Lentiviral vectors and transduction of human cancer B cells

To the editor:

A recent article by Hazan-Halevy et al assessed down-regulation of STAT3 in B-CLL cells (B-CLLs) using VSV-G pseudotyped lentiviral vectors (VSV-G-LVs) encoding for STAT3–small hairpin RNA (shRNA).¹  They reported a transduction of 40% to 70% with VSV-G-LVs in nonstimulated B-CLLs. This is in sharp contrast with our recently published data demonstrating that VSV-G-LVs are not suited for efficient transduction of malignant B cells. Indeed, 4 independent laboratories reported that human B cells, even upon a proliferative stimulus, are highly refractory to VSV-G-LV–mediated gene transfer.^2-5  We confirmed that even upon BCR stimulation, B-CLLs were not efficiently transduced by VSV-G-LVs as reported earlier by Bovia et al.²  Of high importance, we engineered a new generation of LVs carrying at their surface glycoproteins of the measles virus, H and F (H/F-LVs). These new LV-pseudotypes allowed very efficient gene transfer in resting T and B cells as well as in B-CLLs, whereas VSV-G-LVs failed, even at high vector doses.^3,6 

To understand the discrepancy between our results and those of Hazan-Halevy and colleagues,¹  we produced green fluorescence protein (GFP)–encoding VSV-G- and H/F-LVs by ultrafiltration as reported by these authors. We transduced primary B-CLLs⁷  and another B-cell malignancy, marginal zone lymphoma cells (MZLs)⁸  with volumes of vectors concentrated by ultrafiltration and cell densities equivalent to those of Hazan-Halevy et al.¹  Upon 48 hours' transduction, we evaluated the green fluorescence of the cells by FACS analysis (Figure 1A). We clearly detected a shift of transduced cells (H/F or VSV-G-LVs), to a higher mean fluorescence intensity in GFP channel similar to the shift observed by Hazan-Halevy et al. Surprisingly though, 2 consecutive washes of the cells after transduction reduced this GFP⁺ signal almost completely to background levels for VSV-G-LV transduction, whereas a residual GFP⁺ population remained visible upon H/F-LV transduction (Figure 1A). The same picture was found when polybrene was added during transduction. Moreover, when continuing the B-CLL and MZL cultures for 7 days on MS5 cells in the presence of cytokines, only HF-LV transduction resulted in stable GFP expression, whereas no GFP was detected for VSV-G-LV transduction (Figure 1A). Moreover, we confirmed a very low level, close to background, of reverse transcribed (RT) proviral DNA in VSV-G-LV–transduced cells, whereas a 100- to 1000-fold higher RT-DNA level was detected in H/F-LV–transduced cells (Figure 1B), again confirming unproductive VSV-G-LV transduction.

We speculate that the ultrafiltration resulted in vector concentration accompanied by strong protein concentration including GFP present in the viral supernatant. Incubation of B-CLLs or MZLs with ultrafiltration-concentrated LVs may lead to sticking of proteins (including GFP) to the cell surface, resulting in a green fluorescent signal. Note that we use low-speed concentration for our LVs, for which this phenomenon was not observed.^3,6  Clearly, these results were similar for multiple donors: we transduced nonstimulated B cells of 12 different B-CLL donors (Figure 1C; and Frecha et al³ ) and 6 MZL donors (Figure 1D) with low-speed concentrated VSV-G-LVs (multiplicity of infection [MOI] 50) and H/F-LVs (MOI 5). Indeed, VSV-G-LVs do not permit an efficient transduction while H/F-LVs permit high transduction coinciding with signaling-lymphocyte-activating-molecule expression, one of the measles virus receptors (Figure 1C-D; and data not shown^3,6 ). This poor gene transfer profile for VSV-G-LVs was also evident when the cells were prestimulated through the B-cell receptor (Figure 1E-F), whereas HF-LV transduction augmented significantly.³ 

In conclusion, HF-LVs are the candidates of choice for transducing B-CLLs or MZLs, and are highly superior to VSV-G-LVs.