Transgenic shCD6EμLckTg mice were generated in C57BL/6N background. Non-transgenic (NonTg) mice used for comparative purposes came from the same common heterozygous ancestors as the shCD6EμLckTg mice and were kept under the same housing conditions as the latter. The shCD6LckEμTg mice were viable and fertile, and their lifespan and behavior were equivalent to those of NonTg mice. WT C57BL/6N mice were purchased from Charles River Laboratories. All animals were kept under specific pathogen-free (SPF) conditions and used at 7–8 weeks of age.

Spleen, inguinal lymph nodes (LNs) and thymus cell suspensions were obtained by pressing with a syringe plunger through a 40 µm cell strainer (Biologix Research), and erythrocyte-depleted by using red blood cell lysis buffer (eBioscience).

C57BL/6-derived B16.F0, MCA205, EL-4, and RMA-S tumor cell lines were obtained from Dr R. Alemany (Institut Català d’Oncologia, L’Hospitalet de Llobregat, Spain) and Dr P. Lauzurica (Instituto de Salud Carlos III, Madrid, Spain). B16.F0 cells were grown in Dulbecco's Modified Eagle medim (DMEM, Invitrogen) supplemented with 10% fetal bovine serum (FBS) and 100 IU/mL penicillin/streptomycin. MCA205, EL-4, and RMA-S cells were grown in Rosswell Park Medium Institut (RPMI) 1640 medium (Lonza) supplemented with 10% FBS, 10 mM HEPES, 2 mM glutamine, 1 mM sodium pyruvate and 100 IU/mL penicillin/streptomycin.

Production and purification of endotoxin-free purified recombinant shCD6 and shCD5 proteins (in PBS with 10% glycerol, pH 7.4) from SURE CHO-M cell line clones was done following previously reported methods.23 Human (HSA) and bovine (BSA) serum albumin were from Sigma-Aldrich.

The codon-optimized cDNA sequence encoding the whole ectodomain of human CD6 (GenBank X60992.1) from the initiation M1 to R373 was synthesized at GenScript (New Jersey, USA) and cloned into the BamHI site of the p1026x vector (kindly provided by Dr R.M. Perlmutter), just downstream of the lck proximal promoter and the Ig heavy chain enhancer (Eμ).17 Transgenic mice were generated at PolyGene (Rümlang, Switzerland) by injecting the purified NotI-NotI fragments into zygotes from superovulated C57BL/6N female mice mated to C57BL/6N breeder males (Charles River). Injected zygotes were cultured overnight and transferred into pseudopregnant B6xCBAF₁ females (Charles River). PCR screening for transgene integration identified 2 positive out of 14 pups. One of them was bred to homozygosity and maintained in parallel with WT mice from the same original cross. Tail genomic DNA was extracted by the KAPA Mouse Genotyping Kit (Kapa Biosystems) and subjected to PCR amplification with primer pairs specific for the CD6 ectodomain (Fw 5′-ACC TGA CCA GCT CAA CAC-3′ and Rv 5′-GTC ATC GCA CAC TGA TCC-3′) and the MHC class II invariant Ii chain (Fw 5′-TCA CTC AAG GCA ACC TTC CTGC-3′ and Rv 5′-CGA CCT CAT CTC TAA CCA TGA ACAG-3′), resulting in 495 bp and 150 bp fragments, respectively.

Construction and production of recombinant AAV8 virus expressing soluble mouse CD6 (AAV-smCD6) or luciferase (AAV-Luc) proteins under the control of a chimerical liver-specific (EalbAAT) promoter was carried out as previously reported.23

The plasma levels of shCD6 and smCD6 were quantified with the human soluble CD6 PicoKine ELISA Kit (Boster Bio) and the DuoSet mouse CD6 kit (R&D Systems), respectively, following manufacturer’s instructions.

Surface stainings were performed by blocking cells in phosphate-buffered saline (PBS) plus 10% FBS for 30 min at 4°C with rat anti-mouse CD16/32 (2.4G2; Mouse BD Fc Block) and further incubation with the following labeled mAb and appropriate isotype controls: biotin-labeled anti-BP-1 (6C3, eBioscience); fluorescein isothiocyanate (FITC)-labeled anti-CD4 (RM4-5, Tonbo Biosciences), anti-B220 (Ra3-6B2, Tonbo), anti-CD44 (IM7, BD Biosciences), and anti-CD24 (M1/69, BD Pharmingen); phycoerythrin-labeled anti-IgD (11-26, SouthernBiotech), anti-NK1.1 (PK136, BD Biosciences), anti-CD8 (53-6.7, BioLegend), anti-FoxP3 (3G3, Tonbo), anti-CD19 (1D3, Tonbo), and anti-ALCAM (eBioALC48, eBiosciences); PerCP-Cy5.5-labeled anti-CD3 (145-2 C11, Tonbo), anti-CD8 (53-6.7, Tonbo), anti-CD44 (IM7, Tonbo), and streptavidin (BD Biosciences); allophycocyanin-labeled anti-CD3 (145-2 C11, Tonbo), anti-CD62L (MEL-14, Tonbo), anti-CD3ε (145-2 C11, Tonbo), anti-IgM (II/41, eBioscience), anti-LAP (TW7-16B4, Biolegend), and anti-CD25 (PC61.5, eBiosciences); APC-Cy7-labeled anti-CD25 (PC61, BD Pharmingen); Pe-Cy7-labeled anti-CD5 (53–7.3, eBiosciences), anti-CD19 (1D3, Tonbo), and anti-CD4 (GK1.5, BioLegend); violetFluor 450-labeled anti-B220 (RA3-6B2, Tonbo) and anti-CD8 (53-6.7, Tonbo). FoxP3 intracellular staining was performed with the phycoerythrin-labeled anti-mouse/rat Treg Staining Kit (eBioscience), according to the manufacturer’s instructions. Data were analyzed in a FACSCanto II flow cytometer (Becton Dickinson) using FlowJo software (Tree Star).

Detection of in vivo proliferating cells with bromodeoxyuridine (BrdU) was performed as reported elsewhere.24 Treg isolation and suppressive activity assays were performed as previously reported.25

Naïve CD4⁺ T cells (CD4⁺CD25^-CD62L^hiCD44^lo; 1×10⁵) from C57BL/6/J spleens sorted with Dynabeads Untouched Mouse CD4 cells Kit (Thermo Fisher Scientific) were activated for 72 hours in triplicate in 96-well U-bottom plates with coated anti-CD3 (2 µg/mL) plus soluble anti-CD28 (0.5 µg/mL) mAb under standard Treg, Th1, Th2 or Th17 polarization conditions (see https://BioLegend.com/en-us/technical-protocols) in the presence or absence of shCD6 protein (0–10 µg/mL). Cells were stained for surface CD4 (RM4-5) and intracellular FoxP3 (3G3) expression for Treg analysis. For Th1, Th2 and Th17 analysis, cells were restimulated for 5 hours with phorbol myristate acetate (PMA,80 nM) and ionomycin (1 µg/mL) in the presence of 2 µM monensin followed by surface CD4 and intracellular IFN-γ (XMG1.2, Tonbo), IL-4 (11B11, BD Biosciences) or IL-17 (eBio17B7, eBioscience) staining, respectively.

Assessment of local subcutaneous growth and lung metastatization of syngeneic tumor cells in vivo was performed as reported elsewhere.25 26

Total mRNA (0.5 µg) extracted from cell lines by using the PureLink RNA Mini Kit (Ambion, Life Technologies) was retrotranscribed into cDNA by using the high-capacity cDNA Kit (Life Technologies). Galectin mRNA levels were assessed by quantitative real-time PCR (qRT-PCR) analysis (Taqman Fast Universal PCR Master Mix; Life Technologies) using a 7900HT fast real-time PCR System (Applied Biosystems) and the following Taqman probes (Thermo Fisher): Gal-1 (Mm00839408_g1), Gal-3 (Mm00802901_m1) and GAPDH (Mm99999915_g1). Galectin mRNA expression was adjusted by GAPDH expression as 2^ΔCt, where ΔCt=Ct (GAPDH) − Ct (gene of interest).

C57BL/6J splenocytes (1×10⁵) activated for 72 hours in 12-well plates with coated anti-CD3 (145–2 C11; 1 µg/mL, Tonbo) plus soluble anti-CD28 (37.51; 5 µg/mL, Tonbo) mAb were further suspended in FBS-free RPMI medium for 30 min at 37°C in the presence of 100 µg/mL of endotoxin-free glutathione S-transferase (GST) and fusion GST-Gal1 and GST-Gal3 proteins, produced and purified as reported elsewhere.5

Apoptotic cells were detected using the Annexin V-FITC apoptosis detection kit following the manufacturer’s instructions (ImmunoStep).

Tumor cell proliferation was monitored as reported elsewhere.27 Wound healing technique was used to assess tumor cell migration.28

Statistical analyses and data representation were performed with GraphPad Prism V.6 Software.

CD6-mediated interactions were investigated in vivo by generating a transgenic mouse line (shCD6LckEμTg) expressing the soluble ectodomain of human CD6 (shCD6) in a C57BL/6N background. The transgene was under control of lymphocyte-specific transcriptional elements (lck proximal promoter and Eμ enhancer) (figure 1A), which drive shCD6 expression mainly to primary and secondary lymphoid organs. Identification of transgenic mice was performed by PCR amplification of a 495 bp fragment corresponding to the human CD6 ectodomain (figure 1B), and ELISA measurement of plasma shCD6 levels, which yielded 79.2±19.7 and 38.7±6.3 ng/mL for homozygous and heterozygous mice, respectively (figure 1C).

Primary lymphoid organ analyses (online supplementary figure S1A and S1B) showed no differences between shCD6LckEμTg and NonTg mice regarding thymus and bone marrow total cell numbers, as well as the frequency of all B-cell developmental subsets (figure 1D, left and middle). In contrast, shCD6LckEμTg presented significantly higher fractions of double-negative (DN) and CD8 single-positive (CD8SP) thymocytes (figure 1D, right), suggesting a role for CD6-mediated interactions in T cell but not B-cell developmental process.

The analysis of peripheral lymphoid organs revealed decreased total cell numbers in spleen and LN from shCD6LckEμTg mice compared with NonTg controls (figure 1E, left). Annexin V and 7-AAD staining of total spleen and LN cells showed no differences between shCD6LckEμTg and NonTg mice regarding cell apoptosis levels (figure 1E, right). In vivo BrdU-incorporation assays revealed a decrease in the percentage of BrdU⁺ cells in spleen and LN from shCD6LckEμTg mice (figure 1F). A more detailed analysis (online supplementary figure S1C and S1E) showed that lower BrdU incorporation affected the B cell (CD19⁺) compartment in both spleen and LN. These results indicate that a lower proliferative capacity of B cells is behind the reduced cellularity found in the peripheral lymphoid organs of shCD6LckEμTg mice.

Another consistent finding in peripheral (spleen and LN) but not central (thymus) lymphoid organs from shCD6LckEμTg mice was the presence of decreased percentages of CD25⁺FoxP3⁺ cells within the CD4⁺ subset (Treg) (figure 2A and online supplementary figure S2). Moreover, Treg cells from shCD6LckEμTg mice showed lower expression levels of latency-associated peptide (LAP) (figure 2B), a well-known surface marker directly correlating with Treg functionality.29 In agreement with this finding, in vitro suppression assays confirmed that CD4⁺CD25^+/hi T cells (Treg) from shCD6LckEμTg mice exhibited lower suppressive activity on activated CD4⁺CD25⁻ T cells (Tconv) at high Tconv:Treg (8:1) ratios than those from NonTg mice (figure 2C).

The putative effects of shCD6 on T cell polarization were further explored in vitro by culturing naïve CD4⁺ T cells (CD4⁺CD25^-CD62L^hiCD44^lo) from WT mice under Th1, Th2, Th17 and Treg cell polarization conditions in the presence of increasing doses of shCD6 (0.1, 1 and 10 µg/mL). As illustrated in figure 2D, dose-dependent reduction in the percentage of polarized Treg but not Th1, Th2 or Th17 cells was observed. Moreover, a reduction in the surface expression levels of LAP was observed in in vitro polarized Treg cells in the presence of shCD6 (10 µg/mL) (figure 2E), pointing to a role for shCD6 on the reduced frequency and/or functionality of peripheral Treg cells in shCD6LckEμTg and shCD6-treated mice.

To assess the in vivo functional relevance of our immunophenotypical observations, shCD6LckEμTg mice were challenged by subcutaneous injection with syngeneic cancer cells. As shown in figure 3 (left), growth of melanoma B16.F0 tumor cells was slower in shCD6LckEμTg mice than in NonTg controls. Tumor weight at the end of the follow-up period (day 20) was not statistically significant. Similar analyses performed with syngeneic tumor cells of fibrosarcoma (MCA205) and lymphoma (RMA-S) origin also showed reduced tumor growth and weight for shCD6LckEμTg mice (figure 3, middle and right), indicating an enhanced and non-cell lineage-specific antitumor response.

In order to rule out transgenesis artifacts, immunophenotypical analyses were performed following repeated administration of shCD6 protein (1.25 mg/kg, every-other-day intraperitoneal for 15 days) to WT mice. Thymus analyses showed no differences between shCD6-treated and vehicle-treated mice regarding total cell numbers (figure 4A) and the frequency of T-cell developmental subsets (online supplementary figure S3). In the periphery, statistically significant reductions of total cell numbers were detected in spleen but not LN from shCD6-treated mice (figure 4A). The percentage of CD25⁺FoxP3⁺ cells within the CD4⁺ subset was also found significantly decreased in LN but not spleen samples from shCD6-treated mice (figure 4B). Thus, infusion of shCD6 on alternate days for a limited period of time reproduced some of the immunophenotypical changes observed in shCD6LckEμTg.

To investigate whether infusion of shCD6 protein also delays tumor growth in vivo, WT mice were subcutaneously implanted with B16.F0 cells and then intraperitoneally infused with 1.25 mg/kg shCD6 every-other-day from the day of tumor implantation. As illustrated in figure 5, shCD6-treated mice showed a statistically significant reduction in B16.F0 tumor cell growth compared with vehicle-treated mice. Similar results were observed when shCD6-treated mice were challenged with MCA205, EL-4 or RMA-S cells (figure 5), indicating a role for circulating shCD6 in delayed tumor growth in shCD6LckEμTg mice in vivo.

As an approximation to clinical conditions, we explored the effect of every-other-day shCD6 infusion to WT mice bearing ~5–9 mm² B16.F0 tumors (~day 9 post subcutaneous implantation). Again, reduced B16.F0 tumor growth was observed in mice therapeutically treated with shCD6 protein (online supplementary figure S4).

The antitumor responses of WT mice transduced with hepatotropic AAV for sustained circulating smCD6 protein levels were next explored. Previous data from our group show that AAV-smCD6 transduced WT mice present detectable smCD6 levels from day 10 on, which peak at days 15 to 20, and remain sustained for longer than 3 months.23 We reasoned that smCD6 would be less immunogenic than shCD6 in mice and its long-term expression and function would not be impaired by newly appearing neutralizing antibodies. Mice transduced with optimal AAV-smCD6 doses (10¹¹ viral genome (v.g.)/mouse), as determined by dose–response experiments (online supplementary figure S5A), do not displayed delayed growth of subcutaneously implanted B16.F0 tumors compared with control mice (AAV-Luc) (figure 6A). In contrast, significantly lower number of lung metastases and improved survival was observed when AAV-smCD6-treated mice were challenged intravenously (tail vein) with B16.F0 cells (figure 6B and C). As illustrated in figure 6D, the effects of AAV-smCD6 mouse transduction on lung metastatization were dose-dependent, with optimal results at 10¹¹ v.g./mouse. AAV-smCD6 treatment also resulted in reduced numbers of liver metastases when B16.F0 cells were injected into spleen (online supplementary figure S5B).

The possibility that AAV-mediated effects could be in part responsible for the reduced metastatization observed was further ruled out by performing lung metastases assays in which only purified sCD6 protein was infused. In order to reproduce as much as possible the conditions used in the AAV experiments (prophylactic mode and high circulating sCD6 levels), WT mice were infused intraperitoneally with 1.25 mg/kg shCD6 or HSA 1 hour before and after intravenous challenge with 2×10⁵ B16.F0 cells preincubated for 15 min with shCD6 or HSA. Under such experimental conditions, lower number of lung metastases was observed in shCD6-treated mice compared with HSA-treated controls (online supplementary figure S6).

Abnormal tumor expression of Gal-1 and/or Gal-3 is known to influence tumor development, progression and metastasis through different mechanisms, one of them involving T-cell apoptosis.30 All tumor cell lines used in this work express Gal-1 and/or Gal-3 at different levels, as deduced from mRNA analyses (figure 7A). Since CD6 is also known to interact with Gal-1 and Gal-3,5 the influence of shCD6 on galectin-induced T-cell apoptosis was further investigated. To this end, 72-hour activated spleen T cells were exposed to GST-Gal-1 and GST-Gal-3 in the presence of increasing concentrations of shCD6 (10 and 20 µg/mL). As shown in figure 7B, dose-dependent reduction in Gal-1 and Gal-3–induced apoptotic T cells was induced by shCD6, indicating that shCD6 may enhance resistance of tumor-specific effector T cells to apoptosis mediated by either tumor or stroma cell-derived galectins.31

Extensive evidence indicates that ALCAM functions as a risk factor in cancer, with its different functional isoforms (soluble and membrane-bound) correlating with specific clinical outcomes depending on cancer type.32 Moreover, the interference of homophilic ALCAM–ALCAM interactions by mAbs or chimeric proteins has been shown to reduce tumor cell growth in vitro and in vivo.27 33 34 In light of these data, we analyzed the effect of shCD6 on in vitro growth of the tumor cell lines used in this study, which show different surface ALCAM levels (figure 7C and online supplementary figure S7). As illustrated in figure 7D, dose-dependent effects were induced by shCD6 but not HSA on the proliferation of tumor cell lines expressing higher ALCAM surface levels (B16.F0, EL-4, and MC205) compared with RMA-S cells, whose proliferation remained unaffected. Similar inhibitory effects were observed for shCD6 when B16.F0 cell migration was explored by performing wound healing assays (figure 7E).