T‐cell‐derived Hodgkin lymphoma has motility characteristics intermediate between Hodgkin and anaplastic large cell lymphoma

Abstract Classic Hodgkin lymphoma (cHL) is usually characterized by a low tumour cell content, derived from crippled germinal centre B cells. Rare cases have been described in which the tumour cells show clonal T‐cell receptor rearrangements. From a clinicopathological perspective, it is unclear if these cases should be classified as cHL or anaplastic large T‐cell lymphoma (ALCL). Since we recently observed differences in the motility of ALCL and cHL tumour cells, here, we aimed to obtain a better understanding of T‐cell‐derived cHL by investigating their global proteomic profiles and their motility. In a proteomics analysis, when only motility‐associated proteins were regarded, T‐cell‐derived cHL cell lines showed the highest similarity to ALK− ALCL cell lines. In contrast, T‐cell‐derived cHL cell lines presented a very low overall motility, similar to that observed in conventional cHL. Whereas all ALCL cell lines, as well as T‐cell‐derived cHL, predominantly presented an amoeboid migration pattern with uropod at the rear, conventional cHL never presented with uropods. The migration of ALCL cell lines was strongly impaired upon application of different inhibitors. This effect was less pronounced in cHL cell lines and almost invisible in T‐cell‐derived cHL. In summary, our cell line‐derived data suggest that based on proteomics and migration behaviour, T‐cell‐derived cHL is a neoplasm that shares features with both cHL and ALCL and is not an ALCL with low tumour cell content. Complementary clinical studies on this lymphoma are warranted.


| INTRODUC TI ON
Classic Hodgkin lymphoma (cHL) and anaplastic large cell lymphoma (ALCL) both represent lymphomas with CD30+ tumour cells. 1 cHL is frequently diagnosed in early stages of the disease, 2 whereas ALCL is more frequently encountered with advanced-stage disease and has a less favourable outcome. 3 Despite the fact that these lymphomas originate from fundamentally different cells, with a B-cell origin for cHL 4 and a T-cell origin for ALCL, 5 both lymphomas have several features in common: both have lost expression of lineage-specific surface markers and both have constitutively active NFKappaB, NOTCH1, AP1 and JAK-STAT signalling. 6,7 ALCL is classified into ALK + ALCL, with a rearrangement of the ALK gene, and ALK − ALCL, which can harbour rearrangements of DUSP22 or TP63. 8,9 For ALK + ALCL, a Hodgkin-like subtype is described, which is particularly difficult to differentiate from cHL purely by morphology. 10 In this case, ALK1 protein expression is a helpful diagnostic tool. However, ALK − ALCL can also present with a low tumour burden and may thus be difficult to distinguish from cHL since ALK1 is not expressed by ALK − ALCL. Usually, demonstration of nuclear PAX5 staining is very helpful in this setting, as PAX5 is usually expressed in cHL but not in ALCL. 11 However, the classification of some cases that do not express PAX5 remains unclear. PAX5 − cHL cases were shown to have inferior survival compared with conventional cHL cases. 12 Demonstration of clonal B-or T-cell receptor (TCR) rearrangements can be helpful in some cases, but due to the low tumour cell burden and extensive somatic hypermutation of immunoglobulin genes in cHL, clonal rearrangements are frequently missed. Immunohistochemical staining may provide some help in the diagnosis. 13 In a subset of cases, an association with cutaneous ALCL has been observed, with single, scattered, large CD30 + tumour cells in the lymph nodes mimicking cHL. 14 These cases represent a nodal manifestation of a cutaneous ALCL. However, few cases remain, in which no link to a cutaneous T-cell lymphoma can be demonstrated. Based on single-cell polymerase chain reaction (PCR), clonal rearrangements of the TCR could be demonstrated in such cases. 15 However, even two of the recognized cHL cell lines, namely L-540 and HDLM-2, harbour clonal TCR rearrangements, thus originating from T cells. HDLM-2 was established from the pleural effusion of a 74-year-old man with Hodgkin lymphoma (nodular sclerosing; stage IV) in 1982. 16 L-540 was established from the bone marrow of a 20-year-old woman with Hodgkin lymphoma (nodular sclerosing; stage IVB). 17 In gene expression studies, these cell lines clustered very well with cHL. 18 Thus, the biology of cases with the morphology of cHL and clonal TCR rearrangement in the tumour cells is still unclear. The typical morphology and immunophenotype of the tumour cells, the microenvironment, and the crippled B-cell receptor genes are all part of the definition of cHL. However, in these particular cases with cHL morphology and clonal TCR rearrangement in the neoplastic cells, traditional classification criteria do not help any further. As we recently observed that malignant lymphomas differ in their migratory properties, 19 the present study was based on the idea that an analysis of the motility of such tumour cells could help to better understand and classify these lymphomas. Therefore, the present study aimed to better understand the biology and classification of T-cell-derived cHL neoplasms with a focus on the T-cell-derived cHL cell lines L-540 and HDLM-2 and their migratory behaviour.

| Proteomics
The All cell lines underwent quantitative proteomics based on tandem mass tagging (TMT). For this, 5 × 10 5 cells were harvested and lysed in urea buffer as previously described. 20 After protein extraction and digestion with trypsin, peptides were labelled with TMT 10-plex reagents (Thermo Fisher Scientific, Dreieich, Germany). The individually labelled peptides were combined in multiplexed samples. For normalization among multiplexes, an internal reference consisting of peptides from each condition was included. After pre-fractionation using a high-pH C18 reversedphase kit (Thermo Fisher Scientific), the TMT-labelled peptide mixtures were separated by C18-reversed-phase-HPLC on an Ultimate 3000 RSLCnano system (Thermo Fisher Scientific). Peptides eluted from the analytical column were analysed by nano-ESI Orbitrap tandem mass spectrometry on a Q Exactive HF instrument (Thermo Fisher Scientific). The raw mass spectrometric data were processed with the MaxQuant software (version 1.6.17.0; Max Planck Institute of Biochemistry [MPIB]). 21 This included the identification of proteins by running the mass spectra against the Uniprot human protein database (downloaded in February 2019) and collection of common laboratory contaminants, followed by the extraction of TMT reporter ion intensities for protein quantitation. After recalibration, the mass tolerances for precursor and fragment ions were set to 4.5 and 20 ppm, respectively. Oxidation of methionine and acetylation of the protein N-terminus were considered variable modifications, and carbamidomethylation of cysteine was defined as a fixed modification.
The minimal peptide length was set to seven amino acids, allowing up to two missed tryptic cleavages. Both at the peptide and protein levels, a maximum false discovery rate (FDR) of 1% was applied using a reversed decoy database. For downstream data processing with the Perseus software (version 1.6.0.7; MPIB), potential contaminants, proteins identified solely with modified peptides, and hits in the decoy database were removed. After normalization of TMT reporter ion intensities across multiplexed samples, the data were subjected to unsupervised hierarchical clustering based on the Euclidean distance and average linkage method, either using all quantified protein groups or after mapping motility-related proteins. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE 22 partner repository with the dataset identifier PXD031907 (https://www.ebi.ac.uk/pride).

| Microchannel experiments
Polydimethylsiloxane (PDMS) chips with different types of microchannels were produced in moulds provided by Dr. Matthieu Piel.
Straight channels with a diameter of 8 µm and height of 10 µm were tested and found to be most appropriate for the rather large

| Segmentation of movies from microchannel experiments
Segmentation was done with the scikit-image package in Python. 24 First, noise was removed with a median filter, and a background image was calculated by taking the median pixel value for each position over all frames of the video. Then, this background image was subtracted from each frame to remove the microchannels from the image. A mean filter was applied to smoothen the image again, and a black top-head filter was used to enhance the signal of the cells. Finally, the cells were detected with a Yen threshold, and morphological operations were used to improve the result.

| Immunofluorescence
Cytospins from all cell lines were prepared and incubated with a The presented pixel size was 130 nm in each coordinate direction.
Furthermore, the z-step had a size of 0.13 mm. 25,26 3 | RE SULTS

| T-cell-derived cHL cell lines share baseline movement characteristics with both cHL and ALK − ALCL cell lines
Since we previously observed differences in cell migration between ALCL and cHL, 27 we were now interested in studying how  Figures S1 and S2, and Movies S9-S11).

| Inhibitors impair the cells' velocity but have little impact on the mode of migration
Next, we tested the effect of the myosin II inhibitor blebbistatin (15 µM) on all cell lines. Significant differences in velocity were observed here between the ALK − ALCL cell lines and cHL cell lines.
Surprisingly, the T-cell-derived cHL cell lines could even move slightly better after blebbistatin application ( Figure 2D), suggesting that these cell lines may still be able to move after myosin II inhibition in relation to their higher endogenous myosin II levels as observed in the global proteomics screening. After blebbistatin application, the morphology of the migrating cells changed regularly in that the cells had problems with the retraction of their rear, leading to long tails at their rear ends (Figure 3, Supporting Information Figure S3, and Movies S12-S14). However, cells moving in the A2 mode were occasionally encountered, sometimes with a long tail at the uropod

| DISCUSS ION
To our knowledge, this study is the first to assess cell motility in order to classify different lymphoma cell lines. We could demonstrate in previous studies 19,27 that lymphoma entities with similarities in morphology, immunohistochemistry and molecular characteristics show differences in their migratory properties. Although cHL with clonal TCR rearrangements have been described, 15  The rarity of these lymphoma cases was also the reason we did not have the chance to study suspensions of such primary cases in microchannels. Although we tested suspensions of conventional primary cHL and ALCL, the number of CD30+ tumour cells entering microchannels was too low to draw any meaningful conclusions.
The majority of moving cells here represented reactive T cells from the microenvironment. Therefore, we must currently rely on T-cellderived cHL cell lines as a good model for this disease.
T-cell-derived cHL cell lines presented some moving cells with a uropod and typical amoeboid A2 morphology, which we also observed in ALCL cell lines. Formation of an uropod with clustering of TCRs in a dynamic immunological synapse, also called kinapse, 29 is frequently observed for normal T cells. 30 Therefore, the manner of movement of T-cell-derived cHL cell lines reflects their T-cell origin.
However, the generally strongly impaired motility in T-cell-derived cHL cell lines resembles B-cell-derived cHL.
With regard to susceptibility towards different inhibitors, the ROCK inhibitor Y27632 had some effect on the T-cell-derived cHL cell lines. As observed in almost all cell lines, long tails at the rear of the cell were seen, as also previously described in the context of monocyte tail retraction after Rhoa blocking. 31

CO N FLI C T O F I NTE R E S T
The authors do not report any conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The mass spectrometry proteomics data have been deposited to