Screening for new combinations of cell markers for human SSCs and germinal progenitors characterisation

We recently identified a spermatogonial subpopulation highly enriched in SSCs with the phenotype Side population(SP)β2M−ITGA6+THY1+ population using a 5-color panel [16]. Based on this set of markers, we aimed to add new ones to apply a multi-parameter flow cytometry strategy to further characterise human spermatogenesis and improve the delineation of SSCs identity. First, TSPAN33, FGFR3 and SSEA4, previously described to be expressed on SSCs and primitive spermatogonia, were added to our β2M, ITGA6, and THY1 panel of markers, along with the KIT receptor allowing to discriminate differentiating from primitive spermatogonia [6, 9, 16,17,18]. Side population marker has been omitted from this 8-C panel in order to offer new possibilities of fluorochromes to be used on the violet or UV lasers. In fact, the use of Vybrant or Hoechst 33342 DNA dye prevents the detection of a higher number of markers. The 8-C panel is defined as 7 cell surface markers and one viability marker. Viable cells were analyzed using forward and side scatter to remove elongated spermatids and sperm from further analyses, and single cells were selected on FSCA-FSCW (Fig. 1A). After gating for β2M− cells to remove somatic cells, we were able to resolve a subpopulation of ITGA6+ cells expressing SSEA4, FGFR3, or TSPAN33. SSEA4+ population splits into 2 subpopulations according to the expression of KIT (Fig. 1A). The β2M−ITGA6+THY1+ population contains the spermatogonia expressing SSEA4, FGFR3, and the fraction of cells expressing TSPAN33 (Fig. 1B). TSPAN33 expression in the SSEA4+ population clearly discriminated a subpopulation representing 0.26% of germ cells that co-expressed THY1 but were negative for the KIT receptor (Fig. 1C). SSEA4+TSPAN33− cells are a mixture of THY1+KIT− cells and THY1−KIT+ cells.

Fig. 1

Characterisation of SSCs and spermatogonial progenitors using 8-C fluorescence panel. a Representative flow cytometry plots of the 8-C panel analysis on the human spermatogonial progenitors according to forward (FSC) and side scatter (SSC), β2 microglobulin, ITGA6, SSEA4, KIT, THY1, FGFR3, and TSPAN33 parameters. Viable propidium iodide (PI)-negative cells were gated based on morphology to remove cellular debris and doublets. Population frequencies are shown in the graph. b Expression of SSEA4, FGFR3, or TSPAN33 in the β2M−ITGA6+THY1+ population. c Flow cytometric cytogram of TSPAN33 and SSEA4 markers in the β2M− germinal population, and THY1 and KIT expression in SSEA4+TSPAN33+, SSEA4+TSPAN33−, SSEA4−TSPAN33− germinal populations. Arrows indicate the strategy of gating and analysis

Next, we decided to develop a spectral flow cytometric approach allowing to use a larger number of cell surface markers. We first searched for new markers to enrich our 8-C panel. We screened cell membrane markers previously described in transcriptomes or sc-RNAseq from human germ cell populations, or markers of murine SSCs assuming that mouse and human models share some characteristics, or markers of stem cells from other tissues [6,7,8,9,10,11, 18]. For this screening, we studied the expression of the candidate markers we selected (CD7, TSPAN8, CD47, CD77, CD24, CD155, CD184, CD148, CD51/CD61, CD130, EPHB2, CD115, CD146, CD226, CD304/NRP1, CD334/FGFR4, and CD75) in the β2M−ITGA6+THY1+ population. Among these 17 markers, a fraction of β2M−ITGA6+THY1+ cells were positive for the expression of CD47, CD155, CD7, CD130 or CD148 (Figure S1). The other markers were negative for the antibody clones we used for screening. Finally, we identified meiotic and post-meiotic populations based on their DNA content by Hoechst [16], and classified these populations according to the expression of CD148, CD155, CD47 and ITGA6 (Figure S2).

Development of a 16-C panel to characterise human spermatogenesis and discriminate SSCs populations using spectral flow cytometry and high-throughput multi-parametric analysis

To further extend the multi-parametric and in-depth characterisation of human SSCs and spermatogenesis, we associated the 7 surface markers of the first set (β2M, ITGA6, THY1, TSPAN33, FGFR3, SSEA4, and KIT receptor), 2 additional well-known spermatogonial/stem cell marker (EPCAM, CD9) [19, 20] and the 6 newly identified testicular (CD51/61) and germinal (CD47, CD155, CD7, CD130 or CD148) markers. Hence, adult testicular cell samples were profiled by spectral flow cytometry combining 15 antibodies and a viability marker (β2M, CD51/61, ITGA6, SSEA4, TSPAN33, FGFR3, THY1, CD9, EPCAM, CD47, CD7, CD148, CD155, CD130, KIT, DAPI/viability) resulting in a 16-C panel. The different signals were resolved by spectral deconvolution and the autofluorescence of the cells was extracted (Fig. 2 and Figure S3), allowing the analysis of 18 parameters (16 markers and FSC/SSC). Figure 2 shows the manual gating strategy for the 16-color panel used in this study. First, single living (DAPI-negative) testicular cells were selected and spermatozoa were removed from analysis by gating based on FSC and SSCs from 405 nm laser. Testicular cells were then divided according to somatic cell marker β2M, and the integrin complex CD51/CD61 known as αVβ3. CD51/CD61 is expressed in testicular Leydig stem cells and macrophages, and is used in the murine marker panel to negatively discriminate spermatogonia [21,22,23]. As shown in Fig. 2a, human testicular CD51/61+ cells are also positive for β2-microglobulin. Cells negative for CD51/61 and β2M markers were then gated for germ cell analysis. SSEA4+ cells expressed ITGA6 and CD9. Expression of ITGA6, THY1, CD47, TSPAN33, CD7, FGFR3, CD155 (Fig. 2a) and SSEA4 (Figure S3) was detected in EPCAMmed cells, intermediate expression of EPCAM being a marker of primitive spermatogonia [7]. SSEA4+ cells were positive for the CD148 marker, and the majority did not express the KIT receptor, although KIT was found in cells with medium expression of SSEA4. We then subdivided the spermatogonial populations using SSEA4 and TSPAN33 markers (Fig. 2a). The other markers were analysed according to the defined SSEA4+TSPAN33+ (0.09% of the germ population-Fig. 2b), SSEA4+TSPAN33− (1.1% of the germ population-Fig. 2c) and SSEA4− (Fig. 2d) cell populations. We observed EPCAMmed cells with higher expression levels of CD7, CD47, THY1, CD148, FGFR3 and CD155 in the SSEA4+TSPAN33+ population when compared to SSEA4+TSPAN33− cells.

Fig. 2

Flow cytometry profiling of adult testicular cell samples using the 16-C panel. a Representative flow cytometry plots of the spectral flow analysis on the human spermatogonial progenitors according to FSC, SSC, β2M, CD51/CD61, ITGA6, THY1, SSEA4, and TSPAN33 parameters. Testicular cells were gated based on morphology to remove cellular debris, sperm and doublets. b–d Expression of CD7, CD155, CD148, CD47, and KIT markers in the β2M/CD51/CD61−SSEA4+TSPAN33+ (b), the β2M/CD51/CD61−SSEA4+TSPAN33− (c), and the β2M/CD51/CD61−SSEA4−TSPAN33− (d) germinal populations. Arrows indicate the strategy of gating and analysis. Figure S4 shows the most relevant fluorescence minus one (FMO) controls. ITGA6 is marked AL6 in the cytogram axis

To allow comprehensive analysis of the resulting multi-dimensional dataset, the total viable testis cells (341 378 cells) from adult obstructive azoospermia patient with normal spermatogenesis were first projected into the same UMAP space (Fig. 3a). The spermatogonial, meiotic and postmeiotic cell subsets defined on flow cytometric plots, as shown in Fig. 2 and Figure S3, were manually gated, projected and overlaid on the UMAP dimensionality reduction plot (Fig. 3b). All these cell subsets overlapped with well-defined clusters on the UMAP and allowed to define the cell clusters. We discriminated somatic cell populations (β2M+CD51/61+ and β2M+CD51/61− cells) from germ populations. We distinguished meiotic spermatocytes I (4N), and spermatocytes II (2N) and postmeiotic (N) cells from spermatogonial populations. In addition, primitive spermatogonial populations (ITGA6+THY1+ and ITGA6+SSEA4+ cells) were clearly separated from differentiating spermatogonia, demonstrating that the dimensionality reduction visualization was able to separate minor populations of spermatogonia. Heatmap overlay of all the markers on the UMAP plot shows that SSEA4, ITGA6, THY1, CD7, TSPAN33, CD47, EPCAMdim, CD9, FGFR3, CD148 and CD155 markers, but not KIT, were expressed in the primitive spermatogonial ITGA6+THY1+ subset (Fig. 3c). We also observed a continuum of differentiation from the primitive and differentiating spermatogonial clusters to spermatocytes I (4N), spermatocytes II (2N) and spermatid cells (Fig. 3a, b). Unsupervised clustering of β2M+CD51/61− germ cells made by FlowSOM confirmed our manual clustering of germ cells (Figure S5). A FlowSOM tree and a trajectory inference were derived from the unsupervised clustering of the β2M−CD51/61− germinal cells. The FlowSOM tree recapitulated the successive steps of the differentiation process (Fig. 4a and Figure S5). The resulting heatmap from the trajectory inference shows the expression levels of the different markers along this trajectory (Fig. 4b). Based on the expression of markers (Figure S5), the germline clusters were overlaid on the trajectory (Fig. 4b). This confirmed that the trajectory inferred from the unsupervised analysis of the 16-C panels was successful in describing the spermatogonial differentiation.

Fig. 3

Comprehensive analysis of multidimensional 16-C panel dataset of human testicular cells. a High-dimensional UMAP analysis of human testicular cells, (b) Testicular cell subsets defined in Fig. 2 and Fig. S3 are indicated on the UMAP. Spermatocyte I (SPI), spermatocyte II (SPII), spermatids (Sd), and the “S phase” subpopulation (differentiating spermatogonia) were assigned according to the expression of CD148, CD47, CD155 and ITGA6 as previously defined in Figure S2. c Expression patterns of individual spermatogonial markers overlaid onto the UMAP representation of human testis cells. The colour indicates the fluorescence intensity of the marker (red: high to blue: neg)

Fig. 4

Trajectory analysis of the human germ cell differentiation dataset. a FlowSOM tree inferred from unsupervised clustering of β2M+CD51/61− germinal cells using FlowSOM analysis. Germinal subsets are indicated on the FlowSOM tree plot. b Heatmap and Wanderlust trace of the expression of the germinal markers after Wanderlust trajectory inference. The different differentiation stages are indicated on the heatmap and Wanderlust trace to show the expression patterns of each differentiation state

In-depth characterisation of human SSCs and progenitors using the 16-C marker panel

We then focused the analysis on the whole β2M−CD51/61−ITGA6+ spermatogonial population and the β2M−CD51/61−ITGA6+THY1+ primitive spermatogonia. Adult testicular cell suspensions from 3 patients with normal spermatogenesis were phenotyped and the data were concatenated to include a higher number of spermatogonia in the analysis. 141 451 β2M−CD51/61−ITGA6+ were projected into UMAP spaces (Fig. 5a), and the major spermatogonial subsets obtained by manual gating were overlaid on this UMAP plot (Figs. 5b). Three main cell populations were delineated (1) a primitive spermatogonial cluster consisting of ITGA6+SSEA4+ and ITGA6+THY1+ populations, with SSEA4+TSPAN33+ cells at the edge of this cluster, (2) the ITGA6+KIT+ differentiating spermatogonial cluster making a cell state transition, (3) towards the third cluster of more advanced differentiated ITGA6+SSEA4− spermatogonia. Unsupervised FlowSOM analysis allowed to distinguish 6 main clusters (clusters 1, 5, 6, 7, 8 and 10) according to the expression of the 13 markers (Figure S6), allowing to distinguish primitive spermatogonia (clusters K6, K8 and K10) from more advanced spermatogonial progenitors (clusters K1, K5 and K7). Heatmap overlay of all the 13 markers on UMAP plot showed a differential expression of CD47, TSPAN33, CD7, THY1, SSEA4, ITGA6, EPCAM, FGFR3, CD148, CD9 and CD155 markers in the primitive spermatogonial cluster (Figs. 5c, d), allowing the discrimination of different subpopulations in this cluster, with in particular, at its lower edge a prominent expression for CD47, TSPAN33, CD7, ITGA6, FGFR3, CD148, CD9 and CD155, and a medium expression of EPCAM. Higher expression of EPCAM, CD148, CD9, and CD130 was also observed in the more advanced differentiated ITGA6+KIT+ and ITGA6+SSEA4− spermatogonia, together with a lower ITGA6 expression, in line with a progression of spermatogonial progenitors in the differentiation process.

Fig. 5

Comprehensive analysis of multidimensional 16-C panel dataset of ITGA6+ spermatogonia. a High-dimensional UMAP analysis of human ITGA6+ spermatogonia, (b) Testicular cell subsets defined in Fig. 2, Fig. S2 and Fig. S3 are indicated on the UMAP. c Expression patterns of individual spermatogonial markers overlaid onto the UMAP. Colour indicates fluorescence intensity. d Heatmap of differentially expressed markers in the spermatogonial populations

When the 9074 β2M−CD51/61−ITGA6+THY1+ were projected on UMAP (Fig. 6), we observed mainly two clusters, one consisting of SSEA4+TSPAN33+, SSEA4+TSPAN33+CD47high, and SSEA4+ TSPAN33+CD7high cells on the left side, and one consisting of SSEA4+TSPAN33− cells and ITGA6+CD155med on the right side (Figs. 6a–c). Heat maps of marker expression shows a progressive expression of the CD7 marker from high to low levels as one progresses from SSEA4+TSPAN33+ cluster to SSEA4+TSPAN33− one (Fig. 6d). SSEA4+TSPAN33+ cells show the higher expression of CD47, CD7, CD9, CD155 and CD148 markers and intermediate expression of CD130 (Fig. 6d). FGFR3 expression was found in cells at the border of the two clusters, suggesting a transitional step between the two states. Hence, a primitive spermatogonial population with the phenotype β2M−CD51/61−ITGA6+SSEA4+TSPAN33+THY1+CD9medEPCAMmedCD155+CD148+CD47highCD7high is detected (as defined by the marker expression levels in the ITGA6+ population analysis, Fig. 5). Unsupervised clustering of β2M−CD51/61−ITGA6+THY1+ cells by FlowSOM mainly identified three clusters K1, K3 and K4 (Figure S7b), SSEA4+TSPAN33− population being divided in two clusters K1 and K4. The expression profile of K3 corresponded to an immature phenotype (expression of TSPAN33, CD47, CD7, EPCAM, THY1, and FGFR3), K1 to a more mature phenotype (lower expression of TSPAN33, CD47, CD7, SSEA4, THY1, FGFR3, ITGA6, EPCAM, CD155, CD148, and higher expression of KIT and CD130), K4 to an intermediate phenotype (Figure S7c). Surprisingly, the K1 cluster was placed between the K3 and K4 clusters in the FlowSOM tree, although cells from the K1 cluster display the most mature phenotype (Figure S7d). This finding deviates from the expected linear process starting from K3 going to K4, and then to K1 cells. We postulate that K3 and K4 cells could represent alternative stem cell states with different potential, with K3 being the least developmentally advanced SSCs state. K3 and K4 states would converge on the K1 state to commit to differentiation. The existence of different subsets of SCCs that may interconvert between them or have different potentials has yet been described [6, 9]. Finally, we investigated whether the population we defined as SSEA4+TSPAN33+ could correspond to the state 0 and state SSC1-B [6, 9] of the spermatogonial hierarchy previously defined in scRNAseq analysis [6, 9], which are subclusters of the primitive spermatogonial population. Consistent with this, we found that mRNA of C19orf84, PIWIL4, FGFR3, UTF1, and LPPR3 markers of these states were highly expressed in β2M−SSEA4+TSPAN33+ cell fractions compared to β2M−SSEA4+TSPAN33− cells sorted from adult obstructive azoospermia patient with normal spermatogenesis (Fig. 6e), which were flow sorted using the 8-C panel and ARIA flow cytometer (see Fig. 1). Therefore, the β2M−CD51/61−ITGA6+SSEA4+TSPAN33+THY1+CD9medEPCAMmedCD155+CD148+CD47highCD7high cell population should correspond to the most primitive SSCs state 0 or state SSC1-B, or include these primitive states of SSCs [6, 9].

Fig. 6

Comprehensive analysis of multidimensional 16-C panel dataset on the ITGA6+THY1+ spermatogonial population. a High-dimensional UMAP analysis of β2M/CD51/CD61−ITGA6+THY1+ cells (UMAP zoom to main cluster, full UMAP is shown in Figure S6A). b Overlay of SSEA4+TSPAN33+, SSEA4+TSPAN33−, SSEA4+TSPAN33+CD47high and SSEA4+TSPAN33+CD7high populations as defined in Fig. 2 and Fig. S3. c Spermatogonial cell subsets defined in Fig. 2 and Fig. S3 are shown on the UMAP. d Expression patterns of individual spermatogonial markers overlaid onto the UMAP. Colour indicates fluorescence intensity (red: high to blue: neg) (e) Expression of C19orf84, PIWIL4, FGFR3, UTF1 and LPPR3 markers in sorted viable β2M−SSEA4+TSPAN33+ (TSPAN33 +) and β2M−SSEA4+TSPAN33− (TSPAN33-) populations (n = 3)