized in Table 1.

For immunohistochemistry on formalin-fixed sections with antibodies against CD30, CD15, CD20, CD45 leukocyte common antigen, CD3, LMP1 (all from DAKO, Hamburg, Germany), BCL6, CD57 (Novocastra, Newcastle upon Tyne, United Kingdom), CD138 (Serotec, Oxford, United Kingdom), Oct2, and BOB1 (both from Santa Cruz Biotechnology, Santa Cruz, CA) the avidin-biotin complex technique with alkaline phosphatase was applied using Fast Red (DAKO) as chromogen. For micromanipulations frozen tissue sections were stained with anti-CD30 antibody (DAKO). EBER in situ hybridization was performed on frozen and on formalin-fixed sections as described (25).

Single cells were mobilized, aspirated, and transferred into PCR tubes containing 20 μl of Expand high-fidelity PCR buffer (Roche, Mannheim, Germany) with a hydraulic micromanipulator as described previously (26). Cells were stored at −20°C.

After proteinase K digestion [2 h, 50°C, 0.3 mg/ml proteinase K (Roche, Mannheim, Germany) followed by 10 min at 95°C] the whole genomes of 10 single cells of case 4 were amplified several times using a random 15-mer primer at 33.3 μm in 60-μl reactions with 1× Promega PCR buffer A (Promega, Madison, WI), 0.17 mm deoxynucleoside triphosphates, and 2.5 mm MgCl₂ (27). After 2 min at 95°C 5 units of Taq DNA polymerase (Promega) were added at 80°C, followed by 2 min at 37°C and 4 min at 55°C, and 49 times 1 min at 95°C, 2 min at 37°C, and 4 min at 55°C. Temperature increments from 37°C to 55°C were 0.1°C/s.

After a proteinase K digestion (Roche), seminested PCRs for IgH, Igκ and Igλ gene rearrangements were performed for each single cell as described previously (12, 18, 28). For a fraction of the cells of case 4, EBNA1-specific primers were included for detection of EBV-infected cells (14). PCR products were gel purified and sequenced directly. Sequences were analyzed with the EMBL IMGT database.⁵ Unmutated V_κ rearrangements were considered as uninformative regarding mutation analysis, because nonexpressed V_κJ_κ joints are often inactivated by deletion of the Igκ enhancers, which also abolishes somatic hypermutation (29).

Lymph nodes infiltrated by lrcHL showed large nodules composed of small lymphocytes, some of them exhibiting morphological features of follicular mantle cells. Occasionally atropic GCs could be found. In case 4, the HL-infiltrated lymph node region was surrounded by several GCs containing many CD30⁺ cells. In all of the cases the nodular infiltrates were dominated by B cells and varying amounts of T cells, whereas the majority of lymphocytes between the nodules were T cells. Only very few of the T cells were CD57⁺. Histiocytes and occasionally epithelioid cells were detected. The tumor cells showed a morphological spectrum from blasts to classical HRS cells and were often localized in mantle zones. Rosettes of CD3⁺ T cells around HRS cells were only observed in case 5, and CD57⁺ rosettes were never observed.

The number of tumor cells varied from case to case, and the cells displayed the typical immunophenotype of HRS cells of cHL, with most or all cells being positive for CD30 and CD15. Moreover, CD45⁺ HRS cells were observed in only two cases and some CD20⁺ HRS cells only in one case (Table 2; Fig. 1). In three cases, the HRS cells were EBV⁺ by in situ hybridization for EBER transcripts (Table 2) and expressed LMP1 (data not shown). In the seven cases analyzed, no BCL6 expression by HRS cells was observed, and CD138 expression could be detected only in three cases, in two of these cases detection was in only a fraction of the HRS cells (Table 2). The HRS cells did not express BOB1, whereas Oct2 was expressed in the HRS cells of three cases. In all seven of the cases analyzed positivity of various numbers of non-HRS cells for each of the antibodies against BCL6, CD138, BOB1, and Oct2 confirmed successful staining.

Thus, although the HRS cells of lrcHL show the same expression pattern like HRS cells of cHL regarding CD30, CD15, CD20, and CD45, the stainings for BCL6, CD138, Oct2, and BOB1 show differences from both, the HRS cells of cHL and the L&H cells of lpHL. Whereas HRS cells of lrcHL resemble HRS cells of other types of cHL in that they are negative for BCL6 and BOB1 expression, the lack of CD138 expression in four of seven cases, and the expression of Oct2 in most HRS cells of three of six cases is different from the phenotype of HRS cells of cHL. These findings were additionally corroborated by immunohistochemical analysis of six additional lrcHL cases, in all of which HRS cells were BCL6⁻; only one case had CD138⁺ HRS cells (data not shown). However, in all six of the additional cases HRS cells were Oct2⁻.

Sixty to 214 single CD30⁺ HRS cells were micromanipulated from 5-μm sections of frozen tissue of seven cases of lrcHL (Table 3). Buffer aliquots covering the sections during the micromanipulation procedure were taken as negative controls, usually 4 for each 10 micromanipulated cells. From case 4 30 large EBER⁺ cells were also micromanipulated (see below). All of the cells were subjected to seminested amplification of IgH, Igκ, and Igλ rearrangements using V gene family-specific FRI primers and J segment primer mixes. For a fraction of cells of cases 1 and 3–7, IgH rearrangements were additionally analyzed using V gene family-specific leader primers. Fifty-four CD30⁺ cells and 30 EBER⁺ cells of case 4 were additionally subjected to a seminested PCR for a fragment of the EBNA1 gene of EBV.

From the HRS cells of each of cases 1–3 and 5–7, clonally related Ig rearrangements were amplified (Tables 3 and 4). In each of these cases an originally potentially functional V_H rearrangement was obtained. For case 1, in addition, a potentially functional and a nonfunctional V_κ rearrangement were amplified. In case 3 the HRS cells also harbored a nonfunctional V_H rearrangement and two nonfunctional V_κ rearrangements. In addition, the HRS cells of case 5 harbored a D_HJ_H rearrangement on the second IgH allele, a mutated potentially functional V_κ, and two unmutated nonfunctional V_κ rearrangements. Because of the organization of the human Igκ locus, a B cell can carry more than one V_κJ_κ joint on one Igκ allele (30), and the three clonal V_κJ_κ joints amplified from the HRS cells of case 5 can indeed be present on the two Igκ alleles of a B cell. The HRS cells of cases 6 and 7 harbored clonal D_HJ_H rearrangements on their second IgH alleles. In case 6, the HRS cells also carried a mutated potentially functional V_κ rearrangement, whereas in case 7 two mutated, originally potentially functional V_κ rearrangements were obtained.

Besides the clonal rearrangements amplified repeatedly, 15 unique rearrangements and four fragments of unrearranged IgH loci, alone or coamplified with clonal rearrangements, were also obtained from the 466 HRS cells analyzed from cases 1–3 and 5–7 (Table 3). These rare rearrangements and fragments from unrearranged IgHloci (obtained from 4% of the cells) most likely represent cellular contamination from the micromanipulation procedure. Similar low-level contamination frequencies were observed in previous experiments (12) and for buffer controls in the present study (Table 3).

In case 4, a nonproductive V_H rearrangement (C1) was amplified from 67% (41 of 61) of the PCR-positive CD30⁺ cells micromanipulated on three different occasions (Table 3, summarized as experiment 1). Surprisingly, two additional clonal V_H rearrangements were amplified from 5 (C2) and 3 (C3) HRS cells, and from 14 cells 16 unique V gene rearrangements were amplified (Fig. 1). Pairs of the distinct clonal rearrangements were never coamplified from the same cell. Two of the unique V_H rearrangements were coamplified with the dominant rearrangement (C1). The high frequency of these other rearrangements (representing 33% of the PCR-positive cells), the fact that they were obtained in four independent micromanipulations of CD30⁺ HRS cells (three micromanipulations for experiment 1 and one for experiment 3, see below and Table 3), and that the 85 buffer controls of these micromanipulations were all negative strongly support the reliability of these results.

EBV-infected B cells can occasionally acquire an HRS-like phenotype (including CD30 positivity; Refs. 31, 32, 33, 34), and the detection of smaller clones and unique cells in addition to a dominant clone in case 4 could be because of such cells. [Indeed, 1 cell of clone 3 that could be tested retrospectively for EBV infection as it was subjected to a whole genome preamplification (see “Materials and Methods”) was found to be EBV-positive.] Therefore, we also analyzed the clonal composition of EBV-infected (EBER⁺) cells, and the clonal composition and the EBV status of CD30⁺ HRS cells of case 4, the latter by PCR amplification of an EBNA1 fragment of EBV.

Whereas 2 of the 30 micromanipulated EBER⁺ cells were clonally related to each other (C4), none of the nine V_H rearrangements amplified was clonally related to those obtained from the CD30⁺ cells (Table 3, experiment 2). Among the CD30⁺ cells, most V gene-positive cells were also EBV-positive (26 of 30), and nearly all of the IgH and EBNA PCR-positive cells (25 of 26) carried the dominant V_H rearrangement (C1), whereas 1 EBV⁺ B cell with a unique V_H rearrangement was observed (Table 3, experiment 3). One of four IgH PCR-positive but EBNA PCR-negative cells belonged to the dominant clone (C1), whereas the three others carried unique rearrangements. The finding of these “EBV-negative” cells could be because of unsuccessful EBNA1 PCR, or, in the case of the unique cells, it may be speculated that they could be derived from the CD30⁺ GC cells surrounding the HL infiltrate in that case (see “Histology and Immunohistology”; GC structures were not identified on the frozen sections used for micromanipulation).

Taken together, in cases 1–3 and 5–7 single clones were identified among the CD30⁺ cells with HRS cell morphology as expected from previous analyses of many HLs, whereas in case 4, besides a dominant EBV⁺ clone of HRS cells, a population of cells with HRS cell morphology (EBV⁺ as well as EBV⁻), including two smaller clones of which at least one was EBV⁺, were also identified.

All of the informative V gene rearrangements (see “Materials and Methods”; for case 4 only the dominant clone is considered) from cases 1–7 were mutated. The mutation frequencies ranged from 4 to 22.1% (Table 4). Also, the case of lrcHL analyzed previously carried a potentially functional V_H rearrangement with a mutation frequency of 6.1% (24). In case 4 the two small expansions (C2 and C3) and 8 of the 12 EBER⁺ cells, including clone 4, carried mutated V genes.

Like in the one case analyzed previously (24), no intraclonal diversity was observed in cases 1, 2, and 6, although large numbers of rearrangements were analyzed [16 V_H, and 16 and 2 V_κ rearrangements in case 1, 18 V_H rearrangements in case 2, 37 V_H and 25 V_κ rearrangements in case 6 (Table 4), and 14 V_H rearrangements in the case from Irsch et al. (24)]. In case 3, two sequence variants of the V_H1–18 rearrangement were identified, namely a functional V_H1–18 rearrangement and the same rearrangement with a 250-bp deletion, which were amplified from 9 and 10 cells, respectively. Both forms of the V_H1–18 rearrangement were never amplified from the same cell. In case 4 only 1 of 67 V_H gene sequences of the dominant clone carried an additional single point mutation, whereas in case 5, 1 V_H rearrangement carried one additional mutation and another V_H rearrangement carried two additional point mutations compared with the other 14 sequences of the clonal rearrangement. In case 7, 1 of 26 V_H rearrangements carried an additional point mutation.

Crippling mutations rendering originally potentially functional rearrangements nonfunctional were observed in two of the seven cases. In case 3, as described above, half of the HRS cells carried a 250-bp deletion in the V_H1–18 rearrangement expressed originally. The rearrangement on the second IgH allele is rearranged out of frame, confirming that the V_H1–18 rearrangement was the originally functional one. In case 7, all of the HRS cells carried an originally functional V_H rearrangement that was crippled because of one nonsense mutation and two large duplications in the V gene segment (Table 4). A clonal D_HJ_H rearrangement was amplified from the second IgH allele. Moreover, the two V_κ rearrangements that were mutated and originally potentially functional were also rendered nonfunctional by crippling mutations (Table 4).

The ratio of R:S mutations in FRs of potentially functional Ig rearrangements is an indicator as to whether a cell has been selected for expression of a functional antigen receptor during the acquisition of somatic mutations in a GC reaction (17). The ratio of 1.4 (140:98) from 11 informative in-frame rearrangements of seven cases of lrcHL [(six cases from this study and the case from Irsch et al. (24)] is within the range (1.0–1.5) typical for selected cells.

Taken together, the HRS cells in lrcHL are derived from antigen-experienced B cells without or with only very limited intraclonal diversity. The presence of crippling mutations in originally functional rearrangements in two cases shows that the HRS cells, at least in these two cases, are no more dependent on the expression of a functional B-cell receptor, although the R:S ratio from all of the cases is indicative of stringent selection for functionality.

Besides expression of CD30 and CD15, and lack of CD20 and CD45 expression in most cases (22), the HRS cells of lrcHL also resemble cHRS cells in the lack of expression of BCL6 and BOB1, whereas these latter proteins have been described to be expressed in the L&H cells of all of the lpHLs (4, 7). However, whereas HRS cells in cHL usually express CD138 and mostly lack expression of Oct2 (4, 7), we observed CD138 expression by HRS cells in only 4 of 13 cases analyzed, and 3 of 12 cases were Oct2-positive. The lack of BOB1 expression in HRS cells in all of the cases analyzed, and expression of Oct2 in only a fraction of cases may indicate that in lrcHL, rearranged Ig genes are, like in cHL and different from lpHL, not transcribed.

The discrepancy to the work of Kraus and Haley (35), who described BCL6 expression in the HRS cells in each of three cases of lrcHL, is presently unclear. BCL6 negativity of the HRS cells in the present study is unlikely because of a technical failure, as some small BCL6⁺ cells were observed in the tissue sections. Taken together, whereas HRS cells of lrcHL resemble HRS cells of cHL in the expression pattern of many immunophenotypic markers, there are also differences.

With one exception (36), all of the B-cell derived cHLs and lpHLs analyzed thus far carried mutated Ig rearrangements, and this also holds true for the HRS cells of lrcHL (for case 4 only the rearrangement from the dominant clone is considered).

Whereas HRS cells of cHL show an average R:S value in the FRs of 1.7, which is similar to the value of (partially selected) GC centroblasts, L&H cells show an average R:S value of 1.4, i.e., similar to selected memory B cells and plasma cells, and several types of Ig-expressing B-cell non-Hodgkin lymphomas (37). For the 11 potentially productive Ig gene rearrangements of lrcHL the average R:S ratio was 1.4, indicating that, like in lpHL and different from cHL, R mutations in FRs were stringently counterselected.

Mutations that rendered originally productive rearrangements nonfunctional were found in 25% of cases of cHL among >40 cases analyzed (16). Such mutations were detected in only 1 of 17 cases of lpHL, and in that case, only a subclone of the L&H cells harbored the destructive mutation, suggesting that the latter occurred after establishment of the tumor clone (19, 21). The same may also hold true for lrcHL case 3, where half of the HRS cells harbored a crippling mutation. However, in case 7 all of the HRS cells carried crippling mutations, resembling the situation found in cHL.

Intraclonal sequence diversity because of ongoing somatic hypermutation during tumor clone expansion is observed frequently in lpHL (overall in 14 of 22 cases), whereas this is rare in cHL (4 of 16 informative cases; Table 5). Moreover, whereas the intraclonal sequence diversity was restricted to two sequence variants per case in the four cases of cHL showing sequence variants, and variant sequences accounted for 1.3% of all of the sequences, multiple sequence variants were usually observed in the L&H cells of lpHL with 28.4% of sequences representing different variants (Table 5). Although in four of eight lrcHLs variants were observed, these variants accounted only for 2.6% of the sequences. Thus, like in cHL, there is no significant ongoing hypermutation in the HRS cells of lrcHL, establishing a clear distinction to the L&H cells in lpHL. This virtual absence of hypermutation activity in the expanding tumor clones of lrcHL may, besides intrinsic features of the HRS cells, also be related to differences in the microenvironment compared with lpHL. For example, whereas the L&H cells of lpHL are usually surrounded by T cells resembling GC T cells (CD57⁺BCL6⁺), such T cells are absent in cHL and lrcHL (35).

Taken together, the lack of significant ongoing somatic hypermutation during tumor clone expansion clearly distinguishes HRS cells of lrcHL from L&H cells of lpHL. The analysis of mutation pattern regarding selection for expression of a functional B-cell receptor of HRS cells of lrcHL revealed similarities but also differences to HRS cells of cHL. Whereas the presence of crippling mutations, as in cHL, may hint to an origin of the HRS cells from preapoptotic GC B cells, the R:S ratio of 1.4 is indicative of stringent selection for a functional B-cell receptor, while the precursors of the lymphoma clones acquired these mutations. However, as R:S values can show considerable variation (37), the number of cases analyzed from the rare lrcHL may be too small to give conclusive results. Thus, we conclude that the mutation pattern in HRS cells of lrcHL largely resembles the pattern observed in HRS cells of cHL.

The clonality of the HRS cells has been a matter of debate for a long time, but analyses of Ig rearrangements of single HRS cells from ∼60 cases finally lead to the consensus that the HRS cells of cHL as well as the L&H cells of lpHL are monoclonal expansions (11, 13, 20, 38). Whereas the analyses of cases 1–3 and 5–7 were in accord with these previous findings, and, hence, show that also in lrcHL, the HRS cells represent monoclonal populations of B cells, in case 4 besides a dominant clone of EBV-positive HRS cells two smaller clones and a significant fraction of cells with unique Ig rearrangements were identified among the CD30⁺ cells with HRS cell morphology (Fig. 1). The analyses of CD30⁺ cells for EBV infection and of EBER-positive cells collectively show that in case 4, there was a large population of EBV-infected B cells not belonging to the HRS cell clone, and that some of the EBV-infected B cells showed a morphology of HRS cells and expressed CD30.

EBV-harboring HRS-like cells are observed regularly in infectious mononucleosis (31, 32) and occasionally also in the setting of B-CLL (33, 34). In two cases of B-CLL with EBV-positive HRS-like cells, these cells represented clonal populations of cells unrelated to the B-CLL (39), and it may be the particular microenvironment of the lymphomas allowing expansion of EBV-harboring B cells, which are normally tightly controlled by cytotoxic T cells (40, 41). The case presented here may indicate that in rare instances, also in lrcHL, EBV-infected B cells can become a significant population, and that these cells may even acquire an HRS-cell morphology. Increased frequencies of EBV-infected B cells unrelated to the HRS cell clone (42) are indeed often observed in cHL (43, 44). In light of these observations, the finding of unique and oligoclonal cells morphologically indistinguishable from HRS cells in a single case does not challenge the concept of HL as a lymphoma characterized by monoclonal HRS cells.

In conclusion, the immunophenotype and the mutation pattern of rearranged Ig genes of HRS cells of lrcHL revealed that these cells show phenotypic differences to both HRS cells of cHL and L&H cells of lpHL, and that they derive from (preapoptotic?) GC B cells that silenced somatic hypermutation, and, hence, resemble HRS cells of cHL. In light of these findings, the assignment of lrcHL to the group of cHL appears justified.

We thank Christiane Gerhardt, Tanja Schaffer, and Yvonne Blum for excellent technical assistance.