Multiple tracer injections along the mediolateral extent of the dorsal aspect of the third tier cortex

We first investigated whether the lower quadrant representation just medial to upper field DM shows similar or different inter-areal connection patterns to upper field DM. Similar connectivity patterns would suggest that these two visual quadrant representations belong to a single area DM, while different connectivity patterns would indicate that they are parts of different visual areas. To this goal, in case M265, we made three closely spaced injections of different tracers along the mediolateral extent of dorsal cortex just anterior to the medial half of V2d. Fig. 3 illustrates this case. In panel (A), we show the CO-staining pattern in unfolded and flattened visual cortex, with light delineation of the V1/V2 and MT borders based on CO staining, and the CO transition zone at the anterior border of V2. The same CO image is shown again in panel (B) with overlaid injection sites, plots of their resulting cell label, and delineation of areal boundaries. In panel (C), we show the quantitative distribution of labeled cells across extrastriate cortex (except for V2 and DM) resulting from each injection site. Similar conventions are used for all the remaining figures. In this case, FB (blue) was the most medial injection, followed in lateral progression by an injection of CTBg (red) and an injection of CTB488 (green). The CTB488 injection produced label primarily in regions representing the upper visual quadrant, such as V1v, V2v, ventral MT and the ventral halves of extrastriate areas located between V2v and MT (areas VLP, VLA, MTc, as defined by Rosa and colleagues). In many areas, this injection produced an additional small patch of label at near-foveal eccentricities, near the HM representation of the lower visual quadrant (white arrows in Fig. 3B). V1 label resulting from this injection site was located primarily in layers 4B and, less so in 4A (Fig. 4A), a laminar pattern previously shown to result from injections in DM (Rosa et al., Reference Rosa, Palmer, Gamberini, Burman, Yu, Reser, Bourne, Tweedale and Galletti2009; Jeffs et al., Reference Jeffs, Federer, Ichida and Angelucci2013) (see also Materials and Methods, Assessment of Injection Site Location). In summary, both the topography and V1 laminar location of transported label resulting from this CTB488 injection indicated that the injection site was located mostly in upper field DM, but slightly straddled the foveal region of the lower quadrant representation just medial to it (for reference see visuotopic maps of DM in Fig. 1B). Moreover, in areas with well-defined retinotopy such as V1, V2, and MT, the resulting label in ventral cortex occupied much of the territory between the two meridian representations at areal borders, but did not reach these borders, suggesting that the injection site did not extend to the DM anterior or posterior borders. The CTB488 injection also produced a narrow strip of patchy label lining the HM representation at the border between V2d and the lower quadrant representation of the third tier cortex medial to upper field DM (yellow arrows in Fig. 3B). This region corresponds approximately to eccentricities between 6° and 16° (for reference, see visuotopic maps of V2 in Fig. 1). In general, we found these projections after tracer injections located almost anywhere within upper field DM and cortex anterior to it (in upper field DA/DI). Our interpretation is that these represent connections made across discontinuous HM representations (Jeffs et al., Reference Jeffs, Ichida, Federer and Angelucci2009) which border upper field DM posteriorly and medially (and area DA/DI medially and laterally), and are likely due to these areas' small size and large receptive fields, most of which cross these HM representations (see also Jeffs et al., Reference Jeffs, Federer, Ichida and Angelucci2013). In V2, the CTB488 injection site produced cell label that aligned with the dark CO stripes (Fig. 5); while we could not distinguish between thick and thin stripes in this V2v region, the periodicity of patchy label suggested it involved both dark stripe types.

Fig. 4. Laminar patterns of retrograde label in V1 produced by tracer injections in different cortical areas. (A) Case M265 CTB488. V1 label resulting from the CTB488 injection site in upper field DM (or DM+; green injection site shown in Fig. 3B). Plots of retrogradely labeled cells (green dots) in V1v are shown superimposed on an immediately adjacent CO-stained section (the brain was sectioned tangentially to the pial surface and the CO section was immediately deeper to the section from which the cells were plotted). Here and in panels (B–D) white dashed contours delineate layer boundaries, the white solid contour indicates the V1/V2 border, and numbers indicate V1 layers. (B) Case M265 FB. V1 label resulting from the FB injection in third tier cortex medial to DM+ (blue injection site in Fig. 3B). The top panel shows plots of cell label (blue dots) from a section immediately superficial to the CO-stained section shown, therefore the label that appears to align with layer 4C is, in fact, located in layer 4B in the more superficial section that contains the plotted cells. The black box delineates the region shown in the two bottom panels for the label present in two more superficial sections, respectively; the bottom left panel shows cells located primarily in layer 4A and a few cells in layer 3 aligned to the adjacent deeper CO-stained section, whereas the bottom right panel shows all the label that was present in layer 2/3 in a section just superficial to the section in the left panel. Scale bar in (B) applies to all panels in (A) and (B). (C) Case M293 CTBg (Table 1). V1 label resulting from a CTBg injection in a thick CO stripe of V2 (the injection site is shown in Fig. 7B of Jeffs et al., Reference Jeffs, Federer, Ichida and Angelucci2013). Cell label (red dots) was plotted from the same CO-stained section that was also reacted for CTBg. Injections in other stripe types of V2 typically produce a smaller amount of cell label in layer 4B compared to injections in thick stripes (see Federer et al. Reference Federer, Ichida, Jeffs, Schiessl, McLoughlin and Angelucci2009). (D) Case M295 CTB647 (Table 1). V1 label resulting from a CTB647 injection straddling the border between V2d and upper field DM (injection site shown in Fig. 3B of Jeffs et al., Reference Jeffs, Federer, Ichida and Angelucci2013). Plots of labeled cells (pink dots) are superimposed onto a Nissl stain of the same section. To the right of the arrow the label pattern is similar to the one resulting from injections in DM+, i.e., it is heavy in layer 4B, sparse in 4A with few labeled cells in layer 3 (as in panel A); instead, to the left of the arrow the label is heavy in layers 2/3, 4A, and 4B (resembling a combination of the V1 laminar patterns seen after DM and V2 injections). This label pattern is consistent with the interpretation that the label to the right of the arrow represents projections from upper field V1 to the portion of the injection site straddling upper field DM, whereas the label to the left of the arrow represents projections from both upper and lower field V1 across the HM representation to both the DM and V2d portions of the injection site which straddles the HM representation between these areas.

Fig. 5. CO stripe location of retrograde label in V2v produced by a CTB488 injection site in upper field DM. Case M265 CTB488. (A) Enlarged image of CO staining of portions of V2v and adjacent cortical areas from Fig. 3. (B) The same CO image with superimposed plots of CTB488 labeled cells (green dots) in V2v. Yellow arrows point at the same locations in the two images.

The more medial FB injection, instead, landed in cortex representing the lower visual quadrant, as demonstrated by the topography of resulting label which dominated in regions of V1 and extrastriate areas known to represent the lower quadrant, e.g., in V1d, V2d, dorsal MT and in dorsal cortex between V2d and MT (VLP, VLA, and MTc). In addition, in V1 and V2, FB label lay at the V1/V2 border, i.e., at the VM representation, indicating the injection site was also located near a VM representation. Like the CTB488 injection, V1 label resulting from this FB injection was located primarily in layers 4B and 4A (Fig. 4B), and only very sparse label was present in layer 3. This indicated that the injection was not located in V2, since injections in V2 produce strong cell label in V1 layer 2/3 (e.g., Fig. 4C) (Federer et al., Reference Federer, Ichida, Jeffs, Schiessl, McLoughlin and Angelucci2009), nor was it located in DA/DI, since injections in this area produce no label in V1 (see below; see also Rosa et al., Reference Rosa, Palmer, Gamberini, Tweedale, Pinon and Bourne2005; Jeffs et al., Reference Jeffs, Federer, Ichida and Angelucci2013). The V1 laminar pattern of cell label produced by the FB injection also differed from the one typically found after injections straddling the border between V2 and DM, which typically produced a V2/DM mixed laminar pattern, i.e., with dense label in layers 2/3, 4A, and 4B (e.g., the label left of the arrow in Fig. 4D). In summary, the topography and V1 laminar location of label resulting from the FB injection, as well as its distance from the V1/V2 border, all indicated that this injection was located in the third-tier cortex anterior to V2d and medial to upper field DM. The V2 CO stripe pattern was unclear in the V2d region of FB transport due to artifacts introduced by unfolding the medial aspect of the brain in this region; however, the FB label appeared to lie preferentially in the CO-dark regions of V2.

The very small CTBg injection produced label only in V1d and V2d, therefore we did not use data from this injection for the quantitative analysis. However, the topography of cell label in V1 and V2 resulting from it, and its V1 laminar distribution indicated that it was also located near a representation of the lower VM in cortex just anterior to V2d.

In Fig. 3C, we quantify the relative proportion of cell label across extrastriate cortex produced by the FB and CTB488 injection sites. The label produced by the two injections showed similar areal distribution, but was located in complementary quadrant representations. Both injections produced the largest fraction of cell label in dorsal stream areas DA/DI (49 and 20%) and MT (16%), and in parietal area PPd (16 and 28%); areas MTc, PPv, OPt, VLP, and VLA each contained between 2 and 7% of total extrastriate label, while areas MST, FST, and IT each contained <2% of total label. In summary, the similarity in areal and V1 laminar distributions of label resulted from the CTB488 and FB injection sites suggested that they lay in the upper and lower visual quadrant representations, respectively, of the same area DM.

We next compared the inter-areal connectivity patterns resulting from tracer injections in DM with those resulting from tracer injections in the lower quadrant representation located immediately lateral to upper field DM. In a second animal (case M248 – Table 1 and Fig. 6) we made four injections of different tracers in a mediolateral column spanning most of the dorsal aspect of the third tier cortex anterior to V2d. FB (blue) was the most medial injection, followed in lateral progression by injections of CTB488 (green), CTBg (red), and DY (yellow). The CTB488 injection site was located in upper field DM, near its posterior border, as suggested by the topography of resulting label that was located primarily in upper field representing regions, e.g., in V1v, V2v, and ventral MT, near the HM representations of these areas. We believe that the effective tracer uptake zone was larger than depicted in Fig. 6B, and that the injection slightly encroached into V2d. This was suggested by: 1. The V1 laminar pattern (resembling that seen after injections straddling the V2d/DM+ border, e.g., Fig. 4D), 2. the location of transported label at HM representations (e.g., at the anterior V2v border), and 3. the inter-areal pattern of label (heavier label in ventral stream areas VLA and IT than typically seen after injections confined to DM – compare to CTB488 label in Fig. 3C, see also Figs. 11A and S1B) resulting from this injection site.

Fig. 6. A column of four different tracer injections involving upper field DM and cortex medial and lateral to it reveals two distinct areas bordering V2d. Case M248. (A) CO image of unfolded and flattened visual cortex. The pale CO spot (blue arrow) is the location of the FB injection site, whereas the dark spot is the silver reacted CTBg injection site (red arrow). (B) The same CO image as in (A) is shown enlarged with overlaid injection sites and plotted cell label resulting from each tracer injection (FB, blue; CTB488, green; CTBg, red; DY, yellow). The locations of anterogradely labeled fibers are outlined. White arrow on the FB injection site indicates the direction of travel of the injection site from superficial to deeper layers. (C) Proportion of labeled cells in each extrastriate area resulting from the FB, CTB488, and CTBg injections. For abbreviations, see legend of Fig. 1. Other conventions are as in Fig. 3.

The more medial FB injection in case M248 produced label primarily in cortical regions representing the lower visual quadrant, such as V1d, V2d, and dorsal MT, near these areas' HM representation, therefore the injection site was likely located in the lower quadrant representation of DM, near its posterior border. The injection site traveled anteroposteriorly, and the V1 laminar pattern of cell label, as well as the inter-areal connection pattern resulting from this FB injection indicated that it involved layers 1–3 of lower field DM, and layers 4–6 of V2d. Accordingly, cells in V1 were labeled predominantly from the V2 side of the injection site (as feedforward projections from V1 to extrastriate areas terminate mostly in layer 4), dominating in layers 2/3 and 4A, with only few cells labeled in 4B (Table 1). Instead, the inter-areal feedback projections, which terminate most heavily in layers 1 and 6 of downstream extrastriate areas, were labeled by both the DM and the V2 side of the injection, resulting in an areal pattern typical of injections straddling the V2d/DM posterior border (Figs. 6C and S1B). In summary, the similarity in the topography and V1 laminar pattern of label resulting from the FB and CTB488 injections suggested that the two injections were located in the lower and the upper quadrant DM, respectively, and both encroached slightly into V2d. Indeed, the two injections also produced a very similar areal distribution of cell label (Fig. 6C), albeit in complementary quadrant representations, again suggesting that they involved the same area/s. Specifically, both injections labeled the largest fraction of cells in dorsal stream area DA/DI (16 and 20%), ventral stream areas VLA (14 and 23%) and VLP (21 and 9%), and parietal area PPd (13 and 15%); moderate amounts of labeled cells lay in dorsal stream areas MT (7 and 11%) and MTc (7 and 6%). The FB injection produced significant label (19%) in OPt but not in IT (0.4%); in comparison the CTB488 injection produced more label in IT (7%) but less label in OPt (6%). Both injections produced 3.6% of label in PPv, 0.1% in FST, and no label in MST. The small differences in the distribution of inter-areal label resulting from these two injections could depend on the different cortical layers they involved (see Table 1). Notice that this inter-areal pattern of label differed from that produced by the FB and CTB488 injections in DM in case M265 (Fig. 3C), in having a larger fraction of label in VLP and VLA, likely due to their intrusion into V2 (V2 injections label a large fraction of cells in these areas – see Fig. 11C).

The CTBg injection site, in case M248, produced a very different inter-areal pattern of label compared to the two more medial FB and CTB488 injection sites. This injection involved a region of lower quadrant representation at near foveal eccentricities (∼2–3 deg) near a HM representation, as demonstrated by the topography of transported label in dorsal V1 away from the V1/V2 border at near foveal eccentricities (Fig. 6B). Label in extrastriate cortex also lay preferentially in regions of lower field representation (V2d, and dorsal cortex between V2d and MT), except for patchy intra-areal label in the third tier cortex ventral to the injection site and in ventral VLP (V3v, or VP of alternative schemes). Unlike the more dorsal injections in the same case, or the FB and CTB488 injections in case M265 (Fig. 4A and 4B), this CTBg injection produced cell label that largely dominated in layer 2/3 of V1, and very sparse label in 4A and 4B (Fig. 7A, 7B, and 7E). In comparison, V2 injections, which also produce dense layer 2/3 label, typically produce more significant label in layers 4A and 4B (except for injections in pale medial stripes which result in no layer 4B label – Federer et al., Reference Federer, Ichida, Jeffs, Schiessl, McLoughlin and Angelucci2009) (Fig. 4C). Thus, the V1 laminar pattern and the topography of transported label suggested that the CTBg injection site lay in cortex anterior to V2d and lateral to upper field DM. Interestingly, the V1 cell label resulting from this injection site was predominantly located in the CO blobs (Fig. 7A–7D); using the same quantitative analysis as used previously in Federer et al. (Reference Federer, Ichida, Jeffs, Schiessl, McLoughlin and Angelucci2009), we found that 83% of cells lay within CO blobs. In V2, CTBg label resulting from this injection lay primarily in the thin and pale CO stripes (Fig. 8).

Fig. 7. Laminar and CO patterns of retrograde label in V1 produced by a tracer injection in third tier cortex lateral to DM+. Case M248 CTBg. (A) Bright-field image of CO-stained tangential section through V1 layer 2/3 showing the CO blob pattern. The same section was silver reacted to reveal CTBg stained cells; these are visible as black dots in panel (C), which shows a higher power view of the boxed region in (A). Dotted white contour/s in panels (A) and (E) indicate laminar boundaries and numbers indicate layers. (B) Dark-field image of the same section in (A) showing the pattern of CTBg cell label. Green arrows in (A, B, E) point at the same blood vessels. White arrows in (A) and (B) point at the same row of CO blobs (in A) and CTBg-labeled patches (in B). Patches of CTBg retrograde label align with the CO blobs. This is better demonstrated in panels (C) and (D). (C, D) Higher magnification of regions boxed in (A) and (B), respectively, to demonstrate alignment of patches of CTBg labeled cells (in D) with CO blobs (in A). Arrows in (C) and (D) point at the same blood vessels. Scale bar under (D) applies also to (C). (E) A CO stained section 240 µm deeper to the section in (A), showing the same V1 region as in (A, B), with superimposed plots of CTBg stained cells (black dots) from the same section. A total of 14 sections (each 40 µm thick) contained dense CTBg label in layer 2/3. In comparison, much sparser label in layers 4A and 4B was present in only 5 sections, indicating a large dominance of cell label in layer 2/3. Scale bar in (E) applies also to (A) and (B).

Fig. 8. CO stripe location of retrograde label in V2d produced by a tracer injection site in third tier cortex lateral to DM+ (in dorsal VLP). Case M248 CTBg. (A) Enlarged image of CO staining of portions of V2d and adjacent cortical areas from Fig. 6. (B) The same CO image with superimposed plots of CTBg labeled cells (red dots) in V2d and dorsal VLP, and the outline of the injection site in VLP. Yellow arrows point at the same thick CO stripes in the two images. Yellow dotted contours outline the CO stripes.

Cell counts (red bars in Fig. 6C) showed that cell label resulting from this injection site was located predominantly in ventral stream areas IT (37%) and VLA (22%), followed by MTc (16%), and DA/DI (11%); label in areas MT and PPd together amounted to only 1.7% of total label (in contrast, these two areas, together with area DA/DI, are the signature of DM injections – compare red bars in Fig. 6C with green and blue bars in Fig. 3C). Areas PPv, Opt, and FST each contained 3–5% of total label, DM 0.2% and MST contained no label.

This inter-areal pattern of label produced by the CTBg injection in case M248 differed markedly from that produced by the FB and CTB488 injections in the same case, in having a much greater fraction of label in IT and virtually no label in MT and PPd. This pattern also differed markedly from that produced by the FB and CTB488 injections in case M265, in having virtually no label in areas MT and PPd, but a much larger fraction of label in areas VLA, IT, and MTc (compare red bars in Fig. 6C with blue and green bars in Fig. 3C). This areal projection pattern also differed from that produced by the DY injection in the same case, which instead produced patterns of inter-areal and V1 labels typical of V2 injections (Table 1; see also Fig. 11C). Specifically, DY labeled cells in V1 were most numerous in layer 2/3, but significant DY label was also present in layers 4A and 4B. The DY label was located primarily in the V1 interblob regions, particularly at blob borders, a signature of tracer injections located in the thick CO stripes of V2 (Federer et al., Reference Federer, Ichida, Jeffs, Schiessl, McLoughlin and Angelucci2009). Indeed, the DY injection was primarily confined to a thick stripe. Due to significant glial label produced by this injection, we did not quantify cell label in this case. However, qualitative inspection of Fig. 6B reveals a different areal pattern of DY label compared to that produced by the adjacent CTBg injection: strong label in MT which instead had little CTBg label, and little or no DY label in areas that instead had dense CTBg label, such as VLA and many subdivisions of IT. This observation reinforces our interpretation that the CTBg injection was located in cortex anterior to V2d.

In summary, the different V1 laminar and areal distributions of label resulting from the CTBg injection compared to the more medial injections in the same case and the injections in case M265 strongly suggested that this injection lay in a cortical area different from upper field DM, and from the lower quadrant representation medial to it. We interpret this injection as being located in the lower quadrant representation of area VLP. The pattern of intra-areal label arising from this injection was also consistent with the retinotopic maps of VLP shown in Fig. 1B. Specifically, the labeled patchy intra-areal horizontal connections extended up to 5.47 mm laterally to the injection site, but only 1 mm medially to it, and appeared “cut off” at their medial edge (marked by a light dashed contour in Fig. 8B), suggesting the existence of an areal border at this location. The patchy intra-areal label had an anteroposterior width of about 2 mm reaching the VLP posterior border (the HM representation). We interpret this label as connections made across the HM representations that border dorsal VLP posteriorly and medially [according to the retinotopic maps of Rosa & Tweedale (Reference Rosa and Tweedale2000)].

Fig. 9 shows another example case that received an injection of CTB in the third tier cortex lateral to upper field DM, i.e., in dorsal VLP (case M237LH, Table 1). Compared to the CTBg injection in M248, this injection site was located at slightly more peripheral eccentricities (centered at ∼ −4°), approximately midway between the HM and VM representations at the medial and anterior VLP borders, respectively. Label dominated in lower field representing cortical areas (in V1d, V2d, and dorsal MT) at parafoveal eccentricities, extending much of the width of these areas. The inter-areal distribution of retrogradely labeled cells resembled that resulted from the VLP injection in case M248 CTBg, in that label dominated in ventral stream areas VLA (31%) and IT (17%), but was sparse in areas MT (5%) and PPd (1.4%), which are instead heavily labeled after injections located in upper field DM and cortex medial to it (i.e., lower field DM). However, compared to the VLP injection in case M248 CTBg, this CTB injection produced more significant label in parietal areas PPv (25%) and OPt (11%), but less label in areas MTc (3.5%) and DA/DI (6%). Surprisingly, and possibly related to its parafoveal eccentricity, this injection produced only very sparse label in V1, unlike the CTBg injection in case M248; this label, however showed the same laminar distribution as the V1 label in case M248 CTBg, dominating in layer 2/3 with few cells also labeled in the infragranular layers. The few V1 labeled cells lay preferentially at the borders between blob and interblob. Instead, in V2 label dominated in the CO pale stripes (both medial and lateral) and at the borders of the thick stripes (Fig. 9B). The different patterns of retrograde label produced by these two VLP injections (M237LH CTB and M248 CTBg) relative to the CO compartments of V1 and V2 suggest a modular organization within area VLP, as also suggested by the alternating pattern of dark and pale patches visible in CO staining of this cortical region (e.g., Figs. 5 and 9; see Discussion). In this respect, our third VLP injection case (M237RH FR, Table 1) resulted in anterograde label only, which aligned preferentially with the pale CO stripes. The different biases in inter-areal connection patterns resulting from different VLP injections, may reflect different placement of the injections with respect to modules in VLP, or the slightly different eccentricities of the injection sites (the CTBg injection in case M248 was located nearer to the fovea representation than the two injections in case M237).

Fig. 9. A second example of a tracer injection site located in third tier cortex lateral to DM+ (in dorsal VLP). Case M237LH CTB. (A) CO image of unfolded and flattened visual cortex. (B) The same CO image as in (A) is shown enlarged with overlaid CTB injection site and plotted cell label resulting from it (red). Black arrow on the CTB injection site indicates the direction of travel of the injection, from superficial layers (medially) to deeper layers (laterally). (C) Proportion of labeled cells in extrastriate cortex resulting from the CTB injection. For abbreviations, see legend in Fig. 1. Other conventions as in Fig. 3.

In conclusion, our data so far indicated that the lower quadrant representations located in third tier cortex medial and lateral to upper field DM belong to two different cortical areas, DM and VLP, respectively.

Anteroposterior rows of tracer injections in cortex anterior to upper field DM: Area DA/DI

We next compared the patterns of inter-areal label observed after injections placed medial and lateral to upper field DM (in lower field DM and VLP, respectively) with injections located anterior to these areas, i.e., in the cortical region corresponding to areas DA and DI of Rosa and Schmid (Reference Rosa and Schmid1995) (Fig. 1B). Throughout this manuscript, we have termed this cortical region DA/DI because our mapping data did not have sufficient resolution to allow us to parcellate it into two separate areas. Fig. 10 illustrates a case (M298) in which we made four closely spaced injections of different tracers in an anteroposterior row. DY (yellow) was the most posterior injection, followed in anterior progression by injections of CTB488 (green), CTB555 (red), and FB (blue). The most posterior DY injection involved mostly the HM representation at the anterior border of V2d, as indicated by the predominance of resulting cell label near the lower HM representations of V1d, and dorsal VLP, VLA, and MT. In ventral cortex, instead, DY label lined the HM representations of V2v, ventral MT and likely V1v and ventral VLP. However, this injection may have slightly encroached into upper field DM because of the laminar pattern of cell label it produced in V1 (dense in layers 2/3, 4A, and 4B around the HM representation; Table 1), and the heavy label in DA/DI and MT (Fig. S1B), which is a signature of DM, but not V2 injections (see below). However, unlike injections in DM, the DY injection produced no label in PPd, possibly due to its near-foveal eccentricity within upper field DM.

Fig. 10. Tracer injections in area DA/DI. Case M298. (A) CO image of unfolded and flattened visual cortex. The blue arrow points at the location of the FB injection site. (B) The same CO image as in (A) is shown enlarged with overlaid injection sites and plotted cell label resulting from them (DY, yellow; CTB488, green; CTB555, red; FB, blue). Cell counts for the DY injection sites are shown in Fig. S1B, those for the other three injection sites are shown in Fig. 11D.

The three most anterior injections produced label almost exclusively in cortical regions representing the upper visual quadrant, while in lower field regions (e.g., V2d, lower field DM and dorsal VLP and VLA) label only lined the HM representations. This label topography indicates that these injections lay in a region of dorsal cortex representing the upper visual quadrant. That the latter was area DA/DI, rather than DM, was suggested by the absence of retrograde label in V1 (Table 1), and the dense intra-DA/DI label. Labeled cells resulting from these injections dominated in ventral cortex representing peripheral eccentricities (16–20°) in the upper visual quadrant of areas V2, DM, and VLP, and in parietal area PPv (Fig. 11D). The largest fraction of cell label was located in DM (range 20–37%), VLP (range 12–23%), and PPv (range 10–52%). Compared to the two more rostral injections (red and blue), however, the CTB488 (green) injection produced much sparser label in PPv (10% vs. 28 and 52%, respectively) and denser label in MT (21% vs. 5 and 3%, respectively) and MTc (13% vs. 1.3 and 1.4%, respectively). The areal pattern of label produced by this CTB488 injection did not resemble that seen after DM, V2, or VLP injections (Fig. 11). One possible interpretation, thus, is that this injection site was located in area DI, while the two most anterior injections resided in area DA. This would support the proposal of Rosa and Schmid (Reference Rosa and Schmid1995) that the cortical region anterior to upper field DM consists of two areas.

Fig. 11. Quantitative analysis of inter-areal connectivity. (A–D) Areal distribution of the proportion of total retrogradely labeled cells in extrastriate cortex resulting from single tracer injections in DM (A), VLP (B), V2 (C), and DA/DI (D). Different colored bars indicate individual cases (case number and figures illustrating the specific case are indicated in the legend). In the legend of (C), Tn (thin), Tk (thick), P_M (pale-medial), and P_L (pale-lateral) indicate the specific V2 CO stripe injected. Colored asterisks under the x axis in (A–C) indicate statistically significant differences between pairs of areas according to the legends under the x axis. ¹Case M237 FR in (B) was quantified by counting the proportion of anterogradely-labeled fiber patches in each area, since this case produced almost exclusively anterograde label.

Area-specific patterns of corticocortical afferents: Quantitative analysis

We quantified the inter-areal distribution of cell label resulting from each injection site. In addition to label resulting from the injection sites described above, we quantified the cell label resulting from additional injections made in DM, dorsal VLP, and dorsal V2 from previous studies (Table 1). Fig. 11 shows the results of this quantitative analysis for injections in DM VLP, V2, and DA/DI. For comparison, we also quantified cell label resulting from injections straddling the anterior and posterior border of area DM (Fig. S1). The histograms show for each injection case (indicated by different color bars) the proportion of total neurons that was retrogradely labeled in each extrastriate area. We excluded from cell counts labeled cells in V1 and V2 (i.e., the origin of feedforward projections), as well as the intra-areal label (i.e., labeled cells within the area containing the injection site). However, all injections produced the heaviest label intra-areally, and heavy retrograde label in areas V1 and V2, except for injections rostral to DM (in DA/DI) which produced no retrograde V1 label, and two of the VLP injections which produced sparse V1 label.

It is important to note here that the areas we have examined differ in size and relative emphasis on central versus peripheral visual field representations. As a consequence, the fraction of label within each area does not necessarily reflect the actual strength of the projections from that area. Nevertheless, same-area comparisons in the fraction of label produced by different injection sites allow us to determine whether different injection sites reside in the same or different areas, which is the goal of our analysis. However, it is important to take into account the eccentricity location of the injection sites being compared, given prior findings of differing inter-areal connectivity between central and peripheral regions of some cortical areas. In particular, it is known that the central field representations of areas such as V2 and V4 show preferential connections with temporal cortex, while peripheral field representations of these same areas favor connections with parietal cortex (Gattass et al., Reference Gattass, Nascimento-Silva, Soares, Lima, Jansen, Diogo, Farias, Botelho, Mariani, Azzi and Fiorani2005; Ungerleider et al., Reference Ungerleider, Galkin, Desimone and Gattass2008). Therefore, in our analysis below, we point out the potential influences of eccentricity on connectivity.

We found that injections in different areas produced specific inter-areal patterns of labeled inputs. Injections in upper or lower field DM, at 6–12° eccentricities (Fig. 11A), produced heaviest label in V1 and V2, dorsal stream areas DA/DI (mean ± sem: 33 ± 8.4%) and MT (16 ± 0.3%), and in posterior parietal area PPd (25 ± 4.6%). Areas MTc, PPv, OPt, and ventral stream areas VLP and VLA, each contained on average between 2.6 and 5.8% of total label. Sparse label was found in MST and IT (<1.5% in each), and virtually no label in FST. Compared to the adjacent areas, the distinguishing feature of DM injections, at eccentricities between 6° and 12°, was the strong connectivity with DA/DI and PPd. Importantly, with respect to the goal of our study, all our DM injections encompassed similar eccentricities, despite being located in different quadrant representations; therefore, the fact that they all produced similar patterns of inter-areal label suggests that they resided in the same cortical area DM.

Injections in dorsal VLP, at eccentricities between 2° and 5° (Fig. 11B), produced heaviest label in V2, ventral stream areas VLA (31 ± 5.6%) and IT (22.6 ± 7%) and posterior parietal areas PPv (14 ± 5.9%) and OPt (8.9 ± 1.8%); dorsal stream areas DA/DI, MT, and MTc each contained between 2.8 and 8.4% of total label on average. The distinguishing feature of central field VLP injections compared to injections in adjacent areas was the heavy label in IT, and the virtual absence of label in DM (0.9 ± 0.35%). Note that while injections in VLP produced almost no label in DM (Fig. 11A), injections in DM instead, produced sparse, but more significant, label in VLP (2.6% average). This subtle asymmetry in connectivity likely reflects the different eccentricity location of the injections in VLP and DM, and the fact that the representation of the fovea is disproportionally smaller in DM compared to VLP (see Fig. 1B). An important question, therefore, is whether differences in connectivity between central and peripheral visual field representations in VLP could account for the different connectivity patterns observed in Fig. 11A and 11B. This is unlikely, because injections in DM and VLP that were only separated by about 1° in eccentricity (i.e., M265 CTB488, and the two injections in case M237) showed very different connectivity patterns not only with dorsal versus ventral stream areas, but also within the subdivisions of parietal cortex. Moreover, our results in Fig. 11A are consistent with previous results from Rosa et al. (Reference Rosa, Palmer, Gamberini, Burman, Yu, Reser, Bourne, Tweedale and Galletti2009), in which tracer injections placed at all eccentricities within DM also failed to reveal any significant connections with VLA, IT, and PPv, unlike our VLP injections (Fig. 11B), which instead produced large amounts of label in these three areas. Therefore, we are confident that the results of Fig. 11A and 11B reflect differences in connectivity patterns arising from injections placed in different areas (DM and VLP), rather than injections placed at different eccentricities within the same area.

Following injections in V2, within the central 5° of the visual field (Fig. 11C), heaviest label was found in V1, and ventral stream areas VLP (30 ± 4.9%) and VLA (33 ± 4.9%); dorsal stream areas DM, DA/DI, MT, MTc, ventral stream area IT, and parietal area OPt, each contained between 2.5 and 11% of total label (on average). Some inter-injection variability in the areal distribution of label resulting from different V2 injection cases was likely related to the CO stripe location of the injection sites. Thus, for example, label in VLA was least after injections in thick stripes (e.g., M293 CTBg – red bars), and heaviest after injections in thin or pale stripes (e.g., M293 CTB647 – purple bars and DY-yellow bars); instead, DM received denser projections from thick stripes, small projections from thin stripes, and no projections from pale stripes. VLP received projections from all stripe types. However, overall compared to other areas, the distinguishing feature of central field V2 injections was the absence of label in posterior parietal areas PPd and PPv (except for OPt which is a visuotopically organized subdivision of PPv), and in MST and FST. Since all our V2 injections were located at similar eccentricities as the injections in VLP, differences in connection patterns between these two groups of injections (Fig. 1B and 1C) reflect true area differences; this indicates that injections we assigned to VLP could not have been located in V2.

Finally, injections in DA/DI, at eccentricities between 6° and 20° (Fig. 11D), produced the heaviest label in V2, dorsal stream area DM (28 ± 4.9%), ventral stream area VLP (17 ± 3.1%), and posterior parietal area PPv (30 ± 12%); sparse label was seen in ventral stream areas VLA and IT and in MST and FST (<2% in each), no label in V1, and moderate (ranging between 3 and 9.3%) label in the remainder of visual areas (MT, MTc, PPd, OPt). The distinguishing feature of DA/DI injections compared to injections in the other areas was the presence of heavy label in both DM and PPv, and the absence of label in V1. Most relevant to our study is that the different inter-areal label patterns produced by injections in DA/DI versus DM cannot be attributed to eccentricity differences, as the two injection groups overlapped in eccentricity, and individual injections in the two groups that were located at similar eccentricities showed significantly different inter-areal label patterns.

Injections that straddled the posterior or anterior DM+ borders, produced areal patterns of label intermediate between those produced by V2 and DM injections, or by DM and DA/DI injections, respectively (Fig. S1).

Statistical comparison among the four groups of injections (ANOVA with contrasts) revealed that the distribution of label after DM injections differed from that after dorsal VLP injections in having a significantly larger fraction of label in dorsal stream areas DA/DI (P = 0.002), MT (P = 0.004), and in PPd (P = 0.03), and a significantly smaller fraction of label in ventral stream areas VLA (P = 0.006) and IT (P = 0.001) (black asterisks in Fig. 11A indicate statistically significant differences between DM and VLP). DM injections also differed from V2 injections in having a significantly larger fraction of label in dorsal stream areas DA/DI (P < 0.001) and MT (P = 0.001) and in parietal areas PPd (P = 0.03) and PPv (P = 0.04), and a smaller fraction of label in ventral stream areas VLP (P = 0.002) and VLA (P = 0.002) (orange asterisks in Fig. 11A). Compared to DA/DI injections, DM injections produced a significantly larger fraction of label in PPd (P = 0.03), and a smaller fraction of label in PPv (P = 0.013) and VLP (P = 0.03) (cyan asterisks in Fig. 11A).

VLP injections produced a significantly larger fraction of label in IT (P = 0.001), a much larger fraction of label in PPv (P = 0.06) and a smaller fraction of label in DM (P = 0.2) than V2 injections (gray asterisks in Fig. 11B), while they produced significantly more label in ventral stream areas VLA (P = 0.002) and IT (P < 0.001) and less label in DM (P = 0.017), than injections in DA/DI (green asterisks in Fig. 11B). Finally, DA/DI injections, but not V2 injections, produced label in MST, FST, and PPv (P = 0.001), whereas V2 injections produced significantly more label in VLA (P < 0.001) and IT (P = 0.009) than DA/DI injections (red asterisks in Fig. 11C).

These results support the notion that areas V2, DM, VLP, and DA/DI are indeed distinct areas with distinct patterns of inter-areal connectivity. More relevant to the purpose of this study, they indicate that the cortical territories of the third tier cortex located medial and lateral to upper field DM belong to two different areas, namely DM and VLP. The inter-areal connectivity patterns of these areas indicate that the central visual field representation of V2 in marmosets distributes information it receives from V1 to both dorsal and ventral stream areas, with a stronger bias toward the early visual areas of the ventral stream (VLP and VLA). In contrast, DM is heavily connected with the dorsal stream and weakly connected with the ventral stream, while VLP shows a connectivity pattern opposite to that of DM, being more strongly connected with ventral stream, than with dorsal stream, areas. These two areas also show complementary connectivity patterns with parietal cortex, DM being more strongly connected with PPd than PPv, and vice versa for VLP. A summary diagram of the major cortical afferents to DM and VLP is reported in Fig. 12.

Fig. 12. Summary diagram of the major cortical inputs to area DM and VLP. Cortical inputs to (A) DM and (B) VLP from areas of the dorsal (blue) and ventral (pink/purple) streams. Cortical areas are arranged in approximate hierarchical fashion. Color gradients indicate an area's contribution to both streams. Darkest colors indicate strongest afferent connections with DM (in A) or VLP (in B), with lighter shades of color indicating progressively weaker connections. Line thickness also indicates the relative strength of connections. DM is more strongly connected with the dorsal stream (dark blue and lighter pink in A), whereas VLP is more strongly connected with the ventral stream (dark purple and lighter blue in B).