Split Developmental Origin of the Loop of Henle

Renal nephrons are epithelial tubules that consist of a succession of specialized segments. From proximal to distal, these are the Bowman’s capsule, the proximal convoluted tubule, the loop of Henle, the distal convoluted tubule, and the connecting segment, which connects to the urinary collecting duct tree. Some of these contain their own subdivisions; the proximal tubule is usually considered to contain three successive segments (the convoluted S1/S2 and the straight S3), and the loop of Henle contains a thin descending limb, a thin ascending limb, and a thick ascending limb. The entire nephron arises from the metanephric mesenchyme and makes a connection with the ureteric bud–derived collecting duct so that urine can flow from one to the other. This mesenchymal origin of nephrons was first deduced from sections of embryonic material in the 19th century¹ and confirmed in the 1950s by culture systems that used nonrenal epithelia to induce nephrons from metanephric mesenchyme.²

One result of studying kidney development at a molecular level has been a pushing back of the time that nephron segmentation is thought to take place. In the premolecular era, the general consensus was that nephrons first formed as compact but distinctly tubular structures before their different segments acquired distinct identities.³ A combination of 21st century in situ hybridization, microarray, and transcriptomic experiments distinguished differences between segments long before they became histologically apparent and showed that segmental identities arise as epithelialization takes place. When the first cells leave the progenitor niche to make a new nephron, they express markers of distal character; as more cells stream from the progenitor niche to the epithelializing nephron, they express markers of more and more proximal character so that the last to arrive become podocytes and parietal epithelium of the Bowman's capsule.⁴ This “gradual recruitment” mechanism means that distal character is already well established, while cells that will become proximal tubule are still in a nonepithelial, progenitor state.

Most work on the control of nephron segmentation has focused on how a cell “knows” what segment it is in. This work has highlighted the importance, for example, of gradients of β-catenin activation, high at the distal end and low at the proximal, in controlling cell identity along a proximo-distal axis.⁵ Given that the chief source of β-catenin–activating Wnt signaling is the collecting duct stalk, this gradient may emerge naturally from the “gradual recruitment” model because the first cells to leave the progenitor population would be closest to the stalk and spend longest in its vicinity and therefore experience greater β-catenin activation.⁶

Another question is whether all of the detailed segments of the nephron appear one after the other in a temporal sequence ordered by their final sequence in space, or whether coarse divisions (e.g., “proximal”) appear first, and later subdivide into finer zones (e.g., S1, S2, and S3). Classical studies of embryo segmentation provide precedents for both: In the fruit fly, for example, the body is divided first into relatively broad zones, and these then subdivide eventually into segments.⁷ In vertebrates, on the other hand, mesenchyme segments (somites) appear in an orderly sequence, segment by segment, from neck to tail instead of appearing as larger blocks (e.g., thoracic, lumbar) that only later subdivide. In nephron development, there has already been some evidence for division into broad zones first, in the period of “gradual recruitment,” and then later subdivision. Current understanding is shown particularly clearly in Figure 3A of ref. 8, which maps cell types identified by single-cell transcriptomics against physical positions in the developing nephron and deduces progenitor–successor relationships based on position and relationship of transcriptomes.⁸ In the dendrogram summarizing the “family tree” of nephron cell types, for example, one cell type of the distal pretubular aggregate begets two cell types in the later comma-shaped body, which resolve into the connecting tubule, distal tubule, and precursors for both ascending limb and macula densa of the mature nephron. More proximally in the comma-shaped body, there is evidence for two proximal/medial progenitor types and at least one of these gives rise to the proximal tubule. The question of what the other makes was left unanswered in the summary diagram by Lindström et al. referred to above,⁸ but in this issue of JASN, Chung et al.⁹ may well have filled in the blank.

Chung et al. began by making a dataset of single-cell transcriptomes from the kidneys of late fetal or newborn mice in which cells of almost all developmental stages will be present. Cells expressing markers for proximal tubule formed a distinct cluster on a graph of the state space of gene expression; cells expressing markers of the S1 segment made up one part of the cluster, while those expressing markers of the S2 and S3 segments made up adjacent parts. The S3 cells were right on the cluster's edge, near another cluster that represented thin descending limb of the loop of Henle. Detailed analysis of those S3 cells revealed a surprise: as well as expressing classical proximal S3 markers (e.g., Slc5a8, Kap), they expressed Bst and Aqp1, characteristic of the thin descending limb. Immunostaining of developing kidneys confirmed that some cells in developing nephrons do indeed coexpress markers of mature S3 with also markers of mature thin descending limb. This suggested to Chung et al. that the thin descending limb may arise from proximal tubule. To test this, they generated a line of mice expressing constitutive Cre recombinase under the control of the Slc34a1 promoter, active only in proximal cells. Crossed with a reporter for Cre activity that persists even when Cre activity has ceased, the mice showed extensive labeling of cells in segments S1 and S2 of the proximal tubule, and many S3 cells were also labeled. So far, so expected. The surprise was that the label also turned up in cells in the medulla; these expressed Aqp1, a marker for the loop of Henle thin descending limb. It did not appear in cells that expressed Slc12a1 (Na-K-Cl cotransporter), a marker for thick ascending limb. No other parts of the nephrons were labeled, showing the reporter to be “clean.” The implication was clear; at least some cells of the descending thin limb must have arisen from cells that once expressed Slc34a1. This was supported by velocity analysis of their single-cell transcriptomic data, which suggests that some Slc34a1⁺ cells, in the S3 area, are on a path to produce Aqp1⁺ cells.

Chung et al. state this result to mean “This result indicates that thin descending limb cells are descendants of proximal tubule cells.”⁹ Well, perhaps; the data also allow the possibility that proximal tubule and descending thin limb share a common, Slc34a1-positive progenitor that maintains its Slc34a1 expression in its proximal tubule progeny but not in those that make the descending limb. Whether the progenitor of the descending limb deserves to be labeled “proximal tubule,” it might be one of the two identified proximal-precursor cell types (“10” and “11”) in Figure 3 of Lindström summary diagram.⁸ There is an additional reason to suppose that S3 and the thin descending limb might share a common developmental path, and it comes from an evolutionary study. Corkins et al.¹⁰ compared nephrons of the amphibian pronephros with nephrons of the mammalian metanephros to determine how their various segments “pair up” in the sense of being homologous. They found that the pronephros has homologues to proximal S1 and S2, and of straight and convoluted distal segments, but has neither proximal segment S3 nor loop of Henle. This suggests that S3 and the loop may have coevolved as the more advanced type of kidneys appeared, which fits (but does not prove) the idea of their having a common lineage.

What the new data from Chung et al.⁹ definitely do say is that the thin descending limb and the thick ascending limb have different origins. This article shows that the thin descending limb comes from the cells that are Slc34a1⁺ and that the thick ascending limb does not. The earlier work of Lindström et al.⁸ suggested that the thick ascending limb arises from a POU3F3⁺/Irx1⁺ zone that also gives rise to PAPPA2⁺ macula densa. While functional and anatomical considerations cause human researchers to group the descending and ascending limbs together as a “loop of Henle,” it is becoming clear that the developing kidney divides itself up a different way: the descending part being developmentally close to the proximal tubule, and the ascending thick part being developmentally close to the macula densa. It is a reminder that living things do not necessarily split and group their parts in the way that we, who try to make sense of them, do.

There is more to the Chung et al. paper than segmentation; there is also much about dependencies of differentiation on Hnf4a and on cellular velocities in the state space of gene expression. It is sad that one of its authors, the late Steve Potter who did so much to bring high-precision transcriptomics to kidney development, died before he could see this elegant work appear in JASN. He will be remembered with great affection and respect.

Disclosures

Disclosure forms, as provided by each author, are available with the online version of the article at http://links.lww.com/JSN/F353.

Author Contributions

Writing – original draft: Jamie A. Davies.

Writing – review & editing: Jamie A. Davies.

Funding

None.